Novel flow cytometric approach for the detection of adipocyte subpopulations during adipogenesis[S]

The ability of mesenchymal stromal cells (MSCs) to differentiate into adipocytes provides a cellular model of human origin to study adipogenesis in vitro. One of the major challenges in studying adipogenesis is the lack of tools to identify and monitor the differentiation of various subpopulations within the heterogeneous pool of MSCs. Cluster of differentiation (CD)36 plays an important role in the formation of intracellular lipid droplets, a key characteristic of adipocyte differentiation/maturation. The objective of this study was to develop a reproducible quantitative method to study adipocyte differentiation by comparing two lipophilic dyes [Nile Red (NR) and Bodipy 493/503] in combination with CD36 surface marker staining. We identified a subpopulation of adipose-derived stromal cells that express CD36 at intermediate/high levels and show that combining CD36 cell surface staining with neutral lipid-specific staining allows us to monitor differentiation of adipose-derived stromal cells that express CD36intermediate/high during adipocyte differentiation in vitro. The gradual increase of CD36intermediate/high/NRpositive cells during the 21 day adipogenesis induction period correlated with upregulation of adipogenesis-associated gene expression.

BDP is a nonpolar lipid stain that has the ability to stain neutral lipids as well as other nonpolar oils and lipids. BDP emits green fl uorescence when dissolved in these lipids/oils (product insert, https://www.lifetechnologies. com) ( 46,47 ). In contrast, the emission spectrum of the lipophilic dye, NR, changes according to the lipid environment. NR emits yellow-gold fl uorescence (emission wavelength >528 nm) when dissolved in neutral lipids, while it fl uoresces in the deep-red spectrum (emission wavelength >590 nm) when dissolved in amphipathic lipids (lipids with polar as well as nonpolar regions) (48)(49)(50)(51).
In this study, we combined a lipid-specifi c stain with surface marker (CD36) staining to monitor differentiation of ASCs of human origin into adipocytes in vitro. Two fl uorescent lipid-specifi c stains were compared, namely, NR and BDP. By making use of the unique fl uorescent emission profi les of these lipid-specifi c dyes, we show that by applying multi-parameter fl ow cytometry, we are able to distinguish between different cellular stages during adipocyte differentiation. We also demonstrate that adipocyte CD36 surface expression precedes the expected increase in intracellular lipid content associated with adipocyte differentiation. In addition, we demonstrate a linear relationship between the upregulation of the adipogenesis end-stage gene, FABP4, and the increase in CD36 surface expression during adipocyte differentiation/maturation.

Isolation of ASCs from adipose tissue
ASCs were isolated, with minor modifi cations, from human adipose tissue as previously described ( 52,53 ). Subcutaneous adipose tissue was obtained from 10 healthy donors that underwent elective liposuction procedures under general anesthesia. Informed consent was obtained from all donors. The study was approved by the Ethics Committee, Faculty of Health Sciences, University of Pretoria, study numbers 218/2010 and 421/2013. Briefl y, excess oil was removed by washing the harvested adipose tissue in PBS. Adipose tissue was digested by constant agitation for 45 min at 37°C using 0.1% collagenase type I prepared in PBS supplemented with 2% Pen/Strep. The adipose-derived stromal vascular fraction was separated from debris and connective tissue by centrifugation (400 g , 5 min). Pellets were resuspended in VersaLyse and incubated for 10 min at room temperature to lyse contaminating red blood cells. After centrifugation, the cells are also highly dependent on their differentiation state (preadipocytes vs. mature adipocytes) ( 10,19,20 ). In addition, the ratio of preadipocytes to mature adipocytes changes during infl ammation ( 10,22 ). A current limiting factor in fully understanding the different stages of adipogenesis is the inability to accurately defi ne and study intermediate adipocyte subpopulations that exist between MSCs and mature adipocytes ( 23 ). In addition, the majority of adipogenesis studies are performed in vitro using murine cell culture models ( 24 ). During these studies, the involvement of two adipocyte populations, preadipocytes and mature adipocytes, has mainly been investigated. It is also important to recognize that adipose tissue distribution in rodents and humans is different ( 25 ). Therefore, translating observations made on rodent cell culture models to humans should be undertaken with caution. Primary human preadipocyte cultures may be more representative of adipogenesis in humans, and the ability to differentiate MSCs into adipocytes in vitro serves as a proxy for the study of adipogenesis in vivo. Development and improvement of techniques that more accurately represent adipocyte subpopulations during adipogenesis may assist in a better understanding of the adipogenesis process, as well as how other processes, like infl ammation, infl uence the repertoire of the various cell populations present in adipose tissue.
