Regulation of Transforming Growth Factor–Beta in Diabetic Nephropathy: Implications for Treatment

Abstract

The recognition of the drivers of matrix accumulation as a therapeutic target for diabetic nephropathy is accepted by the Nephrology and pharmaceutical community. Interventions focused around Transforming Growth Factor–beta (TGF–β) will likely be an important area of clinical investigation in the near future. Understanding the various pathways involved in stimulating TGF–β in the diabetic kidney is of paramount importance in devising strategies to combat the development and progression of diabetic nephropathy. In this review we highlight the major pathways involved in stimulating TGF–β production by elevated glucose and discuss the therapeutic implications.

Transforming growth factor–β has been closely linked to the development and progression of diabetic nephropathy in cell culture, animal models, and the human condition ^1,2. It is likely that there will be several strategies to target TGF–β production and action as a novel means of therapy for diabetic nephropathy within the next ten years. In this review, we will highlight several aspects regarding the regulation of TGF–β in the context of diabetic kidney disease. An aspect that will be highlighted is that there are multiple pathways by which TGF–β may be stimulated in diabetic kidney disease thus complicating approaches that attempt to block one pathway and not others.

Regulation of TGF–β by glucose

Elevations in glucose drives stimulation of TGF–β in cell culture, animal models and human studies. We recently demonstrated that a glucose infusion for 2h to raise plasma glucose to the 200-250 mg/dl range led to stimulation of urinary levels of TGF–β1 in normal human subjects ³. This study demonstrates that transient elevation of blood glucose, even in healthy individuals, is sufficient to cause production of TGF–β. As urine levels of TGF–β1 increased and plasma levels were stable, the study suggested that the kidney may indeed be responding to hyperglycemia with increased TGF–β production. The mechanisms involved in this response remain to be proven in humans, however several pathways are likely involved based on experiments in cell culture and animal models.

Protein kinase C (PKC), reactive oxygen species (ROS), HETES, hexosamines, and the ERK-p38MAPK have all been implicated in mediating glucose induced stimulation of TGF–β. Protein kinase C has been found to be stimulated by glucose elevations in mesangial cells and vascular smooth muscle cells ^4-6. The isoform responsible for regulating TGF–β has not been conclusively demonstrated, although there is convincing data that the PKC–β isoform is intimately involved. The use of PKC–β inhibitors has been demonstrated to block TGF–β stimulation in animal models of diabetic kidney disease, as well as block matrix accumulation ^7,8. These pre-clinical studies have contributed to recognizing the possible role of PKC–β inhibition in human diabetic kidney disease ⁹. The PKC-a isoform is likely not involved in TGF–β stimulation as PKC-a knockout diabetic mice did not exhibit reduction of renal TGF–β levels as compared to wild type diabetic mice ¹⁰.

Reactive oxygen species (ROS) production appears to be critical in the pathophysiology of diabetic vascular complications. High glucose induces ROS in mesangial cells ¹¹ and ROS upregulates TGF-β and extracellular matrix (ECM) expression ¹². In addition, Nath et al. reported that H₂O₂ was able to induce TGF-β mRNA expression in rat kidneys and isolated fibroblasts ¹³. Scavenging of ROS by alpha-lipoic acid and manganese SOD dismutase leads to decreased TGF–β production in diabetic kidney disease and amelioration of diabetic renal pathology and albuminuria ^14,15. Specific pathways that lead to ROS production by high glucose in renal cells will need to be determined and may pave the way for directed therapies to focus on ROS-induced TGF–β production. One such pathway is NADPH oxidase and the isoform Nox4. Specific inhibition of Nox 4 blocks glomerular matrix accumulation in diabetic rats ¹⁶ and may be involved in stimulation of TGF–β. On the other hand, TGF-β itself could cause ROS production via NADH/NADPH oxidase ^17-19. We reported that TGF-ß1 induced ROS production in vascular smooth muscle and endothelial cells via NADPH oxidase and that Nox4 is involved in cytoskeletal alterations by TGF–β ^18,19.

Lipoxygenases (LO) are a family of enzymes that insert molecular oxygen into polysaturated fatty acids. They are classified as 5-, 8-, 12-, and 15-LO. 12-LO activation can produce 12(S)-hydroxyeicosatetraenoic acid [12(S)-HETE] ²⁰ and have been demonstrated to stimulate TGF–β in human macrophages ²¹. High glucose has been demonstrated to stimulate 12/15-LO expression, and 12(S)-HETE (12-LO product) induces cellular hypertrophy and fibronectin expression in rat mesangial cells ²². Recently, the Natarajan group ²³ clarified the cross-talk between the TGF-β and 12/15-LO pathway in MCs. Direct addition of 12(S)-HETE to rat mesangial cells stimulated the murine TGF–β1 promoter, TGF-β1 mRNA and protein expression, along with p-Smad 2/3 activation. Reciprocally, TGF-β treatment of rat mesangial cells increased 12/15-LO mRNA expression and 12(S)-HETE production, significantly. In addition, mesangial cells from 12/15-LO knock out mice expressed less TGF-β, and mesangial cells overexpressing 12/15-LO produced more TGF-β. The authors suggested that 12/15-LO and TGF-β could cross-talk and activate each other during the initiation and progression of diabetic kidney disease ²³.

