| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Division of Neonatology (L.M.S.), Children’s National Medical Center, George Washington University School of Medicine, Washington, D.C. 20010; BioWhittaker Inc. (M.K.), Walkersville, Maryland 21793; and Diabetes Branch (M.M., D.L.), National Institute of Diabetes & Digestive & Kidney Diseases, National Institutes of Health, Bethesda, Maryland 20892-1758
Address all correspondence and requests for reprints to: Derek Le Roith, M.D., Ph.D., Chief, Diabetes Branch, Room 8D12, Building 10, National Institutes of Health, MSC 1758, Bethesda, Maryland 20892-1758. E-mail: derek{at}helix.nih.gov.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
IR-3 antibody, which inhibits IGF-I receptor activity. We also describe the pattern of differentiation with regard to cell growth, lipid accumulation, and morphologic changes and define the changes in these parameters that are influenced by IGF-I. Finally, peroxisome proliferator activating receptor-
immunoreactivity was also increased in response to IGF-I, and this effect was blocked in cells treated with the IR-3 antibody. Taken together, these findings suggest that IGF-I plays a critical role in adipocyte differentiation and lipid accumulation. |
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Through the use of various experimental models, including immortalized cell lines (reviewed in Ref. 4), adipose tissue-derived precursors (5), whole animals (6), and patients (7), substantial progress has been made in our understanding of the process of adipocyte differentiation. At the transcriptional level, interactions between the CCAAT/enhancer binding protein (C/EBP) and the peroxisome proliferator activating receptor (PPAR) families of transcriptional regulators have been shown to play pivotal roles during adipocyte differentiation. In an early response to differentiation factors, the transcription factors C/EBPß and C/EBP
are increased. Expression of C/EBPß and C/EBP, in turn induce the expression of PPAR. In the presence of ligand, PPAR interacts with the retinoic acid receptor X to induce C/EBP, which in turn acts in concert with one or more of these transcription factors to induce adipocyte-specific genes (8).
In most adipocyte model systems, multiple cell signaling factors are required to induce differentiation. These factors include glucocorticoids, agents that modulate cAMP levels, and, in some models, such as 3T3-L1 cells, micromolar levels of insulin. This suggests that the IGF-I receptor (IGF-IR) may play a critical role in this process because micromolar concentrations of insulin can act through IGF-IRs (9). In undifferentiated 3T3-L1 cells, IGF-IR expression predominates, whereas insulin receptor (IR) expression increases only as differentiation progresses, inducing the formation of hybrid receptors (10). These and other data have revealed a great deal about the effect of IGF-I and insulin on adipocyte differentiation. However, the specific mechanisms by which IGF-I or insulin stimulate differentiation of adipocytes have not been clarified.
Recently bone marrow-derived human mesenchymal stem cells (HMSCs) have been developed as a new model system to study adipocyte differentiation. These primary cells are obtained from healthy human volunteers, are uniform in type, and retain the ability to undergo multiple (but not infinite) cell divisions. HMSCs can be induced to differentiate into adipocytes, chondrocytes, and osteocytes in vitro (11), as well as other cell lineages in vivo, including thymic cells, cardiac and skeletal muscle cells, glial cells, and others (12). In the series of states of differentiation that exist between the undifferentiated multipotent embryonic stem cell and the fully differentiated adipocyte, these cells are further removed than the typical primary precursor cell derived from fat tissue. When transplanted into fetal sheep marrow, HMSCs differentiate and incorporate into many normal adult tissues, including adipose tissue (12). It is likely that some of the cells in adult adipose tissue originate from cells derived from this pool. Thus, delineating the pathways by which HMSCs differentiate could help us to understand how adipose tissue is generated by this mechanism in vivo. Relatively little is known about what determines the phenotype of these cells or what regulatory factors may be involved in differentiation of HMSCs.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
HMSCs, prescreened fetal bovine serum (FBS), and growth media were obtained from Cambrex (previously BioWhittaker, Walkersville, MD). DMEM was from Invitrogen (Carlsbad, CA). Antibodies were obtained as follows: IGF-IR antibody (Ab-1 or IR3 antibody) from Calbiochem Immunologicals (San Diego, CA), anti-IRß subunit (IRß, C-19), anti-IGF-IRß subunit (IGF-IRß, C-20), anti-PPAR (H-100) and nonspecific rabbit and mouse IgG were all from Santa Cruz Biotechnology (Santa Cruz, CA). [125I]IGF-I was obtained from Amersham Biosciences, Inc. (Piscataway, NJ). Insulin was from Sigma (St. Louis, MO), IGF-I was from Genentech (South San Francisco, CA) and LongR(3)IGF-I was from GroPep Ltd. (Adelaide, Australia).
Cells and cell culture conditions
HMSCs were obtained from a single donor bone marrow aspirate by density gradient centrifugation and selective culturing techniques (11). Cells were expanded according to an established protocol (BioWhittaker/Cambrex). In brief, cells were quick thawed and suspended in HMSC growth medium (HMSCGM), which includes a proprietary basal medium, 10% prescreened FBS, 200 mM glutamine, 25 U penicillin, and 25 µg streptomycin per 500 ml. Cells were grown to 90% confluence in 75-mm tissue culture flasks at 37 C in 5% CO2/95% air (Corning Inc., Acton, MA). Cells were then detached from the flask surface with 0.25% trypsin for 2 min, washed in HMSCGM, and replated at a density of 7000 cells/cm2 (
4:1 split). Cells underwent two further expansions before aliquots were stored in 40% FBS, 50% HMSCGM, and 10% DMSO and frozen at –70 C for further experiments.