Cluster of differentiation (CD)36 is a multifunctional transmembrane protein that is highly expressed on the adipocyte cell surface ( 38 ) as well as various other cell types, including monocytes, macrophages, endothelial cells, and platelets (38)(39)(40)(41). One of the main functions of CD36 is to serve as a fatty acid translocase by facilitating the uptake of free long-chain fatty acids into cells (38)(39)(40)(41). The uptake of these free fatty acids is followed by esterifi cation into triglycerides, which are then stored within cytosolic lipid droplets ( 42 ).
Adipogenesis is a complex multi-step process that involves various transcription factors in a well-described sequential manner. The process of adipogenesis begins with the up-and downregulation of various transcription factors during the clonal expansion phase. These transcription factors in turn activate other transcription factors, such as CCAAT/enhancer binding protein (C/EBP) ␣ and PPAR ␥ . The upregulation of C/EBP ␣ and PPAR ␥ are essential during the differentiation stage of adipogenesis. The latter stages of differentiation are accompanied by the activation of various genes associated with adipocyte maturation, including fatty acid binding proteins (FABPs) such as FABP4 (43)(44)(45). into overlay images. For NR, three single channel images were captured. The fi rst single channel image was captured using Filter Set 9 (excitation BP 450-490, emission LP 515; Carl Zeiss) to visualize yellow-gold fl uorescence. A second single channel image was captured using Filter Set 00 (excitation BP 530-585, emission LP 615; Carl Zeiss) to visualize deep-red fl uorescence emission of lipid droplets. A third single image was captured using Filter Set 49 (excitation G 365, emission BP 445/50; Carl Zeiss) to visualize nuclei stained with DAPI. For BDP, two single color images were captured using Filter Set 9 and Filter Set 49, respectively. Images were initially captured using AxioVision software (Version 4.8.2).
In order to optimally visualize lipid droplets, all images were enhanced, but not manipulated, post-acquisition using Image J imaging software ( 54 ). Enhancement of images was done by adjusting contrast and brightness settings.

RNA isolation and RT-quantitative PCR
Total RNA was extracted from postconfluent ASCs using the RNeasy mini kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. cDNA was generated using the iScript™ reverse transcription supermix (Bio-Rad Laboratories, Inc., Hercules, CA). For RT-quantitative (q)PCR, LightCycler ® 480 SYBR Green I Master Mix (Roche, Basel, Switzerland) was used. PCR reactions were performed in 10 l volumes, where the primer concentrations were 400 nM and the cDNA concentration was 20 ng/ l. qPCR was performed on a LightCycler 480 II instrument (Roche) using the following conditions: denaturation at 95°C for 5 min, 45 cycles of amplifi cation at 95°C for 30 s, 62°C for 30 s, 72°C for 30 s. After amplifi cation, a melt curve was performed at 95°C for 30 s, 40°C for 30 s, and ramped at 0.1°C/s. The primers (IDT, Coralville, IA) for the genes of interest and the reference genes (internal controls) are indicated in Table 1 .

