Visceral fat-specific regulation of plasminogen activator inhibitor-1 in aged septic mice

Abstract

Elevated plasma levels of plasminogen activator inhibitor-1 (PAI-1) are documented in patients with sepsis and levels positively correlate with disease severity and mortality. Our previous work demonstrated that visceral adipose tissues (VAT) are a major source of PAI-1, especially in the aged (murine endotoxemia), that circulating PAI-1 protein levels match the trajectory of PAI-1 transcript levels in VAT (clinical sepsis), and that PAI-1 in both VAT and plasma are positively associated with acute kidney injury (AKI) in septic patients. In the current study utilizing preclinical sepsis models, PAI-1 tissue distribution was examined and cellular sources, as well as mechanisms mediating PAI-1 induction in VAT, were identified. In aged mice with sepsis, PAI-1 gene expression was significantly higher in VAT than in other major organs. VAT PAI-1 gene expression correlated with PAI-1 protein levels in both VAT and plasma. Moreover, VAT and plasma levels of PAI-1 were positively associated with AKI markers, modeling our previous clinical data. Using explant cultures of VAT, we determined that PAI-1 is secreted robustly in response to recombinant transforming growth factor β (TGFβ) and tumor necrosis factor α (TNFα) treatment; however, neutralization was effective only for TNFα indicating that TGFβ is not an endogenous modulator of PAI-1. Within VAT, TNFα was localized to neutrophils and macrophages. PAI-1 protein levels were fourfold higher in stromal vascular fraction (SVF) cells compared with mature adipocytes, and among SVF cells, both immune and nonimmune compartments expressed PAI-1 in a similar fashion. PAI-1 was localized predominantly to macrophages within the immune compartment and preadipocytes and endothelial cells within the nonimmune compartment. Collectively, these results indicate that induction and secretion of PAI-1 from VAT is facilitated by a complex interaction among immune and nonimmune cells. As circulating PAI-1 contributes to AKI in sepsis, understanding PAI-1 regulation in VAT could yield novel strategies for reducing systemic consequences of PAI-1 overproduction.

Keywords: PAI-1; adipose tissue; aging; kidney injury; sepsis.

Figure 1.. Adipose tissues are a major site for PAI-1 synthesis during sepsis.

Polymicrobial sepsis was induced by CLP in aged mice (24 months old) which were euthanized 24h later for tissue collection. RNA and proteins were extracted from major organs and processed for qRT-PCR and Western blot analyses. (a) PAI-1 mRNA levels in various tissues of CLP mice, data are expressed relative to 18SrRNA transcript levels (tissues without a common letter (i.e. a, b, c) significantly differ, analyzed by paired t-tests, p<0.05); (b) PAI-1 mRNA levels in eFat of sham and CLP mice, data are expressed relative to 18SrRNA mRNA levels; (c) PAI-1 protein levels in in eFat of sham and CLP mice, intensity of each band was normalized to total protein content of each lane. *p<0.05 assessed by Brown-Mood median test (d) correlation between mRNA and protein levels in VAT assessed by Pearson correlation coefficient. Data are shown as individual values in box plots (n=5 each group) or correlation plots. eFat: epididymal fat; mFat: mesenteric fat; rFat: retroperitoneal fat; sFat: subcutaneous fat.

Figure 2.. Sepsis-induced visceral adipose tissue synthesis of PAI-1 is reflected at the protein level in both visceral adipose tissue and plasma and levels correlate with markers of acute kidney injury.

Sepsis was induced in middle-aged mice (12 months old) by cecal slurry (CS) injection and mice were euthanized 24, 48, and 72h later for tissue and plasma collection. RNA and protein were extracted from visceral adipose tissues (epididymal depot) and processed for qRT-PCR or Western blot, respectively. (a) PAI-1 mRNA levels in VAT assessed by qRT-PCR, data are expressed relative to 18S mRNA levels; (b) Western blot and quantification for PAI-1 protein in VAT, intensity of each band was normalized to total protein content of each lane; (c) correlation of PAI-1 mRNA and PAI-1 protein levels in VAT for data points through 48h; (d) Western blot and quantification for PAI-1 protein in plasma, intensity of each band was normalized to total protein content of each lane; (e) correlation of PAI-1 mRNA in VAT and plasma PAI-1 VAT for data points through 48h; (f) plasma NGAL concentration measured by ELISA; (g) correlation of PAI-1 mRNA in VAT and plasma NGAL levels for data points through 48h; (h) correlation of PAI-1 protein in plasma and plasma NGAL levels for data points through 48h. All data are shown as individual values in box plots (n=4 each group) or correlation plots. Statistical differences were determined by Brown-Mood median test: overall p-values are depicted in upper left or right corner of plot, **p<0.01 indicate multiple comparison tests vs ctrl. Correlations were assessed by Pearson correlation coefficient. Ctrl: control; NGAL: neutrophil gelatinase-associated lipocalin; VAT: visceral adipose tissue.

