Stromal Cells Are the Main Plasminogen Activator Inhibitor-1-Producing Cells in Human Fat
Evidence of Differences Between Visceral and Subcutaneous Deposits
Elevated plasma plasminogen activator inhibitor (PAI)-1 observed during insulin resistance has been connected with an excessive PAI-1 adipose tissue secretion mainly by visceral fat. Our aim was to compare the localization of PAI-1 in human visceral and subcutaneous fats. PAI-1 secretion was also investigated in vitro during human adipocyte differentiation. PAI-1 antigen and mRNA were localized in the stromal area of the tissue and were also present in a few CD14-positive monocytes, in direct contact with adipocytes. In addition, in subcutaneous tissue, PAI-1 mRNA contents, determined by using real-time polymerase chain reaction, were higher in the stromal fraction than in the adipocyte fraction. PAI-1 mRNA-positive cells were 5-fold more frequent in the visceral area than in the subcutaneous stromal area (P=0.004). Such a difference was also observed for PAI-1 mRNA content between both whole adipose tissues. In contrast to leptin, during adipocyte differentiation, PAI-1 secretion did not follow adipocyte maturation. In situ hybridization in culture did not reveal PAI-1 mRNA in lipid-filled cells. Our results demonstrate that PAI-1 production is mainly due to stromal cells, which were more numerous in the visceral than in the subcutaneous depot. These results could explain the strong relationship observed between circulating PAI-1 levels and the accumulation of visceral fat.
Plasminogen activator inhibitor (PAI)-1 is a fibrinolytic inhibitor whose elevated plasma concentration is thought to contribute to the increased susceptibility to atherogenesis described in insulin-resistant patients with obesity.1,2
Interestingly, some reports have underlined the independent association of body adiposity and plasma PAI-1 levels, mainly visceral fat, in mice3,4 and in humans.5–9 PAI-1 expression has been reported to be upregulated in adipose tissue from obese mice and humans concomitantly with increased plasma PAI-1 levels,10–12 and it has been proposed that the increase in adipose tissue PAI-1 observed in obesity could be the result of tumor necrosis factor-α disturbance, which specifically accompanies insulin resistance.13
The precise origin of PAI-1 expression in adipose tissue is not definitively established, because PAI-1 synthesis has been attributed or not to adipocytes, depending on which model was used. Several groups have documented PAI-1 synthesis by murine adipocyte cell lines.10,14–16 Recently, Crandall et al17 observed that human stromal cells issued from adipose tissue produced PAI-1 after 24 hours of incubation in medium containing serum, whereas Birgel et al18 identified PAI-1 antigen in stromal preadipocytes differentiated into adipocytes. In untreated CB6 mice, Samad et al10 localized PAI-1 mRNA by in situ hybridization in the smooth muscle cells within the small vessels observed in adipose tissue, whereas no specific signal was observed in the fat cells. In contrast, after endotoxin or tumor necrosis factor-α injection, both cell types produced PAI-1 mRNA.10 In obese mice, a strong PAI-1 mRNA signal was localized in the smooth muscle cells of the tunica media of the vessels, but a positive signal was also observed in >50% of the cells that morphologically resembled adipocytes.19 In humans, we have previously shown, by use of RNase protection assay, that PAI-1 mRNA was mainly present in the stromal fraction of freshly collected adipose tissue from obese patients, whereas, conversely, immunolocalization performed with isolated adipocytes also evidenced the protein at the adipocyte level.20 Until now, the precise localization of PAI-1 in human adipose tissue sections has not been evaluated. Several studies have emphasized the relationship between visceral fat and plasma PAI-1.5–9 In mice, Shimomura4 described a strong PAI-1 expression in visceral fat. Using human adipose tissue explants maintained in culture or differentiating adipocytes, we and others have observed a higher PAI-1 secretion by the visceral fat.11,20,21
The purpose of the present work was to identify the localization of PAI-1 synthesis from cryosections of freshly collected human adipose tissue from obese patients or from human adipose cells in primary culture by immunolocalization and in situ hybridization and to compare the levels of PAI-1 expression between visceral and subcutaneous fat territories.