Cell differentiation
Cells were plated in tissue culture plates at a density of 21,000 cells/cm2 and grown to confluence in HMSCGM. To induce differentiation, the medium was switched to DMEM with 10% prescreened FBS, 200 mM glutamine, 25 U penicillin, and 25 µg streptomycin per 500 ml (DMEM basal medium) plus 1 µM dexamethasone, 0.2 mM indomethacin, and 0.5 mM 3-isobutyl-1-methylxanthine (IBMX) (hereafter this mix of three differentiation factors is referred to as D), plus 0.01 mg/ml (1.7 µM) insulin. The control medium included each of the above components except D. The protocol for full differentiation required three phases of induction, during which the cells were exposed to the full differentiation medium (with D), interspersed with two maintenance phases, when the cells were exposed to DMEM basal medium plus insulin only (without D). Each induction phase lasted approximately 4 d, and each maintenance phase lasted approximately 2 d.
Induction of differentiation by replacing insulin with IGF-I
To determine whether IGF-I could replace insulin in the induction of differentiation, various treatment conditions were compared, as described in the legend to Fig. 1
. When insulin, IGF-I, or IGF-IR blocking antibody was included in the treatments, they were used consistently throughout, in both the treatment and maintenance phases of the experiment. The IR-3 antibody (1 µg/ml), which competitively inhibits activation of the IGF-IR, was included in some experiments to control for activation of the IGF-IR by endogenous IGF-I. In some experiments the IR3 antibody at the same concentration was present in combination with either IGF-I or insulin to assess its ability to block pharmacologic levels (20 nM) of ligand. Initial experiments were with a D-only control for IGF-I activity and IR3 effects. To rule out nonspecific immunoglobulin effects, we compared D only-treated cells with cells treated with D plus IgG. Observing no difference between the conditions, we used the D plus IgG as the more comprehensive control in subsequent experiments.
|
Cells were cultured and differentiated as described above, washed twice with ice cold PBS, and snap frozen. At the time of analysis, cells were suspended in PBS or a solution of Nile Red (2.5 µg/ml) in PBS in a volume sufficient to cover the cells. Cells were viewed in an Axiophot inverted microscope (Carl Zeiss Inc., Thornwood, NY). Images were captured with a PentaMAX camera (Princeton Instruments Inc., Trenton, NJ) and IP Labs software (Scanalytics, Inc., Fairfax, VA) and processed with Adobe Photoshop (Adobe Systems Inc., Mountain View, CA).
Lipid accumulation
HMSCs were differentiated as described above. At the end of the treatment period, plates were placed on ice and washed twice with ice-cold PBS. Excess PBS was removed and cells were stored sealed with Parafilm at –20 C before assay. Nile Red (Molecular Probes, Eugene, OR) was dissolved in DMSO at a concentration of 1 mg/ml and stored at –20 C. To assess lipid content, Nile Red stock was diluted to 2.5 µg/ml in PBS, and 200-µl aliquots were placed in each well. Cells were then kept at 4 C to equilibrate for 2 h. Fluorescence was then determined in a fluorescent plate reader (Molecular Dynamics, Sunnyvale, CA) with an excitation wavelength of 485 and emission wavelength of 595 angstroms with six samplings per well at a fixed gain of 60. A low level of background fluorescence was observed even in the absence of lipid droplets, and this background was proportional to the number of cells in each well. This background (calculated on the basis of cell number) was subtracted from all readings before calculating the lipid content of cells.
CyQuant assay for DNA content
DNA determinations were performed on duplicate plates to those used in the lipid assay described above. The CyQuant assay (Molecular Probes) was performed according to the manufacturer’s protocol. Plates were read in a Molecular Dynamics fluorescent plate reader with an excitation wavelength of 485 and emission wavelength of 535 Å with nine samplings per well at a fixed gain of 60. Background was subtracted from raw values before analysis for each treatment condition.
Western blot analysis
HMSCs were grown on 100-mm tissue culture plates (Corning). At various time points in the treatment protocol, cells were harvested in 0.7 ml of ice-cold extraction buffer (containing 50 mM Tris, 0.5% Nonidet P-40, 1 mM dithiothreitol, 0.3 M sodium chloride, 0.1 mM EDTA, 10% glycerol, plus protease inhibitors). Samples were then sonicated, and aliquots containing 20 µg of protein extracts were subjected to SDS-PAGE on 8% polyacrylamide gels. Samples were then transferred to nitrocellulose membranes, and the membranes were blocked with 5% nonfat milk in TBS-Tween 20 buffer [20 mM Tris-HCl (pH 7.6), 130 mM NaCl, 0.1% Tween 20] for 1 h at room temperature. Membranes were then incubated with the corresponding primary antibody in TBS-Tween 20 buffer with 5% nonfat milk for 16 h. After one rinse and three 30-min washes, the membrane was incubated with a horseradish peroxidase-coupled secondary antibody (1:5000) in the hybridization buffer. After washing, immunoreactive bands were visualized using the ECL-plus detection kit (Amersham Biosciences, Inc.), according to the manufacturer’s instructions.