Hexosamines, such as glucosamine-6-phosphate, can be formed under the control of the rate limiting enzyme glutamine:fructose-6-phosphate amidotransferase (GFAT). Immunostaining of GFAT demonstrated increased staining in diabetic kidneys, primarily in glomerular mesangial and epithelial cells ²⁴. Inhibition of GFAT by chemical inhibitors can block TGF–β expression in mesangial cells treated with high glucose ²⁵. The PKC and p38 pathway appears to be involved in mediating hexosamine induced TGF–β production in human mesangial cells ²⁶.

ERK as well as p38 MAPK are activated in mesangial cells exposed to high glucose and in rat glomeruli of early type 1 diabetes ^27,28. In a type 2 diabetic model, ERK activity was reported to be significantly activated in renal cortex of db/db mice as compared with non-diabetic mice ²⁹. ERK activation may well be due to upstream PKC activation as PKC inhibition also inhibits ERK activation ³⁰. There is a growing body of evidence supporting a role for p38 in diabetic kidney disease ³¹ and its regulation of TGF–β ³². The mechanism of high glucose-induced p38 MAPK in monocytes and mesangial cells may be mediated via reactive oxygen species ^33,34 and possibly independent of PKC. High glucose has been shown to stimulate p38 in mesangial cells, podocytes, endothelial cells ³⁵ as well as glomeruli in diabetic kidney disease ³⁶. Inhibition of p38 may mediate renal TGF–β production in several models of kidney disease ³⁷ and in proximal tubular cells ³².

Transcription factors involved in glucose induced TGF–β stimulation

The major transcription factors involved in mediating glucose induced TGF–β1 promoter activity has been postulated to include the AP-1 complex and the family of Upstream Stimulatory Factors (USF). Activation of PKC is well known to stimulate the c-fos and c-jun proto-oncogenes that form complexes for the AP1 binding site of the human and murine TGF–β1 promoter ^38,39. Mutagenesis of either one of the two or both AP-1 binding sites abolished the high glucose and PMA effect in the human promoter ³⁸. Furthermore, addition of the AP-1 inhibitor curcumin blocked the glucose response ³⁸. Interestingly, curcumin treatment of diabetic rats ameliorates kidney disease although no measurements of TGF–β1 was performed ^40,41. However, to data there are no in vivo studies to clearly implicate AP1 in glucose mediated renal TGF–β production.

Our group had initially identified the CACGTG element or E–βox as an important site for glucose regulation in the murine TGF–β1 promoter ⁴². High glucose increased binding of mesangial cell nuclear proteins to this E–βox element. Additional studies by our lab and others have now clearly determined that USF1 and USF2 are involved in binding the murine TGF–β1 promoter ⁴³ and the human TGF–β1 promoter ⁴⁴. In human mesangial cells, there is evidence of binding of both USF1 and USF2 to the human promoter ⁴⁴. High glucose induced stimulation of the human TGF–β promoter via USFs and the E–βox appears has been demonstrated to involve the hexosamine pathway ⁴⁴. In murine mesangial cells, chromatin immunoprecipitation assay revealed in vivo binding of USF1, but not USF2, to a glucose-responsive region of the TGF-ß1 promoter. Furthermore, glucose fluctuation with high carbohydrate feeding led to stimulation of renal TGF–β1 mRNA level in wild type and USF2 knockout mice but not in the USF1 knockout mice ⁴³. Therefore, there is convincing data that both USF1 and USF2 are involved in regulating TGF–β1 production by high glucose in mesangial cells and, at least in the murine system, the available evidence favors a dominant role for USF1. Future studies in various diabetic models are required to identify modulators of USF1 and USF2 to better understand the role of this family of transcription factors in the development of diabetic nephropathy.

TGF-β Activation: LAP, thrombospondin, and decorin

TGF-β is usually secreted in large latent complexes without biological activity. It consists of three components: a disulphide bonded homodimer of mature TGF-β, noncovalently bound to the latency-associated protein (LAP; homodimers of the N terminal fragment of precursor TGF-β) and a covalently attached molecule of latent TGF-β binding protein (LTBP) ^45-47. LAP and TGF-β compose the small latent complex. In this latent complex, TGF-β cannot bind to its surface receptors. Thus, the dissociation of TGF-β from LAP is a critical regulatory mechanism ⁴⁵. Administration of LAP reduces glucose induced fibronectin production in mesangial cells ⁴⁸, however no studies to date have comprehensively examined regulation of latency associated peptide with active TGF–β in the setting of diabetic kidney disease.