Population and Sample Collection
Visceral and subcutaneous adipose tissues were obtained during gastroplasty from 26 patients (22 women, 4 men) aged 41±12 (mean±SD) years, with mean±SD body mass index (BMI) 41±6.3 kg/m2, which is weight (kilograms) divided by height squared (meters squared). Mean±SD (range) values for plasma PAI-1 antigen and activity levels were 52±36 (12 to 165) ng/mL and 26±29 (1 to 140) IU/mL, respectively. Fasting insulinemia was 7.3±6.7 (mean±SD) μU/mL. The fasting triglyceride level was 1.56±0.65 (mean±SD) g/L. All subjects were white and did not suffer from any ongoing disease (infection, cancer). Investigations were conducted according to the principles expressed in the declaration of Helsinki.
After resection, adipose tissue was immediately put into dry ice and rapidly transported to the laboratory. The separation of mature adipocytes from stromal cells was performed with subcutaneous adipose tissue samples obtained from 8 women during abdominal lipectomy; mean±SD (range) values for BMI and age were 23±3.7 (20 to 31) kg/m2 and 33±10 (18 to 54) years, respectively.
Venous blood samples were obtained just before anesthesia. For PAI-1 antigen and activity, samples were drawn into chilled trisodium citrate tubes, centrifuged as previously described11 to obtain platelet free plasma, and stored at −80°C until use.
PAI-1 Antigen Determination
Supernatant PAI-1 antigen was assayed as previously described.22 Plasma PAI-1 antigen and activity were assayed by using Asserachrom PAI-1 (Diagnostica Stago) and Chromolize (Biopool) kits, respectively.
Total RNA was extracted by using the method of Chomczinski and Sacchi.23 The integrity of the RNA was confirmed by electrophoresis in ethidium bromide-containing agarose gels, and the RNA concentration was determined spectrophotometrically.
Quantitative RT-PCR TaqMan
Reverse transcription (RT) was performed as previously described.11 Amplification was performed in a final volume of 25 μL with an Abiprism 7700 (Perkin Elmer). Amplification was performed in 25 μL on an AbiPrism 7700 (Perkin Elmer). The polymerase chain reaction (PCR) conditions were as follows: 2 minutes at 50°C, 10 minutes at 95°C followed by 40 cycles of 2-step PCR reaction denaturation at 95°C for 15 seconds, and annealing extension at 60°C for 60 seconds. Each sample contained 2 μL cDNA in 1× TaqMan Universal PCR Master Mix (Applied Biosystems) and 400 mmol/L of primers (Life Technologies) and probe (FAM labeled, Applied Biosystems). Primer express software (Applied Biosystems) was used to design primers and probes for PAI-1 mRNA and 18S rRNA (housekeeping gene). For PAI-1 mRNA quantification, primers were 5′-CAG AAA GTG AAG ATC GAG GTG AAC-3′ and 5′-GGA AGG GTC TGT CCA TGA TGA T-3′, and the probe was 5′-ACG GTG GCC TCC TCA TCC ACA GC-3′ (GenBank accession No. M16006). The primers for 18S rRNA were 5′-CTA CCA CAT CCA AGG AAG GCA-3′ and 5′-TTT TTC GTC ACT ACC TCC CC-3′, and the probe for 18S rRNA was 5′-CGC GCA AAT TAC CCA CTC CCG AC-3′ (GenBank accession No. X03205).
The number of cycles required to generate a threshold of 0.03 fluorescence units was determined in triplicate for each sample. Results were accepted if the coefficient of variation was <0.5 for the number of cycles required. Each experiment included a standard curve for the individual amplicon. Results were expressed as femtograms of PAI-1 mRNA per nanograms of 18S rRNA.
Separation of Mature Adipocytes From Stromal Cells and Cell Culture
Stromal and adipocyte cell fractions were isolated by collagenase digestion as previously described.20 Semiquantitative RT-PCRs for von Willebrand factor (vWF) and leptin mRNA were performed on each fraction to determine their degree of purity. For cell cultures, the stromal cell fraction was incubated with erythrocyte-lysing buffer that contained 154 mmol/L NH4Cl, 10 mmol/L KHCO3, and 0.1 mmol/L EDTA for 10 minutes. Cells were repeatedly washed and resuspended in DME/F-12 Ham medium (vol/vol) supplemented with 15 mmol/L NaH2CO3, 33 μmol/L biotin, and 17 μmol/L pantothenate. After cell attachment (SonicSeal Slides, Nunc), the medium was switched to a differentiation mixture that consisted of the same medium enriched with 252 μmol/L isobutylmethylxanthine, insulin transferrin selenium (ITS, Sigma Chemical Co; 1:100), 10 nmol/L dexamethasone, and 1 nmol/L triiodothyronine. After 4 days of incubation, isobutylmethylxanthine was retrieved. Medium was then changed every 72 hours. Differentiated adipocytes were defined as cells, the cytoplasm of which was completely filled with lipid droplets. Leptin was quantified in the culture medium (Human Leptin Quantikine, R&D Systems) as a marker of adipocyte maturation. After 21 days of culture, cells were fixed in 4% paraformaldehyde.