Detection of IR and IGF-IR hybrids
Cells were grown to confluence in 100-mm tissue culture plates and then grown under two conditions. Control cells, in DMEM, 10% FBS, and IGF-I (IGF-I only), and differentiated cells in DMEM 10% FBS plus the D mix of factors plus IGF-I. After the relevant treatment, cells were extracted with 300 µl lysis buffer containing Tris 10 mM, NaCl 150 mM, NaF 100 mM, Na vanadate 10 mM, Na pyrophosphate 10 mM, EDTA (pH 8) 1 mM, EGTA 1 mM, Triton X-00, 1%, Nonidet P-40 1%. Extract was placed on a rotary wheel at 4 C for 1 h and then centrifuged at 4 C at 13,000 rpm for 15 min in a Sorval refrigerated bench top centrifuge. Then 250-µl aliquots of the supernatant were diluted 1:1 with H2O and precipitated with anti-IGF-IR antibody and rotated at 4 C for 24 h. Fifty microliters of protein A Sepharose beads were added and rotated at 4 C for 16 h. Samples were then centrifuged at 4 C at 13,000 rpm and the supernatant discarded. Beads were washed with ice-cold lysis buffer/H2O (1:1) three times before the final wash was discarded. Beads were suspended in 20 µl reducing protein loading dye, heated to 95 C for 5 min subjected to 8% SDS-PAGE and Western blotted onto nitrocellulose membranes as described above. Membranes were first immunoblotted for IR, stripped with Ponceau stain, and then immunoblotted for the IGF-IR.
Scatchard analysis
HMSCs were grown to confluence in 12-well tissue culture plates (Costar, Cambridge, MA) in preparation for competitive binding studies and Scatchard analysis (10). Three different analyses were carried out. First, we assessed IGF-I binding sites; second, we determined whether nonreceptor binding sites, presumably IGF binding proteins (IGF-BPs), were present; and third, we assessed insulin binding. Cells were treated either with 20 nM IGF-I only (control) or D-plus-IGF-I (differentiated) through three full rounds of induction and maintenance, as described above. Culture plates were then placed on ice and washed twice with ice-cold PBS. For the first two analyses, HMSCs were then incubated with [125I]IGF-I in the presence or absence of either unlabeled IGF-I, which binds both IGF-IR and IGF-BPs, or LongR(3)IGF-I, which has high affinity binding only for IGF-IR. Each of these ligands was aliquoted in a graduated series of concentrations for displacement of the IGF-I tracer. This experiment was run in parallel with NWT B3 cells, which were engineered to overexpress IGF-IRs. The assay buffer included 100 mM HEPES (pH 7.9), 150 mM NaCl, 0.05 mM KCl, 1 mM MgCl2, 1 mM EDTA, 5 mg/ml insulin-free BSA. Cells were then washed with ice-cold PBS to remove free IGF-I. Cells and bound IGF-I were harvested in 0.5 ml of 0.4N NaOH, and radioactivity was determined in these samples with a Cobra II -counter (Packard Instrument Co., Downers Grove, IL). For the insulin binding, parallel analyses were carried out on duplicate plates of HMCSs and in 3T3-IR cells (which overexpress IRs) with [125I]insulin as tracer and unlabeled insulin as displacer.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
HMSCs can be induced to differentiate into adipocytes upon exposure to a set of specific factors. The adipocyte induction medium includes dexamethasone, IBMX, indomethacin (mixture D), and insulin at micromolar concentrations (11). To determine whether activation of the IGF-IR induces differentiation of adipocytes, we compared the effects of various nanomolar concentrations of IGF-I (D plus 2 nM, 20 nM, or 200 nM IGF-I) vs. 1.7 µM insulin (D plus insulin). Phase-contrast microscopy was used to assess lipid accumulation at the end of the third induction phase, as shown in Fig. 1
. Exposure to D plus IGF-I at all three nanomolar concentrations induced adipocyte differentiation of HMSC cells to a degree similar to that induced by D plus 1.7 µM insulin. Cells in the D only-treated group exhibited a consistent reduction in lipid accumulation, compared with D plus insulin or D plus IGF-I treatment. In contrast, neither insulin nor IGF-I treatment-induced lipid accumulation in the absence of mixture D. Cells in the insulin-only group maintained a fusiform mesenchymal cell appearance, whereas all cells exposed to mixture D underwent morphological changes. HMSCs exposed to D plus IGF-I at all three concentrations or D plus insulin developed lipid droplets of similar densities and sizes. Cells exposed to mixture D alone exhibited fewer cells with well-developed lipid droplets. In subsequent experiments, D plus IGF-I treatment always included 20 nM IGF-I.