Statistical analysis
Results are expressed as means ± SD. Outliers were identifi ed using the ROUT statistical test and excluded from statistical analysis. The nonparametric Mann-Whitney test was used to determine statistical signifi cance between groups. The degree of correlation between different groups was performed using the nonparametric Spearman rank correlation test. GraphPad PRISM 6 (Version 6.07) and INSTAT software Version 3.06 (GraphPad Software Inc., La Jolla, CA) were used for all statistical analyses.
For the qPCR data, relative gene expression was calculated using the comparative C T method. Relative fold-increase in gene expression ( ⌬ C T ) was reported as an increase in gene expression relative to the following housekeeping genes: glucuronidase ␤ (GUSB), peptidylprolyl isomerase A (PPIA), TATA binding protein (TBP), and tyrosine 3-monooxygenase/tryptophan 5-monooxygenase activation protein zeta (YWHAZ) ( Table 1 ).

RESULTS
Adipocytes are fragile cells ( 55,56 ). During fl ow cytometric analysis, intact cells were identifi ed according to their intermediate/high forward scatter characteristics (see supplementary Fig. 1 ). To confi rm that these cells (intermediate/high forward scatter) were viable, the cells were stained with the nuclear stain, DAPI (5 M, fi nal concentration). The viability of the cells identifi ed as intact were resuspended in culture medium. The culture medium consisted of ␣ -MEM supplemented with 10% FBS and 2% Pen/ Strep. The cell suspension was fi ltered through a 70 micron Falcon cell strainer (BD Biosciences, San Jose, CA) and was seeded (37°C, 5% CO 2 ) at a density of 5 × 10 3 cells/cm 2 in 25 cm 2 Nunc culture fl asks (Nunc, Rosklide, Denmark). After 24 h, the fl asks were rinsed with PBS to remove nonadherent cells, followed by the addition of fresh expansion medium.

Expansion of ASCs
The plated cells were maintained at 37°C/5% CO 2 in culture medium. At 80-90% confl uence, cells were trypsinized for 10 min at 37°C using 0.25% trypsin-EDTA. Cells were replated at a density of 5 × 10 3 cells/cm 2 and cultures were expanded for 6 to 10 passages. All cultures were phenotyped at each passage. The cells used in differentiation experiments were positive for CD73, CD90, and CD105 and negative for CD34 and CD45.

Adipocyte quantifi cation using fl ow cytometry
A DAPI working solution was prepared in staining buffer [100 mM Tris (pH 7.4), 150 mM NaCl, 1 mM CaCl 2 , 0.5 mM MgCl 2 ] to achieve a concentration of 10 g/ml. NR and BDP working solutions were prepared in absolute ethanol to achieve concentrations of 2 g/ml and 20 g/ml, respectively. Prior to staining, the cells were trypsinized and washed using PBS supplemented with 10% FBS and 2% Pen/Strep. After centrifugation (184 g , 5 min), the pellets were resuspended in supplemented PBS and then stained with NR (20 ng/ml fi nal concentration) or BDP (200 ng/ml), mouse anti-human CD36-APC, and DAPI (5 g/ml fi nal concentration). After 20 min incubation at room temperature, cells were analyzed using a Gallios fl ow cytometer (Beckman Coulter, Miami, FL). DAPI was excited with a 405 nm laser and fl uorescence emissions were collected using the fl uorescence detector (FL)9 [450/40 nm band pass fi lter (BP)] detector. NR was excited with a 488 nm laser and fl uorescent emission signals were collected using FL2 (575/30 nm BP) and FL5 [755 nm long pass fi lter (LP)] detectors. Noninduced ASCs (day 1, 24 h post adipogenic induction) were used to optimize the photomultiplier tube voltage settings, as well as to set the signal-to-background threshold of the NR and BDP fl uorescence. Compensation settings were optimized using a differentiated culture (day 21) by removing any fl uorescent spill over into the FL2 and FL5 channels. After initial optimization, all instrument settings were kept constant for the duration of the study (supplementary Fig. 2). Flow Check Pro (Beckman Coulter) fl uorospheres were run daily to validate instrument performance. Flow cytometry data were analyzed using Kaluza fl ow cytometry data analysis software (Version 1.3; Beckman Coulter).