Sections (20 to 25 μm) were cut in a cryostat microtome (Leica) at −20°C and put onto slides (Superfrost Plus, CML). Dried slides were fixed for 5 minutes in acetone and then washed for 5 minutes in PBS. Labeling was performed with the DAKO LSAB kit according to the manufacturer’s instructions. A mix of 4 monoclonal antibodies kindly provided by Prof P. Declerck (Faculty of Pharmaceutical Sciences, Leuven, Belgium) was used at a final concentration of 5 μg/mL to detect PAI-1. Two monoclonal antibodies directed against CD68 and CD14 (Dako) allowed the identification of cells from the monocyte/macrophage lineage. They were diluted at 1:100 and 1:20, respectively. Two antibodies directed against vWF (rabbit polyclonal antibody from Dako, dilution 1:300) and CD34 (Dako) were used to evidence endothelial cells. Anti-α-actin (rabbit antibody, diluted 1:50, Sigma) and anti-cytokeratin M18 (mouse monoclonal antibody, diluted 1:100, Sigma) antibodies were used to identify smooth muscle cells and mesothelial cells, respectively. Antibodies against leptin were kindly provided by Prof R. Negrel (Center of Biochemistry, Université de Nice Sophia-Antipolis, Nice, France). Slides were then counterstained with Harris’ hematoxylin.
Double labeling was performed by combining a first incubation step with anti-CD14 antibodies (1:20). Then, slides were subjected to goat anti-mouse antibodies linked to alkaline phosphatase. Staining was performed by using the fast blue substrate (Vectastain) from Vector in the presence of levamisole during 20 minutes at obscurity. Slides were then rinsed with PBS and incubated for 15 minutes in normal mouse serum (1:10). Slides were then incubated for 30 minutes with the mix of 4 anti-PAI-1 monoclonal antibodies coupled to peroxidase. After a further wash, peroxidase was revealed by using amino-9-ethylcarbazole (AEC) in the presence of H2O2 for 20 minutes. Controls without antibodies were assessed under the same conditions.
In Situ Hybridization
PAI-1 sense and antisense riboprobes were produced as previously described.24 Leptin riboprobes were used to obtain a labeling of adipocytes. The leptin probe was a 301-bp fragment (from base 57 to 357) subcloned in pPCRscript (Stratagene) linearized with BamHI (antisense) and with NotI (sense). Tissues were processed and hybridized with 35S-labeled PAI-1 or leptin riboprobes as previously described.25 Slides were subsequently dipped into nuclear emulsion (Ilford K5, Ilford) diluted 1:2 and exposed for 30 days. After development, slides were counterstained with neutral nuclear red. PAI-1 mRNA-labeled cell bodies were manually counted, whereas the total stromal surface was automatically calculated by using NIH image software.26 Four visceral and subcutaneous sections (mean surface 1 [range 0.3 to 1.7] cm2) were analyzed per subject.
Results were expressed as mean±SD (range). The value n represents the number of tissue samples. In the study aimed to compare visceral and subcutaneous expression, the between-group comparison was tested by using the nonparametric paired Wilcoxon test. Significance was defined as a value of P≤0.05.
Immunolocalization of PAI-1 Antigen in Human Adipose Tissue
Sixteen paired adipose samples of visceral and subcutaneous origin were immunostained for PAI-1 antigen. The population was composed of 13 women and 3 men aged 38±12.5 years, with a wide range of BMI (41±6.7 [range 27 to 50] kg/m2). PAI-1 antigen was localized mainly in the stromal compartment of the tissue whatever the territory tested (visceral or subcutaneous) and whatever the PAI-1 activity plasma level (which was 30±35 [range 2 to 140] UI/mL). PAI-1 antigen was detected in purely stromal area (Figure 1A) as was α-actin (Figure 1B), and PAI-1 antigen was also present in small cells in direct contact with adipocytes (Figure 1C) as were monocytes (or macrophages) immunostained with antibodies against CD14 (Figure 1D) or CD68 (data not shown). The background staining was similar to that obtained after having complexed the PAI-1 antibodies with purified PAI-1 (data not shown). This pattern of PAI-1 localization differs from that observed for vWF antigen (Figure 1E) and for cytokeratin M18, which was localized at the periphery of the adipose lobules in the visceral tissue only (data not shown). Leptin was localized in the small rim of the adipocyte cytoplasm as well as in the stromal area (Figure 1F). Control experiments performed without primary antibodies did not exhibit any specific labeling (data not shown).