To further evaluate the relative role of IRs and IGF-IRs in the differentiation of HMSCs into adipocytes, we used binding assays to determine the number and affinity of IRs and IGF-IRs in undifferentiated (IGF-I only) HMSCs and differentiated (D plus IGF-I) HMSCs at the end of the induction treatment period. As judged by displacement of [125I]IGF-I tracer by IGF-I (Table 1), the levels of IGF-IR binding were relatively high and the dissociation constant (KD) values were similar to both those of the NWT B3 cells included in the assay (Table 1) and those reported for IGF-IRs in 3T3-L1 cells (10). Differentiation did not significantly alter either the binding site or KD of IGF-IRs in HMSC cells. A duplicate experiment yielded equivalent results. The presence of the IGF-IR in both differentiated and undifferentiated states was confirmed by Western blot (Fig. 2). However, there was only partial displacement of IGF-I tracer by LongR(3)IGF-I even at the highest concentrations (legend to Table 1), suggesting that most of the IGF-I binding was to IGF-BPs and not to an IGF-IR. For NWT B3 cells the 125I-IGF-I tracer was displaced as would be expected by both IGF-I and LongR(3)IGF-I, with IGF-I having roughly a 3-fold higher affinity than the LongR form. Parallel binding studies were carried out using 125I-insulin as tracer and insulin for displacement. 3T3-IR cells, which overexpress the IR, exhibited robust insulin binding and yielded the predicted results in Scatchard analyses. No appreciable insulin binding was detected in differentiated or undifferentiated HMSCs, even in the absence of competing unlabeled ligand. However, IR protein was detected by Western blot (Fig. 2).
|
|
IGF-IR blocking antibody Ab, IR3, suppresses adipocyte lipid accumulation
Accumulation of lipids was consistently lower in cells exposed to D only, compared with cells treated with D plus IGF-I. However, certain morphological changes (rounding up) characteristic of adipocyte differentiation and some lipid accumulation were induced by exposing HMSCs to D only. It was possible that these differentiation-related changes observed in the absence of exogenous IGF-I in cells treated with D only could be due to traces of IGF-I (either derived from the 10% FBS in the culture medium or by endogenous production). To test this, cells were treated with D and the IGF-IR blocking antibody IR-3 (D plus Ab). To permit more sensitive and precise comparisons than afforded by phase microscopy or even Oil Red O staining (not shown) for microscopy, the neutral lipid binding fluorescent dye Nile Red (13, 14) was used for both fluorescence microscopy and a fluorescence-based assay of lipid content. Figure 3 (upper panel) shows fluorescent microscopy of cells exposed to various treatment conditions. The IGF-I only-treated cells exhibited low-grade diffuse background cytosolic staining with Nile Red, whereas D plus IGF-I treated cells exhibited bright staining of individual cells with densely clustered lipid droplets of varying sizes. In contrast, HMSCs exposed to D plus the IR-3 antibody showed limited numbers of very small lipid droplets. Cells treated with D only and cells treated with D plus IgG were indistinguishable and intermediate between cells treated with D plus IGF-I or D plus IR3 Ab, with regard to the density and size of lipid droplets. Inhibition of the IGF-IR seems to limit lipid accumulation but not the proportion of cells that initiate droplet formation.
|
. Cells were grown and treated through the third induction phase in 96-well plates, stained with Nile Red, and read in a fluorescent plate reader as described in Materials and Methods. Consistent with the fluorescent microscopy results, cells treated with IGF-I only showed low-level background staining. Of the cells treated with the D mix of factors, those treated with D plus IGF-I showed the greatest lipid accumulation, those treated with D plus Ab showed the least (about 70% reduction in lipid content), and those treated with D plus IgG (or D only) were intermediate (25–30% reduction from D-plus-IGF-I). The overall ANOVA (the four conditions shown) was highly significant (P < 0.0001), and differences between all of the experimental groups reached significance by the Bonferroni t test for multiple comparisons. When lipid accumulation in response to 20 nM IGF-I and 20 nM insulin was compared quantitatively, there was 20% less lipid accumulation in response to insulin (P < 0.001) and D plus insulin had 15% more than was seen in D plus IgG-treated cells (Fig. 4). Whereas it is clear from the above that the IR3 Ab at 1 µg/ml can suppress endogenous IGF-I activity (compare D plus IgG and D plus IR3 Ab), adding IGF-I at 20 nM was able to overcome this inhibition (D plus IGF-I and Ab in Fig. 4). However, insulin at 20 nM did not completely overcome the ability of the antibodies to inhibit lipid accumulation (D plus insulin and Ab in Fig. 4). To block the IR, we used serum from a patient with insulin resistance due to the presence of IR blocking antibody. This serum had previously been found to inhibit receptor at a 1:500 dilution. When this receptor was added at a dilution of 1:100 to D plus IGF-I or D plus insulin conditions, it had no suppressive effect (data not shown).
|
We next used the Nile Red-based fluorescence assay to progressively follow lipid accumulation as HMSCs differentiated into adipocytes. Exposure to IGF-I alone had little effect on lipid accumulation at any of the time points examined (data not shown). However, in each the three groups treated with mixture D, there was an increase in lipid accumulation, starting after the first induction phase. Lipid accumulation increased most rapidly in HMSCs treated with D plus IGF-I, with slower accumulation in HMSCs treated with D only and even slower lipid accumulation in cells exposed to D plus the IR3 Ab.