Fluorescence microscopy and analysis
Cells (both noninduced and induced) were cultured in 6-well plates as described above. Prior to fl uorescence microscopy imaging, culture medium was removed and the wells were rinsed with PBS to remove nonadherent cells. PBS supplemented with 10% FBS and 2% Pen/Strep was added to the wells. Both noninduced and induced cultures were stained with 2.5 g/ml (fi nal concentration) DAPI. The cultures were incubated overnight in a 5% CO 2 incubator to allow optimal staining of all nuclei. The next day cultures were stained with either NR (fi nal concentration 50 ng/ml) or BDP (fi nal concentration 500 ng/ml). Fluorescence images (20× objective; 256 × 256 pixels) were captured after a 20 min incubation at room temperature, using an AxioVert A1 inverted fl uorescence microscope (Carl Zeiss, Gottingen, Germany) equipped with an AxioCam Cm1 camera (Carl Zeiss). Single channel images were captured and subsequently converted with NR at various time points (days 0, 1, 7, 14, and 21) after adipocyte differentiation was induced. The percentage of cells that emitted yellow-gold fl uorescence due to the formation of intracellular lipid droplets increased gradually over the 21 day period ( Fig. 1B ). On day 7, 8.57 ± 5.05% ( P < 0.0001 compared with day 1) intact cells emitted yellow-gold fl uorescence. This increased to 16.66 ± 8.88% on day 14 ( P < 0.0001 compared with day 1). By day 21, the percentage of intact cells that emitted yellowgold fl uorescence had further increased to 21.46 ± 9.44% ( P < 0.0001 compared with day 1). Initially, approximately 1% of cells in the noninduced cultures emitted low levels was 96.57 ± 3.33% (noninduced) and 93.87 ± 8.5% (induced), respectively. Our results confi rm previous fi ndings that larger mature adipocytes seem to be more fragile and susceptible to damage during sample processing and analysis ( 55 ).
The fi rst indication of adipocyte differentiation is the appearance of lipid droplets in the cytoplasm of differentiating preadipocytes ( 49 ). The lipid droplet core consists mainly of neutral lipids, triglycerides, and cholesterol esters ( 27,57 ). The lipophilic dye, NR, emits yellow-gold (emission wavelength >528 nm) fl uorescence when dissolved in neutral lipids ( 48,50 ). Adipocytes were stained Genes of interest Preadipocyte factor 1 ( Pref-1 )  1A ).The median yellow-gold fl uorescence intensities observed in induced ASCs were signifi cantly higher at day 14 ( P = 0.016) and day 21 ( P < 0.008) when compared with the MFIs observed for the noninduced cells at the corresponding time points. No changes were observed in the median yellow-gold fl uorescence intensities observed for the noninduced cultures during the 21 day culture period ( Fig. 1A ).