Double labeling performed with PAI-1 and CD14 showed that few small cells in close contact with adipocytes expressed both PAI-1 and CD14 antigen. However, the majority of CD14-positive cells were negative for PAI-1 (data not shown). No colocalization was evidenced between PAI-1 and vWF or CD 34 (data not shown).
In Situ Hybridization and Quantification of PAI-1 mRNA in Visceral and Subcutaneous Human Adipose Tissue
Paired adipose samples of visceral and subcutaneous origin were subjected to in situ hybridization. The population was composed of 12 women and 3 men (aged 42±13 [range 23 to 66] years, BMI 42±6.8 [range 33 to 57] kg/m2) with a wide range of plasma PAI-1 levels (47.8±36 [range 12 to 165] ng/mL). The pattern of PAI-1 expression was compared with that of leptin, which is produced by adipocytes. PAI-1 expression differed from that of leptin because it was mainly expressed in the stromal area from visceral tissue (Figure 2A) as well as from subcutaneous tissue (Figure 2B) and in small cells in close contact with adipocytes but not within adipocytes (Figure 2C), whereas leptin was observed mainly in the small cytoplasmic rim of adipocytes (Figure 2D). Slides hybridized with PAI-1 mRNA sense probe were negative (data not shown).
The counterstaining performed allowed quantification of the number of PAI-1 mRNA-positive cells per square millimeter of the stromal surface from both territories. The percentage of the quantified surface did not differ between the 2 fat territories. Interestingly, the stroma of the visceral tissue contained 5-fold more positive cells than the stroma of the subcutaneous tissue (67.6±61.2 [range 9.5 to 166.8] versus 13.6±24.2 [range 0 to 66.2] cell/mm2, respectively; P=0.004; n=15). This difference was also observed when the amounts of PAI-1 mRNA were measured by quantitative RT-PCR in both tissues. Indeed, the visceral tissue contained 4.7-fold more PAI-1 mRNA than did the subcutaneous tissue (141±155 [range 7.6 to 637 ] versus 29.8±29.4 [range 0.7 to 102 ] fg PAI-1 mRNA/ng 18S rRNA, respectively; P=0.001; n=21).
Immunolocalization and In Situ Hybridization of PAI-1 mRNA in Cultured Human Adipocytes
Adipocyte differentiation was followed for 21 days. A 4-fold increase in PAI-1 antigen accumulation was observed between days 4 and 10. Then, after a plateau of 6 days, PAI-1 antigen accumulation progressively decreased. This pattern of production is in contrast to that of leptin, which continued to increase (Figure 3). The decrease of PAI-1 antigen accumulation was not due to modification of total or lipid-filled cell number, inasmuch as they did not decrease throughout the culture period (data not shown).
Cell types in culture proved to be heterogeneous. We could distinguish some round cells labeled with antibodies against CD68 (Figure 4C) and identified as macrophages. Leptin was found in cells full of lipids (Figure 4D) as well as in more largely spread cells free of lipids (data not shown). PAI-1 antigen was detected in several types of cells: fully differentiated adipocytes, cells free of lipids (Figure 4A), and cells that morphologically resembled those positive for CD68 (Figure 4B). The intensity of the PAI-1 signal obtained at day 14 was higher than that observed at day 21 (data not shown).
In situ hybridization for PAI-1 was performed on cultures at different periods of time. PAI-1 mRNA labeling was mainly observed in small cells but not in lipid-filled cells. As for PAI-1 antigen, we observed a decrease in PAI-1 mRNA signal intensity preceding that of PAI-1 antigen (Figure 2E through 2F)
PAI-1 Expression in Mature Adipocytes and Stromal Cells From Human Subcutaneous Adipose Tissue
Subcutaneous adipose tissues (n=3) were subjected to collagenase digestion, as described in Methods, to separate the adipocytes from the stromal cells. A complete depletion of the stromal cells from the adipocyte fraction was not obtained, inasmuch as vWF mRNA was detected in both fractions, whereas leptin mRNA was only detected in the adipocyte fraction (data not shown). Total RNA was then obtained to quantify PAI-1 mRNA by quantitative RT-PCR. The stromal fraction contained much more PAI-1 mRNA than did the adipocyte fraction. Stromal cells versus adipocyte fractions were, respectively, as follows: 6.05 versus 1.43, 116.82 versus 11.57, and 16.47 versus 0.35 fg PAI-1 mRNA/ng 18S rRNA.