IGF-IR activation promotes cell division in HMSCs
Cell division is often observed before adipocyte differentiation, but it is uncertain whether this cell division is required for differentiation (15, 16). To determine whether HMSCs undergo cell division before differentiation and to assess the importance of IGF-I in regulating such division, cell proliferation was progressively assayed at various times during differentiation. Replicates of the 96-well plates were assayed at different time points for DNA content and the results are shown in Fig. 5 (upper panel). Cellular proliferation was observed at the first induction phase (light gray bars), and proliferation reached a plateau after the first maintenance (hatched bars) or second induction phases (dark gray bars) in all treatment conditions. The two groups treated with exogenous IGF-I showed the largest increase, with a 61% increase for the IGF-I-only group and a 65% increase for the D plus IGF-I group in the course of the experiment, suggesting less than a single cell division per cell on average took place. In the 3T3-L1 cell model, two to three rounds of replication can be seen. The addition of IGF-I induced growth equally well in the presence or absence of the D mix. This increase was greater than that seen for the D-only group, which in turn was greater than that seen for the D plus Ab group, again suggesting an endogenous activity. At the end of the third induction, the ANOVA for the four treatment conditions was highly significant (P < 0.0001), and all individual comparisons, except for that between the IGF-I only and D plus IGF-I groups, reached significance (P < 0.05) by the Bonferroni t test for multiple comparisons.
|
Lipid content per cell is regulated by IGF-I
To determine the extent to which differences in lipid accumulation were due to changes in the number of cells that were differentiating, we calculated the concentration of lipid per cell or per unit DNA. These results are shown in Fig. 5 (lower panel). In each treatment group, cellular proliferation preceded lipid accumulation. Therefore, the concentration of lipids stayed level or fell during the first induction phase (light gray bar). There was little recovery in the cells treated with IGF-I only, as expected from the low level of lipid accumulation under these conditions. In the cells exposed to mixture D, the rate of lipid accumulation increased as proliferation reached a plateau, resulting in increased lipid concentrations only at the end of induction 2 (dark gray bar). As expected, the highest levels of lipids were detected in cells exposed to D plus IGF-I, followed by D only and D plus the IR3 Ab. The differences in lipid concentration per unit DNA were smaller than the differences in total lipid accumulation (data not shown) because IGF-I promotes both cell growth and lipid accumulation. The difference between the D plus IGF-I and D-only groups narrowed considerably (30% decrease in total lipid levels vs. a 14% decrease in lipid per cell) but remained statistically significant. At the end of the third induction, the ANOVA for the four conditions was highly significant (P < 0.0001), and all individual comparisons reached significance (P < 0.05) by the Bonferroni t test for multiple comparisons. This experiment was performed four times, each with similar results. In each of two replications, replicate plates were assayed for DNA and lipid. In two later replications, the method had been worked out so that DNA and lipid assays were performed in series on the same plate, allowing more direct estimates of lipid per cell changes. The latter showed that in reference to the D-plus-IGF-I group, a decreases of about 20% was seen when exogenous IGF-I was removed (D only), and a decrease of about 60% was seen when IGF-I was replaced by the IR3 Ab (D-plus-Ab; data not shown).
Treatment with IR3 decreases phosphorylation of AKT, a downstream target of IGF-IR activation
To document that IR3 was having the intended effect of blocking IGF-IR activation, we examined the phosphorylation state of AKT/protein kinase B, thought to be a key element in the IR and IGF-IR signaling pathways regulation of adipocyte differentiation (Fig. 6). Unlike the usual protocols for studying phosphorylation in signaling cascades, these cells were not serum starved before stimulation. They were maintained in the same 10% FBS containing treatment media used in all of our differentiation experiments. The conditions were the same: IGF-I only, D plus IGF-I, D plus Ab, or D plus nonspecific IgG. In one experiment cells were incubated for 30 min in IGF-I only and D plus IGF-I and for 2 h in D plus IR3 and D plus IgG before being harvested. In another experiment all groups were harvested at about 4 h of incubation; the results were indistinguishable. Figure 6 shows a representative blot. Looked at as the absolute number of densitometric units of phosphorylated AKT and normalizing the D plus Ab group as 1, the D plus IgG group was 2.0-fold, the D plus IGF-I group 2.8-fold, and the IGF-I only group 5.4-fold greater. Total AKT levels were as high or higher under treatment conditions that suppressed phosphorylation. Normalized first to total AKT, the relative numbers for phosphorylation of AKT remain similar: D plus Ab (1), D plus IgG (2.1), D plus IGF-I (3.5), and IGF-I only (6.6). Therefore, even in the context of 10% FBS, the IR3 Ab decreased and IGF-I increased the level of AKT phosphorylation. Unexpectedly, phosphorylated AKT signal for the IGF-I-only group was always about 2-fold greater than for the D plus IGF-I group. This finding suggests that one or more components in D depress AKT phosphorylation induced by IGF-I.
|
The DNA and lipid accumulation profiles shown in Fig. 5 suggest that cell division precedes differentiation. However, differentiation is a multifaceted process, and lipid accumulation occurs relatively late during differentiation. Figure 7 shows representative phase-contrast images of cells exposed to various treatments approximately 36 h into the first induction phase of the differentiation protocol. The HMSCs treated with IGF-I alone maintained their mesenchymal appearance, similar to what was observed at the end of the third induction phase, as shown in Fig. 1. In contrast, the cells treated with D plus IGF-I exhibited an early differentiation phenotype, namely rounding up. All cells treated with the D mix of factors underwent this rounding-up process, and IGF-I had no discernible effect on it. Lipid droplets were never detected during the first 48–72 h of induction and were only rarely observed at 96 h. Between the end of the first maintenance phase and the middle of the second induction phase (5–9 d), discernibly greater numbers of sparse and very small droplets were observed in HMSCs exposed to D plus IGF-I, compared with cells treated with D plus the IR3 Ab (data not shown). If the D plus IGF-I-treated cells are cultured in maintenance medium for several weeks after the third induction phase, the cells continue to accumulate lipid and the droplets increase in size, with a few cells having only two or three large droplets (data not shown).