A second neutral lipid-specifi c dye was included, namely, BDP. The sensitivity of NR and BDP to detect, using fl ow cytometry, intracellular lipid droplets during adipogenesis was directly compared in ASC cultures from seven independent donors ( Fig. 3 ). BDP is a new class of lipidspecifi c fl uorescent dyes and has the capacity to stain neutral and other nonpolar lipids. It is described as being more sensitive than NR for staining lipid droplets ( 61 ). In of yellow-gold fl uorescence ( Fig. 1B ). By day 21, the proportion of nondifferentiated ASCs in noninduced cultures that emitted low levels of yellow-gold fl uorescence had increased to 4.70 ± 5.20% ( Fig. 1B ). This observed increase in the percentage of noninduced cells that emitted low levels of yellow-gold fl uorescence was not statistically signifi cant and may have contributed to the high levels of variance observed between the ASC cultures that were investigated. The noninduced cultures were confl uent on day 21, resulting in contact inhibition. The observed increase in yellow-gold fl uorescence emitted by these nondifferentiated ASCs is consistent with previous fi ndings that lipid droplet biogenesis occurs in cells under stress, i.e., during contact inhibition ( 58,59 ). Our results support previous fi ndings ( 60 ) that lipid droplets present in cells other than adipocytes are usually very small in size (<1 m in diameter) ( Fig. 2A, C ), compared with the intracellular lipid droplets that form during adipocyte differentiation DAPI was used to visualize the nuclei. Images were obtained 21 days post induction. Images were initially captured as single channel images and then merged using Image J software.  64 ) demonstrated, using the murine 3T3 cell line, that CD36 is directly involved in adipocyte differentiation and adipogenesis. In this study, we combined CD36 surface expression staining with a lipidspecifi c fl uorescent stain to monitor adipocyte differentiation in vitro. According to the International Fat Applied Technology Society and the International Society of Cellular Therapy guidelines, ASCs should dimly express CD36 ( 3 ). In this study, 97.17 ± 4.92% (n = 10; MFI: 2.52 ± 2.14) of the ASCs expressed CD36 dimly. A subpopulation of ASCs (5.13 ± 4.22%) expressed CD36 at a higher intensity (MFI: 14.01 ± 11.52) (supplementary Fig. 3). These CD36 intermediate/high cells were the cells of interest in this study. The proportion of CD36 intermediate/high cells, as well as the expression level of CD36, increased gradually over the 21 day culture period when ASCs were induced to general, a greater proportion of cells emitted fl uorescence associated with increased neutral lipid content when stained with BDP compared with NR ( Fig. 3A, B ). Day 1 cultures (noninduced and induced, respectively) were used to set the positive detection limits for both NR and BDP ( Fig. 3A, B ). Although the differences that were observed between NR and BDP were not statistically signifi cant, NR staining resulted in a signifi cantly higher proportion of cells that emitted yellow-gold fl uorescence post induction when compared with the noninduced cells at day 7 ( P = 0.036), day 14 ( P = 0.0006), and day 21 ( P = 0.0006). A gradual increase in fl uorescence was observed over the 21 day period when the noninduced cultures were stained with BDP, resulting in nonsignifi cant differences between the noninduced and induced cultures ( Fig. 3B ). In order to compare the proportion of cells with increased neutral lipid content associated with adipogenesis as detected by NR and BDP, the detection limits were adjusted to compensate for the increase in fl uorescence observed in the noninduced cultures ( Fig. 4 ). When the detection limits were set according to each time point's control (noninduced cultures), both NR and BDP detected a similar proportion of cells ( Fig. 4A ; r 2 = 1.0; P = 0.08). In order to compare the relative fl uorescence intensities of NR and BDP, the signal (emitted fl uorescence) to background fl uorescence ratios was calculated using the following formula: signal:background ratio = MFI of cells emitting fl uorescence/MFI of cells only emitting background fl uorescence. Although variable, we found that NR provided a better resolution between cells emitting fl uorescence and cells that did not emit fl uorescence (signal:background ratio) when compared with the signal:background ratio observed for BDP ( Fig. 4B ; supplementary Fig. 2).