The present results show that in human adipose tissue, stromal cells are, in vivo, the main source of PAI-1, even in obese patients with high plasma PAI-1 levels. In contrast to Samad and Loskutoff,19 we did not find PAI-1 mRNA in mature adipocytes. Indeed, PAI-1 mRNA was never detected in the cytoplasmic part of adipocytes but rather in small cells in close contact with the latter. Moreover, by use of quantitative RT-PCR assay on isolated cell fractions from adipose tissue, PAI-1 mRNA levels were observed to be higher in the stromal than in the adipocyte fractions. It could be hypothesized that the remaining PAI-1 found in the adipocyte fraction was due to contaminating stromal cells. Indeed, we never obtained a complete depletion of the stromal cells from the adipocyte fraction, because vWF mRNA was detected in both fractions (data not shown), but a production by adipocytes lower than the limit of sensitivity of the methods we have used cannot be excluded. Thus, PAI-1 antigen and mRNA were mainly found in the stroma of human adipose tissue. This result is original, inasmuch as most of the groups interested in studying PAI-1 synthesis by adipose tissue have focused on adipocyte rather than stromal cells.21,27,28
Visceral fat was observed to express almost 5-fold more PAI-1 mRNA than did abdominal subcutaneous adipose tissue by use of real-time PCR TaqMan. This method is very accurate compared with the semiquantitative method that we used previously, which did not allow demonstration of such a difference.11 Furthermore, in situ hybridization showed that the number of PAI-1-expressing cells was higher in visceral than in subcutaneous fat. These results confirm previous reports showing a higher PAI-1 antigen secretion by visceral adipose tissue explants compared with subcutaneous explants.20,21,28 All these data are in favor of a high PAI-1 synthesis capacity of the visceral fat. These data are in contrast to the results published by Eriksson et al12 who, after normalizing the results by the number of fat cells, found a higher level of PAI-1 expression in subcutaneous than in visceral tissue. In the present study, the PAI-1 expression levels were normalized by the number of fat cells. The absence or low contribution of adipocytes in PAI-1 production could probably explain these results. Our results contrasted with the lack of difference between the adipose tissue PAI-1 antigen content of subcutaneous and visceral fats that we reported previously.11 This could be attributed to a retention of PAI-1 antigen in extracellular matrix, which could impair the demonstration of a difference because the quantification of surfaces labeled for PAI-1 antigen did not differ between territories (data not shown).
A complete identification of the stromal cells involved in PAI-1 synthesis could not be undertaken. However, the colocalization of CD14 and PAI-1 revealed that some of these cells could be of monocyte origin. The similar pattern of PAI-1 antigen and α-actin is also in favor of a contribution of smooth muscle cells, as proposed by Samad et al.10 Nevertheless, a contribution of other stromal cells as preadipocytes cannot be excluded. To analyze this point, we have focused on the localization of PAI-1 synthesis during human adipocyte differentiation. In situ hybridization experiments localized PAI-1 mRNA in small cells free of lipids. This result was in contrast to the PAI-1 immunostaining, which revealed the presence of PAI-1 antigen in few lipid-filled cells. This phenomenon is not fully understood but could be due to a diffusion of the PAI-1 protein to the adhesion surface of the adipocyte unless, as underlined by Cousin et al,29 it reflects a phagocytic activity of these cells. Several groups,14,16 together with ours,20 have demonstrated that murine cell lines subjected to differentiation produce PAI-1. Few data have been published regarding human adipocyte cell culture. Crandall et al17 observed synthesis and secretion of PAI-1 after a 24-hour incubation of isolated stromal cells in medium containing 10% serum. This was in agreement with a PAI-1 synthesis by preadipocytes or nonadipocyte cells. More recently, Birgel et al18 demonstrated, like ourselves, that PAI-1 protein secretion increased 1.9-fold during the course of adipocyte differentiation, but no localization of PAI-1 mRNA was performed. The present results demonstrate that in human adipose tissue, stromal cells appear to be the main cells involved in PAI-1 production. Although mature adipocytes have not been found to produce PAI-1, a contribution of preadipocytes cannot be excluded. PAI-1 expression is almost 5-fold higher in the visceral than in the subcutaneous fat because of the increase in stromal PAI-1producing cells. These results could explain the strong relationship observed between circulating PAI-1 levels and the accumulation of visceral fat and suggest that visceral fat is subjected to a particular environment that stimulates PAI-1 synthesis.