|
Inhibition of IGF-IR activation by the IR3 blocked adipocyte differentiation, as determined by lipid accumulation (Figs. 3–5). We next asked whether this effect was mediated by alterations in the expression of the PPAR family of transcription factors. Levels of this transcription factor were evaluated at the end of the first induction phase because the differentiation process that precedes lipid accumulation is well underway at this time (see Fig. 7). Western blot analysis showed that very little PPAR is present at time 0 (Fig. 8). Treatment with D plus IGF-I increased levels of PPAR isoforms 1 and 2 at 24 h in some experiments (data not shown), but this occurred more strongly and consistently at 96 h after initiation of the first induction phase (Fig. 6). At 96 h PPAR1 increased for all cells treated with D set of factors. The addition of IGF-I tended to increase and the IR3 Ab tended to decrease PPAR1, but these changes were small (10–20%). In some experiments treatment with IGF-I alone resulted in detectable levels of the PPAR1 but never the PPAR2 isoform (five experiments). An increase in PPAR2 seemed dependent on the presence of both D and IGF-IR activation. Using arbitrary densitometric units and normalizing to the actin signal, the values for PPAR2 in Fig. 8 were: IGF-I only, 0.1; D plus IGF-I, 2.0; D plus Ab, 0.5; and D plus IgG, 2.6. In four of the five experiments we performed, PPAR2 is higher in D plus IGF-I than in D plus IgG. Thus, regulation of PPAR immunoreactivity parallels the cell growth and lipid accumulation induced by the various treatment conditions and likely mediates the latter.
|
). |
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
HMSCs are primary cells derived from adult bone marrow that can differentiate into adipocytes, chondrocytes, or osteocytes in vitro through manipulation of culture conditions in the absence of contact with any other cell type (11). When 6 million to 20 million HMSCs were transplanted into immune tolerant fetal sheep, the cells survived for at least 13 months, were broadly distributed, and gave rise to mature cells that appeared to have a normal phenotype in several tissues, including liver, cardiac muscle, skeletal muscle, bone marrow stroma, adipose tissue, and others (12). Interestingly, the transplanted HMSCs were resistant to rejection even when the transplant occurred after immune competence developed, raising the possibility that these cells may be of considerable clinical value. Two recent articles (17, 18) identified cell fusion as the actual cause of what had appeared to be trans differentiation in coculture of adult and embryonic stem cells. They suggested that other cases of transdifferentiation could also be due to fusion. However, this has so far been seen as an in vitro phenomenon of low frequency with the less differentiated phenotype being dominant (19). Moreover, in the studies reported here, a uniform population of cell was studied, and differentiation was driven by exogenous factors. In complex systems (i.e. in vivo), controls for fusion would be appropriate, at least until its relevance is better defined.
It is possible that the endogenous counterparts of HMSCs serve as one source of precursors for normal adipose tissue in vivo. Developmentally, HMSCs are between the far removed embryo-derived cell lines (such as 3T3-L1 cells) and the proximal adipose tissue-derived preadipocytes that have been used as models for studying the process of differentiation. The time required for HMSCs to differentiate into adipocytes is longer than in other cell models, although it is not known whether this is due to the fact that these are stem cells or that they are human cells, as opposed to mouse cells. In both 3T3-L1 cells (20) and mouse embryo fibroblasts (3), full differentiation occurred within 6–8 d of treatment. In HMSCs, very small and sparse lipid droplets were detected within 5–8 d, but abundant lipid accumulation occurred only between the end of the second maintenance phase and third induction phase, after 12–16 d of treatment (see Figs. 3 and 5). We characterized the process of adipocyte differentiation in these cells and the role played by IGF-I in this process.
The media supplements required for HMSCs to differentiate into adipocytes include glucocorticoid, IBMX, indomethacin, and high levels of insulin (1.7 µM). Similar concentrations of insulin are required in various other protocols for adipocyte differentiation in other model systems (4). In an early paper describing the 3T3-L1 cell model (21), IGF-IRs were found to be abundant, whereas IRs were sparse in undifferentiated cells; however, the level of IGF-IR remained constant, whereas IR levels increased by about 25-fold in differentiated cells. In a later paper (9), the same group demonstrated that pharmacological doses of insulin were required because it was acting through the IGF-IR and not the IR. In those experiments, it was shown that IGF-IRs with characteristic binding affinity were present at about the same modest level (13,000 receptors/cell) both before and after differentiation and that even at differentiation (d 6), IGF-I induced lipogenic enzymes with 40- to 60-fold more potency than insulin. A more recent study (10) reported that 3T3-L1 cells express larger numbers of IGF-IRs (about 40,000 receptors/cell) and that IGF-IR/IR hybrid receptors formed as the IR protein levels increased. This was accompanied by a change in the KD from 0.9 nM in undifferentiated cells to 3.6 nM in differentiated cells.
In comparison we found a very large number of high-affinity IGF-I binding sites in both undifferentiated and differentiated HMSCs. Using LongR(3)IGF-I, we were able to determine that the majority of these sites were probably IGF-BPs. The affinity of IGF-IR/IR hybrids for LongR(3)IGF-I is unknown, but the possible number of hybrid receptors as judged by our coimmunoprecipitation experiments was far too small to account for the number of binding sites found (400,000/cell). NWTB3-positive control cells showed the expected IGF-IR binding and displacement for both IGF-I and LongR(3)IGF-I, with complete displacement at high concentrations of LongR(3)IGF-I. We were unable to demonstrate insulin binding by HMSCs using methods that yielded the expected binding and displacement in the 3T3-IR-positive control cells. In comparison, insulin binding was undetectable in both undifferentiated and differentiated HMSCs but was abundant in 3T3-IR cells that were engineered to overexpress the IR. Although IR binding could not be demonstrated, IR protein was observed by Western blot (not shown).
