Endothelial dysfunction contributes to the development of acute kidney injury (AKI) in animal models of ischemia reperfusion injury and sepsis. There are limited data on markers of endothelial dysfunction in human AKI. We hypothesized that Protein C (PC) and soluble thrombomodulin (sTM) levels could predict AKI. We conducted a multicenter prospective study in 80 patients to assess the relationship of PC and sTM levels to AKI, defined by the AKIN creatinine (AKI Scr) and urine output criteria (AKI UO). We measured marker levels for up to 10 days from intensive care unit admission. We used area under the curve (AUC) and time-dependent multivariable Cox proportional hazard model to predict AKI and logistic regression to predict mortality/non-renal recovery. Protein C and sTM were not different in patients with AKI UO only versus no AKI. On intensive care unit admission, as PC levels are usually lower with AKI Scr, the AUC to predict the absence of AKI was 0.63 (95%CI 0.44-0.78). The AUC using log10 sTM levels to predict AKI was 0.77 (95%CI 0.62-0.89), which predicted AKI Scr better than serum and urine neutrophil gelatinase-associated lipocalin (NGAL) and cystatin C, urine kidney injury molecule-1 and liver-fatty acid-binding protein. In multivariable models, PC and urine NGAL levels independently predicted AKI (p=0.04 and 0.02) and PC levels independently predicted mortality/non-renal recovery (p=0.04). In our study, PC and sTM levels can predict AKI Scr but are not modified during AKI UO alone. PC levels could independently predict mortality/non-renal recovery. Additional larger studies are needed to define the relationship between markers of endothelial dysfunction and AKI.

Background

Diabetic cardiomyopathy (DM CMP) is defined as cardiomyocyte damage and ventricular dysfunction directly associated with diabetes independent of concomitant coronary artery disease or hypertension. Matrix metalloproteinases (MMPs), especially MMP-2, have been reported to underlie the pathogenesis of DM CMP by increasing extracellular collagen content.

Purpose

We hypothesized that two discrete MMP-2 isoforms (full length MMP-2, FL-MMP-2; N-terminal truncated MMP-2, NTT-MMP-2) are induced by high glucose stimulation in vitro and in an experimental diabetic heart model.

Methods

Rat cardiomyoblasts (H9C2 cells) were examined to determine whether high glucose can induce the expression of the two isoforms of MMP-2. For the in vivo study, we used the streptozotocin-induced DM mouse heart model and age-matched controls. The changes of each MMP-2 isoform expression in the diabetic mice hearts were determined using quantitative real-time polymerase chain reaction (qRT-PCR). Immunohistochemical stains were conducted to identify the location and patterns of MMP-2 isoform expression. Echocardiography was performed to compare and analyze the changes in cardiac function induced by diabetes.

Results

Quantitative RT-PCR and immunofluorescence staining showed that the two MMP-2 isoforms were strongly induced by high glucose stimulation in H9C2 cells. Although no definite histologic features of diabetic cardiomyopathy were observed in diabetic mice hearts, left ventricular systolic dysfunction was determined by echocardiography. Quantitative RT-PCR and IHC staining showed this abnormal cardiac function was accompanied with the increases in the mRNA levels of the two isoforms of MMP-2 and related to intracellular localization.

Conclusion

Two isoforms of MMP-2 were induced by high glucose stimulation in vitro and in a Type 1 DM mouse heart model. Further study is required to examine the role of these isoforms in DM CMP.