This work was supported by grants from INSERM and from Ministère de l’Education Nationale, de l’Enseignement Supérieur, et de l’Insertion Professionnelle (Contrat Quadriennal). D. Bastelica is a recipient of the French Atherosclerosis Society (SFA). We thank O. Georgelin-Geel and M. Verdier for excellent technical assistance.
Received May 18, 2001; revision accepted September 13, 2001.
Juhan-Vague I, Pike S, Alessi MC, Jespersen J, Haverkate F, Thompson SG. Fibrinolytic factor and the risk of myocardial infarction of sudden death in patients with angina pectoris. Circulation. 1996; 94: 2057–2063.
Mavri A, Stegnar M, Krebs M, Sentocnik JT, Geiger M, Binder BR. Impact of adipose tissue on plasma plasminogen activator inhibitor-1 in dieting obese women. Arterioscler Thromb Vasc Biol. 1999; 19: 1582–1587.
Giltay EJ, Elbers JM, Gooren LJ, Emeis JJ, Kooistra T, Asscheman H, Stehouwer CD. Visceral fat accumulation is an important determinant of PAI-1 levels in young, nonobese men and women: modulation by cross-sex hormone administration. Arterioscler Thromb Vasc Biol. 1998; 18: 1716–1722.
Mykkanen L, Ronnemaa T, Marniemi J, Haffner SM, Bergman R, Laakso M. Insulin sensitivity is not an independent determinant of plasma plasminogen activator inhibitor-1 activity. Arterioscler Thromb. 1994; 14: 1264–1271.
Alessi MC, Bastelica D, Morange P, Berthet B, Leduc I, Verdier M, Geel O, Juhan-Vague I. Plasminogen activator inhibitor 1, transforming growth factor-beta1, and BMI are closely associated in human adipose tissue during morbid obesity. Diabetes. 2000; 49: 1374–1380.
Samad F, Uysal KT, Wiesbrock SM, Pandey M, Hotamisligil GS, Loskutoff DJ. Tumor necrosis factor alpha is a key component in the obesity-linked elevation of plasminogen activator inhibitor 1. Proc Natl Acad Sci U S A. 1999; 96: 6902–6907.
Lundgren CH, Brown SL, Nordt TK, Sobel BE, Fujii S. Elaboration of type-1 plasminogen activator inhibitor from adipocytes: a potential pathogenetic link between obesity and cardiovascular disease. Circulation. 1996; 93: 106–110.
Morange PE, Aubert J, Peiretti F, Lijnen HR, Vague P, Verdier M, Negrel R, Juhan-Vague I, Alessi MC. Glucocorticoids and insulin promote plasminogen activator inhibitor 1 production by human adipose tissue. Diabetes. 1999; 48: 890–895.
Birgel M, Gottschling-Zeller H, Rohrig K, Hauner H. Role of cytokines in the regulation of plasminogen activator inhibitor-1expression and secretion in newly differentiated subcutaneous human adipocytes. Arterioscler Thromb Vasc Biol. 2000; 20: 1682–1687.
Alessi MC, Peiretti F, Morange P, Henry M, Nalbone G, Juhan-Vague I. Production of plasminogen activator inhibitor 1 by human adipose tissue: possible link between visceral fat accumulation and vascular disease. Diabetes. 1997; 46: 860–867.
Declerck PJ, Alessi MC, Verstreken M, Kruithof EKO, Juhan-Vague I, Collen D. Measurement of plasminogen activator inhibitor 1 (PAI-1) in biological fluids with a murine monoclonal antibody, based on enzyme-linked immunosorbent assay. Blood. 1988; 71: 220–225.
Grino M, Zamora AJ. An in situ hybridization histochemistry technique allowing simultaneous visualization by the use of confocal microscopy of three cellular mRNA species in individual neurons. J Histochem Cytochem. 1998; 46: 753–759.
Cousin B, Munoz O, Andre M, Fontanilles AM, Dani C, Cousin JL, Laharrague P, Casteilla L, Penicaud L. A role for preadipocytes as macrophage-like cells. FASEB J. 1999; 13: 305–312.