Both IGF-I and, to a lesser extent, insulin had biological activity, and both IGF-IR and IR proteins were detectable on Western analysis in differentiated and undifferentiated cells. It is possible that IRs are expressed but do not reach the cell surface in numbers that allow detection by tracer displacement methods. We demonstrated that features of the adipocyte such as rounding up, lipid droplet formation, and lipid droplet growth in size are separable. Perhaps cell surface receptor repertoire is also such a feature, so that newly differentiated HMSCs have a shape and lipid content typical of other adipocytes differentiated in vitro, whereas its cell surface receptor repertoire is different in its paucity or lack of IRs. Perhaps such a cell would remain incomplete as an adipocyte with regard to this feature, or perhaps it will develop an abundance of IRs over time. We did note that the lipid droplets continued to increase in size and decrease in number over time. Whereas 3T3-L1 cell-derived adipocytes maintain multiple small droplets, true adipocytes that develop in vivo have one or two very large droplets in their cytoplasm.
Low concentrations of IGF-I (2 nM) induced differentiation in HMSCs that was indistinguishable from that induced by an approximately 1000-fold higher dose of insulin (1.7 µM) and by 20 nM IGF-I. This suggests that we are operating close to the plateau for responsiveness to IGF-I at 2 nM. Residual activity in the media not containing added IGF-I became apparent only after receptor activity was inhibited with IR3 Ab, which blocks IGF-IR activation. Subsequently this antibody was used to better characterize the influence of IGF-I in the differentiation process. Finally, in presence of the D mixture, 20 nM IGF-I, which selectively acts on IGF-IRs, induced significantly more accumulation of lipids than did 20 nM insulin, which selectively acts on IRs. Therefore, it is reasonable to conclude that whereas much of the impact of insulin and IGF-I on growth and differentiation in HMSCs occurs through binding to the IGF-IR, insulin seems to have some independent activity. We do not exclude the possibility that there are IRs on the cell surface that are in numbers that are below the threshold of detection by a receptor binding assay but still able to mediate biologically activity.
Typically, adipocyte precursors are grown to confluence, at which point they stop dividing due to contact inhibition. On exposure to differentiation factors, these precursors undergo cell division again in a process called clonal expansion. In the 3T3-L1 cell model, the cells generally undergo two to three cell divisions (16). Because treatments that inhibit these cell divisions also tend to inhibit differentiation, it has been argued that these cell divisions may be required for subsequent differentiation (15).
IGF-I was believed to be necessary for both clonal expansion and differentiation. The IGF-IR when activated leads to receptor autophosphorylation on specific receptor tyrosine residues and then phosphorylation of a number of other cellular proteins. This initiates a set of well-described signaling cascades that involve changes in protein-protein interactions through SH2/SH3domains as well as activation/suppression of a number of cellular enzymes and transcription factors (22). Two major branches of the signaling cascade initiated by the IGF-IR are the MAPK/ERK and the phosphatidylinositol 3-kinase (PI3K) branches. It is emerging that during adipocyte differentiation, stimulation of the MAPK branch of the cascade induces cell division in the cell cycle-arrested confluent cells. Boney et al. (20, 23) showed that IGF-I treatment of undifferentiated 3T3-L1 cells resulted in MAPK activation through the Shc/GRB2 pathway intermediates and that this resulted and cell division. Identical treatment of already differentiated cells did not show this MAPK activation and cell division did not occur. This was supported by subsequent findings also in the 3T3-L1 system that showed that blocking MAPK/ERK pathway with PD98059-blocked clonal expansion but not differentiation (16). In those studies the authors seriously challenged the link between clonal expansion and differentiation, showing each could occur in the absence of the other. The PI3K branch of IR/IGF-IR signaling was definitively linked to differentiation in a model system that used mouse embryo fibroblasts from insulin receptor substrate (IRS)-1–/–, IRS-2–/–, and IRS-1–/–-IRS-2–/– knockout embryos (24). It was shown that the degree of differentiation seen in response to treatment with differentiation factors and insulin (5 µM, so again likely via the IGF-IR) was proportional to PI3K activation, which in turn was dependent on the availability of IRS-1 and to a lesser extent IRS-2 to act as substrate for the IGF-IR/IR. Experiments that used the PI3K inhibitor LY294002 reinforced this finding. MAPK was not influenced by the IRS knockouts.