Figure 3.. TNFα is a major endogenous inducer of PAI-1 in visceral adipose tissue.

Explant cultures of VAT from naïve aged mice (21–24 months old) were prepared in multiple separate experiments (tissues for each experiment were derived from 1–2 mice) and secretion of PAI-1 protein to the medium was assessed by ELISA. (a) Cultures were stimulated with LPS (10μg/mL), heat-killed cecal slurry (HK-CS, 2.4 × 10⁴ cells/mL); PMA and Ionomycin (81nM PMA and 1.3μM ionomycin), and recombinant proteins IL-1β (5ng/mL), IL-17A (20ng/mL), TNFα (10ng/mL), TGFβ (5ng/mL), and angiotensin II (AngII, 100nM) and culture medium was sampled at 6 and 24h. Experiments were replicated at least twice for each condition with n=6–9 technical replicates for each treatment. (b) Cultures were stimulated with HK- CS, treated with a neutralizing antibody for TNFα (1.3 μg/mL), and culture medium sampled at 6 and 24h; experiment was repeated four times (3–4 technical replicates in each experiment) with similar trends. (c) Cultures were stimulated with heat-killed CS, treated with a neutralizing antibody for TGFβ (3 μg/mL), and culture medium sampled at 6 and 24h. Experiment was repeated twice (3 technical replicates for each experiment) with similar results. PAI-1 levels were normalized for adipose tissue DNA content of each well. Data are expressed as the mean ± SEM. Statistical differences were determined Brown-Mood median test or one-way ANOVA as described in Methods. (*p<0.05, **p<0.01, ***p<0.001). (d) Gating scheme for identification of the cellular source of TNFα in HK-CS stimulated adipose SVF cells. (e) Percent of cells positive for TNFα among different cell subsets. (f) Mean fluorescence intensity (MFI) in TNFα expressing cells.

Figure 4.. Sepsis-induced PAI-1 protein in visceral adipose tissue is localized to both immune and non-immune cells of the stromal vascular fraction (SVF).

VAT from naïve control and CS-injected aged mice (21–24 months old) were digested with collagenase and adipocytes (Ad) and stromal vascular fraction (SVF) cells purified by repeated centrifugation. (a) Proteins were extracted from each fraction (n=1 non-injected control mouse, n=3 CS-injected mice) and PAI-1 quantified by Western blot analyses for CS-injected mice only, each band represents proteins from an individual mouse, intensity of each band was normalized to total protein content of each lane; (b) SVF cells obtained from pooled VAT were magnetically separated into CD45 positive and CD45 negative fractions; proteins were extracted from each fraction (n=1 non-injected control sample pooled from 8 mice, n=3 pools of VAT from CS-injected mice (n=6–8 mice for each pool)) and PAI-1 quantified by Western blot analyses for CS-injected mice only), intensity of each band was normalized to total protein content of each lane. Statistical differences were determined by one-sided paired t-tests as only increases were of interest, *p<0.05. (c) SVF cells were magnetically separated into macrophages (M, F4/80 positive), T cells (T, CD3 positive), and remaining CD45 positive cells (O, other immune cells depleted of CD3 and F4/80 positive cells); proteins were extracted from each fraction, and PAI-1 quantified by Western blot analyses. Data are representative of two experiments, each with pooled cells derived from 8 CS-injected mice. (d) SVF cells were depleted of CD45 positive cells, then by a series of positive selections, magnetically separated into endothelial cells (E, CD31), preadipocytes (P, CD34), and remaining CD45 negative cells; proteins were extracted from each fraction, and PAI-1 quantified by Western blot analyses. Data are representative of two experiments, each with pooled cells derived from 13 CS-injected mice.

FIGURE 5.. Mechanism and downstream effects of PAI-1 induction in VAT of septic mice.

Our collective data suggest that, during sepsis, TNFα in VAT is derived predominantly from macrophages and neutrophils; by paracrine signaling TNFα derived from these cells stimulates PAI-1 production and secretion from macrophages, preadipocytes, endothelial cells, and mature adipocytes. Adipose-derived PAI-1 is secreted into the circulation from the visceral adipose depots where it acts by inhibiting fibrinolysis. Consequently, due to the established effects of PAI-1 on promoting the continuation of thrombosis, intravascular coagulation and reduced blood flow result, contributing to organ injury, in particular acute kidney injury.