In order to confi rm that the increase in neutral lipid content observed in the noninduced samples was not due to the spontaneous differentiation of ASCs, gene expression studies were performed. C/EBP ␣ and PPAR ␥ are two of several transcription factors that play an important role in adipocyte differentiation. These transcription factors are involved in a well-described cascade of molecular events that leads to the expression of proteins, such as FABP4, that are characteristic of mature adipocytes ( 45,62,63 ). The baseline level of expression of these genes in ASCs is very low. The relative baseline gene expression of C/EBP ␣ , PPAR ␥ , and FABP4 in ASC cultures (n = 6) prior to adipogenic induction (day 0) was 0.24 ± 0.53, 0.16 ± 0.10, and 0.005 ± 0.005, respectively. None of these genes were signifi cantly upregulated in noninduced ASC cultures during the 21 day culture period ( Fig. 5A-C ), confi rming that the increase in lipid accumulation observed at day 21 is not due to spontaneous differentiation of the ASCs into adipocytes. In contrast, all three genes were signifi cantly upregulated in the induced cultures ( Fig. 5A-C ). C/EBP ␣ and PPAR ␥ were signifi cantly increased from day 14 onwards ( Fig. 5A, B ), while FABP4 was signifi cantly upregulated from day 7 onwards ( Fig. 5C ). The relative fold-increase for FABP4 gene expression was very low (<0.15) in the noninduced cultures. FABP4 is mainly expressed by mature adipocytes and macrophages ( 65 ). Similar to CD36, FABP4 binds longchain fatty acids and, in so doing, facilitates their transport into cells, after which the fatty acids are esterifi ed and stored as triglycerides within intracellular lipid droplets ( 42 ). The upregulation of FABP4 is associated with the latter stages of adipocyte differentiation, and is therefore an indication of the adipocyte maturation process. We investigated the relationship between FABP4 gene expression and CD36 intermediate/high expression, as well as FABP4 gene differentiate into adipocytes ( Fig. 6A, B ). At day 1 (24 h after induction was initiated) 10.4 ± 4.5% CD36 intermediate/highpositive cells were observed. The proportion of the CD36 intermediate/high -positive cells increased to 11.63 ± 7.03% on day 7 ( P = 0.34 compared with day 1), 20.15 ± 10.86% on day 14 ( P = 0.0099 compared with day 1), and 25.25 ± 13.95% on day 21 ( P = 0.007 compared with day 1) ( Fig. 6B ). CD36 expression (as indicated by the MFI) was statistically signifi cant on day 14 ( P = 0.05) and day 21 ( P = 0.05) when compared with the baseline level at day 0 ( Fig. 6A ). The observed increase in CD36 intermediate/high cells correlated ( r 2 = 1.0, P = 0.08, n = 4) with the proportion of cells with increased neutral lipid content. Our results also show that
The outer layers of lipid droplets consist of amphipathic lipids, mainly cholesterol and phospholipids. As preadipocytes differentiate into mature adipocytes, the lipid droplets merge to eventually form one large unilocular droplet that occupies around 90% of the cytoplasm ( 27,49 ). Greenspan, Mayer, and Fowler ( 48 ) showed that NR emits deep-red fl uorescence (emission wavelength >590 nm) when it binds to amphipathic lipids. We postulated that these unique emission properties of NR would allow us to identify a more mature adipocyte phenotype during the end-stages of adipocyte differentiation in vitro. The maturation of adipocytes is associated with an increase in the number, as well as in the size, of intracellular lipid droplets, which consequently contribute to an increase in amphipathic lipid quantities present in adipocytes ( Fig. 9A-H ). Our results show that an increase in amphipathic lipids results in the emission of deep-red fl uorescence (FL5+) when stained with NR ( Fig. 10A ; supplementary Fig. 5; days 14 and 21). Due to the absence of relatively large lipid droplets in the noninduced cultures, these cultures did not emit fl uorescence in the deep-red spectrum in fl ow cytometric assays to quantify adipocyte differentiation, there is no consensus in reporting fl ow cytometric results after staining adipocytes with NR. Most investigators report the fl ow cytometric results as both percentages of cells that emit yellow-gold fl uorescence after staining with NR, as well as the mean or median fl uorescence intensity of the emitted yellow-gold fl uorescence . Other investigators choose to only report the mean or median fl uorescence intensity of the emitted yellow-gold fl uorescence. Our results confi rm that yellow-gold median fl uorescence intensity seems to be a more sensitive indication of adipocyte formation. We also observed that NR resulted in a better signal:background resolution compared with BDP ( Fig. 4B ).