In other model systems, the pathways mediating the IGF-I effect on growth and differentiation seem to be reversed, with the MAPK/ERK pathway mediating differentiation and the PI3K pathway mediating cell division and suppressing differentiation (25). Whereas Miki et al. (24) showed definitively that IRS-1 or IRS-2 were required for adipocyte differentiation, Morrione et al. (25) showed suppression of IRS-1 is required for neuronal differentiation. Insightful work on neuronal differentiation has established that there are complex interactions between the AKT and MAPK pathways. For example, inhibitors of one PI3K/AKT pathway increases tyrosine phosphorylation of IRS-2 but does not increase MAPK activity or differentiation (26). There are newly described kinases such as salt-inducible kinase-2 found in adipocytes that are regulated by insulin in what appears to be a negative feedback modulation of the IRS-1 pathway (27)
Such observations force one to think beyond linear pathways in which proximal inhibition influences one branch of a pathway exclusively. It is important to remember that differentiation into two disparate cell types, even from the same stem cell lineage, involves sequential activation of disparate sets of specific genes that determine the specific phenotype. Each set may have specific, even opposite requirements for activation that are downstream of the regulatory pathways of signaling kinases (28)
Our studies of HMSCs to date do not speak to such signaling complexities. They had the more modest purpose of establishing the extent to which IGF-IR activation was necessary for adipocyte differentiation, which features of the differentiated phenotype were under IGF-IR control, and how IGF-IR action in these cells compared with that seen in other model systems. To those ends we have demonstrated that IGF-I had an effect on both cell growth and certain aspects of differentiation in a way consistent with other model systems. The data unequivocally demonstrated that postconfluent cell growth was dependent on the level of IGF-I R activity (because cell growth was highest in cells treated with D plus IGF-I and lowest in cells treated with D plus the IR-3 Ab) and independent of the presence of the differentiation factors. A similar pattern was recently reported in 3T3-L1 cells (16).
The impact of IGF-I on differentiation is complex because differentiation is a gradual and multifaceted event. Cell division in the first 2–3 d was regulated by IGF-I, but the concurrent morphological changes (Fig. 4) were not because there was no distinguishable difference between cells treated with D plus IGF-I and those treated with D plus Ab at that time. The formation of lipid droplets began at some point between d 4 and 6, and IGF-I accelerated this process. Lipid accumulation progressed slowly at first and then the rate increased during the second induction. This process was critically dependent on IGF-I. Cells treated with D plus IR3 Ab had the ability to form rudimentary lipid droplets. The subsequent increase in size and number of these droplets was dependent on IGF-IR activation. Two previous studies (16, 24) effectively isolated the effect of IGF-I from other components of the differentiation mixture. Miki et al. (24) used mouse embryonic fibroblasts from IRS-1 and IRS-2 knockout mice to block the downstream effect of receptor activation, and Qiu et al. (16) withheld IGF-I from the induction cocktail for 3T3-L1 cells. Each of these studies reported that IGF-I played a role in lipid accumulation that is consistent with our findings. The IGF-I-dependent changes in PPAR that we demonstrate here are similar to those reported in the embryonic fibroblast system by Miki et al. (24).
In summary, the data presented in this paper expand our understanding of the process of differentiation in the HMSC model system by showing that this system parallels the differentiation process observed in other models in several ways. These findings also demonstrate the IGF-IR mediates the differentiation induced by both insulin and IGF-I in this system. Finally, these findings also delineate which aspects of differentiation are dependent on insulin or IGF-I from those that are dependent on the combination of dexamethasone, IBMX, and indomethacin (D mixture). It remains to be seen whether the tools used to manipulate aspects of IGF-IR signaling can be applied to this system: will the cells tolerate adequate dose and time exposure to inhibitors; will they transiently or stably transfect expression vectors well. If so, they should make a stable, renewable, and potentially clinically useful cellular model in which to work through the mechanistic complexities of IGF-IR signaling during adipocyte differentiation.
|
Footnotes |
|---|
Received September 25, 2003.
Accepted April 6, 2004.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
and RXR agonists. Diabetes 51:718–723 in regulating a cascade expression of cyclin-dependent kinase inhibitors, p18(INK4c) and p21(Waf1/Cip1), during adipogenesis. J Biol Chem 274:17088–17097This article has been cited by other articles:
 |
|
 |
|
  K. K Ong, M. Langkamp, M. B Ranke, K. Whitehead, I. A Hughes, C. L Acerini, and D. B Dunger Insulin-like growth factor I concentrations in infancy predict differential gains in body length and adiposity: the Cambridge Baby Growth Study Am. J. Clinical Nutrition, July 1, 2009; 90(1): 156 - 161. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. Kloting, L. Koch, T. Wunderlich, M. Kern, K. Ruschke, W. Krone, J. C. Bruning, and M. Bluher Autocrine IGF-1 Action in Adipocytes Controls Systemic IGF-1 Concentrations and Growth Diabetes, August 1, 2008; 57(8): 2074 - 2082. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. U. Freda, W. Shen, S. B. Heymsfield, C. M. Reyes-Vidal, E. B. Geer, J. N. Bruce, and D. Gallagher Lower Visceral and Subcutaneous but Higher Intermuscular Adipose Tissue Depots in Patients with Growth Hormone and Insulin-Like Growth Factor I Excess Due to Acromegaly J. Clin. Endocrinol. Metab., June 1, 2008; 93(6): 2334 - 2343. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. Matsushita, Y. Wu, Y. Okamoto, R. E. Pratt, and V. J. Dzau Local Renin Angiotensin Expression Regulates Human Mesenchymal Stem Cell Differentiation to Adipocytes Hypertension, December 1, 2006; 48(6): 1095 - 1102. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Hashimoto, Y. Kariya, and K. Miyazaki Regulation of Proliferation and Chondrogenic Differentiation of Human Mesenchymal Stem Cells by Laminin-5 (Laminin-332) Stem Cells, November 1, 2006; 24(11): 2346 - 2354. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