Our results indicate that BDP is more sensitive than NR in detecting changes in intracellular lipid content ( Fig. 3B ).
differentiation. To our knowledge, this is the fi rst study that has used BDP to monitor adipocyte differentiation using fl ow cytometry.
Currently, all NR-based fl ow cytometric studies make use of the ability of NR to emit yellow-gold fl uorescence (emission between 505 and 585 nm) when dissolved in neutral lipids after excitation by a 488 nm light source ( 33,35,49,60,66,68,74 ). The emission of yellow-gold fl uorescence is associated with the total neutral lipid content of cells when stained with NR. Lipid droplets are present in all cells, but are generally larger in size, as well as present in higher quantities, in adipocytes and macrophages. The core of these intracellular lipid droplets consists mainly of neutral lipids, triglycerides, and cholesterol esters ( 27,57 ). Although NR has increasingly been used ; the proportion of cells that highly express CD36 and simultaneously emitted yellow-gold fl uorescence (NR-associated) (D); and proportion of cells that highly express CD36 and simultaneously emit green fl uorescence after staining with BDP (E). Cultures were terminated on days 0, 1, 7, 14, and 21 after induction of adipogenesis. Gene expression levels were normalized to day 0. All results are expressed as mean ± SD and are from fi ve independent ASC cultures from fi ve different donors. an increase in intracellular cholesterol levels . Diaz et al. ( 79 ) suggest cell-to-cell contact changes the ratio of triglycerides to cholesterol esters present in the lipid droplet core by causing a decrease in the triglyceride content of the lipid. The fl uorescence emission profi le of NR seems to be sensitive to changes in the hydrophobic strengths of lipids ( 79 ). The predominant lipids present in the lipid droplet core are triglycerides, followed by cholesterol esters ( 69 ). A possible explanation for the differences in the detection sensitivity of NR and BDP may thus be due to the specifi city of NR to emit yellow-gold fl uorescence when dissolved in triglycerides. This results in NR being potentially more sensitive for adipogenesis-associated changes to the intracellular lipid content. BDP seems to be less specifi c to the hydrophobic strength of lipids and thus measures all neutral lipids irrespective of their hydrophobic strength. This hypothesis, however, needs to be tested directly in further studies.
CD36, a fatty acid translocase, plays an important role in fatty acid metabolism during lipid droplet formation The increase in sensitivity observed for BDP seems to result in the detection of nonadipogenesis-related changes in the neutral lipid content of the cells. This increase in intracellular lipid content observed in noninduced cells is not due to spontaneous differentiation of ASCs, as no upregulation of adipogenesis-associated transcription factors was observed.
Many investigators have shown that cell-to-cell contact leads to changes in the lipid content of cells. Thus, using murine 3T3 fi broblasts, it was observed that contact inhibition led to an increase in intramembraneous structures in these cells (75)(76)(77). The exact lipid composition of these intramembraneous structures is unknown, but interestingly, Cansell et al. ( 78 ) did not fi nd any difference in the total phospholipid content between actively proliferating and confl uent endothelial cells. It has also been reported that in addition to preventing proliferation, contact inhibition leads to changes in lipid droplet composition ( 78,79 ). Cansell et al. ( 78 ) reported that contact inhibition led to an increase in CD36 expression precedes an increase in intracellular neutral lipid content. In addition, we observed an improved correlation of total neutral lipid content due to the intracellular accumulation of lipid droplets, with upregulation of FABP4, when the simultaneous detection of CD36 high expression and an increase in neutral lipid content was used ( Fig. 8E ). This observation would suggest that the combination of CD36 expression with intracellular lipid quantifi cation allows for a more sensitive means of monitoring adipocyte differentiation.
Our data suggest that BDP is a sensitive indicator of the overall lipid content of cells. However, this increased sensitivity may result in the overestimation of changes in lipid content associated with adipogenesis. It is therefore important to normalize the data to a corresponding noninduced control when BDP is used as an indicator of adipogenesis-associated lipid content. The neutral core of (38)(39)(40)(41)(42). In this study, we show that by combining a lipidspecifi c stain with CD36 surface marker staining, we are able to identify three main subpopulations during adipocyte differentiation. The gene expression data confi rm that both the increases in intracellular neutral lipid content, as well as the upregulation of CD36 cell surface expression, are associated with adipogenesis. PPAR ␥ expression controls the expression of FABP4 during the later stages of adipocyte differentiation ( 26,80 ). The exact function of FABPs, including FABP4, is not yet fully known, but one of the proposed functions of these proteins, also known as lipid chaperones, may be to facilitate the transport of lipids to specifi c compartments in the cell, such as to the lipid droplet for storage ( 80 ). CD36 is highly expressed by mature adipocytes ( 38,81 ). The expression of CD36, therefore, may be used to monitor the transition from adipocyte differentiation to maturation, as we observed that fl ow cytometric parameters and gene expression studies are combined. Therefore, the proposed model may contribute to a better understanding of the role of various intermediate phenotypes in the adipocyte differentiation process, both in physiological and pathological settings. lipid droplets is surrounded by an outer layer of amphipathic lipids, such as phospholipids and cholesterol ( 27,57 ). In this study, we observed that adipocytes emit deep-red fl uorescence, when stained with NR, as they mature. Based on the observation reported by Greenspan and Fowler ( 51 ) indicating that NR emits fl uorescence in the deepred fl uorescence spectrum when dissolved in amphipathic lipids , we postulate that the observed emission of deepred fl uorescence might be due to the higher levels of amphipathic lipids that surround larger lipid droplets. We therefore postulate that the observed increase in deep-red fl uorescence emission serves as an indicator of the presence of larger lipid droplets within adipocytes, and thus is an indicator of adipocyte maturation. Therefore, the unique emission spectrum of NR provides an opportunity to detect a more mature stage of adipocyte differentiation. Our results suggest that simultaneous staining with NR and CD36 allows for the identifi cation of an even more mature adipocyte phenotype (CD36 high /NR FL2+/FL5+ ) that displays an increased level of cellular complexity ( Fig. 10 ). However, we recognize that quantifi cation of the ratio of neutral versus amphipathic lipids in the various adipocyte subpopulations should be performed in future studies to confi rm this hypothesis.
The differentiation process from mesenchymal stem cells to mature adipocytes is not synchronized, and various cell populations that are at different stages of differentiation are present at a given time point. To date, these intermediate cell populations have been poorly described, mainly due to lack of the necessary tools required to distinguish between the different populations, and this is therefore a limiting factor in fully understanding the different stages of adipogenesis ( 23 ). Currently two main populations are studied in the adipose differentiation process, i.e., preadipocytes and mature adipocytes. At a cellular level, investigators distinguish between these two cell populations based on cellular complexity and lipid content. Preadipocytes do have a lower degree of cellular complexity and contain less total lipid than mature adipocytes that display an increased level of cellular complexity and contain larger intracellular lipid droplets. Our data suggest that adipocytes can be classifi ed into various subpopulations according to their CD36 expression profi les and intracellular neutral lipid content. In general, BDP seems to be more sensitive in detecting changes in intracellular lipid content, but NR's unique fl uorescence emission spectrum has the potential to allow for the identifi cation of a more mature adipocyte phenotype.
Various investigators have shown that preadipocytes and mature adipocytes differ in their physiological function, levels of protein expression, etc. ( 11,12,19 ). It has also been shown that infl ammation infl uences the ratio of preadipocytes to mature adipocytes in adipose tissue. However, the role and magnitude of intermediate adipocyte populations in infl ammatory conditions is largely unknown. In this study, we provide a means of identifying and studying intermediate adipocyte populations. Our data clearly show that it is possible to monitor the distribution of various adipocyte subpopulations when several