Thrombospondin 1 (TSP1) is a homotrimeric multifunctional glycoprotein expressed by a variety of cell types such as platelets, vascular smooth muscle cells and mesangial cells. It is frequently expressed at sites of inflammation and wound healing ⁴⁹, overexpressed in diabetic vessels ⁵⁰, diabetic kidneys ⁵¹ and secreted by mesangial cells ⁵². Murphy-Ullrich et al. ⁵³ first reported that TGF-β could bind to TSP1 under physiological conditions. More detailed studies identified two important sites in the TSP1 molecule that were responsible for this complex interaction. One is the WxxW (WSHW, WSPW or WGPW) motif from the type I repeats of the TSP1 that binds active TGF-β and the other site is the (K)RFK-sequence that binds the N-terminal LSKL-sequence of the LAP ^54,55. Additional studies indicate that high glucose-induced activation of TGF–β is largely dependent on TSP-1. There is convincing data that high glucose stimulates TSP1 gene transcription via the USF2 transcription factor ⁵⁶ and TSP1 then binds to the LAP thus dissociating active TGF–β. Interruption of this step may have therapeutic implications.

Decorin is a multi-functional extracellular proteoglycan ⁵⁷, and its core protein neutralizes TGF-β and antagonizes its prosclerotic effect ^58,59. Exogenous decorin suppressed TGF-β m RNA expression of the kidney ⁵⁸. Mogyorosi et al. reported a rapid and sustained increase of decorin mRNA expression in an animal model of type 1 diabetes ⁶⁰. Furthermore, decorin expression was enhanced by high glucose in mesangial cells and proximal tubular cells ⁶⁰. Stimulation of decorin by high glucose has been shown by several studies ^61,62 and appears to be via a CREB dependent pathway ⁶³. It was notable that decorin expression of mesangium and proximal tubular cells were suppressed by exogenous TGF-β treatment, both in high and normal glucose condition ⁶⁰. As both decorin and TGF-β were upregulated in diabetic state, Mogyorosi et al. speculated that TGF-β and decorin act in negative feedback loop with each other ⁶⁰.

In a human study, Schaefer et al. revealed that small proteoglycans including decorin mRNA were upregulated in all stages of diabetic nephropathy, both in the tubulointerstitium and in glomeruli ⁶⁴. The authors pointed out the glomerular expression and protein accumulation of decorin were not prominent, compared with mRNA expression. The authors suggested that this phenomenon could be explained by assuming that decorin were secreted into the mesangial matrix and then cleared via the vasculature or the urinary tract, in part as complexes with TGF-β. In advanced diabetic nephropathy, decorin deposition was found in fibrotic area and was co-localized with deposits of type I collagen ⁶⁴. Decorin may modulate progression of TGF-β mediated renal fibrosis through the formation of complexes of decorin-type I collagen-TGF-β. Decorin deficiency could result in an imbalance of decorin and TGF-β counter-regulation and lead to excess TGF-β activity and progressive matrix accumulation. Further studies are needed to clarify the relative role of decorin in mediating TGF-β activity and progression in diabetic nephropathy.

Role of Macrophage in TGF-β production

Most studies have focused on mesangial cells and proximal tubular cells as the cells responsible for TGF–β production in diabetic kidney disease ^32,65-67. However, there are a variety of other cell types that may contribute as much or more to renal TGF–β production in diabetic kidney disease. Macrophage are a rich source of TGF–β and it is now clear that macrophage infiltration is a characteristic feature of diabetic nephropathy ^{68 69}.

The role of macrophages in diabetic kidney disease remains to be clarified. Depletion of leukocytes by irradiation in the diabetic rat leads to reduced a3 (IV) collagen mRNA expression in the glomeruli ⁷⁰. Moreover, diabetic ICAM-1 knock out mice demonstrated reduced macrophage infiltration and decreased TGF-β and type IV collagen expression coincident with reduced mesangial matrix expansion and reduced albuminuria ⁶⁸. It is likely there is an important cell-cell interaction between macrophages and mesangial cells, as it has been reported that the culture supernatant of macrophages stimulates mesangial cells to produce fibronectin ⁷¹. Together with these findings, one could speculate that infiltrated glomerular macrophages and mesangial cells conspire to both secrete TGF-β and promote a positive feedback loop. A potential protective role may also apply to macrophages in early diabetic kidney disease. Glomerular mesangial cell production of hyaluronan leads to attraction of macrophages in early diabetes in the rat ⁷². Macrophages appear to be responsible for removal of glomerular hyaluronan and macrophages may then depart. If macrophages remain it is likely that they may contribute to pathology. As it is likely true for all infiltrating cell types the context in which the cell is present is critical in understanding its role. A complex dual role for macrophages has also been suggested in the development of atherosclerosis ⁷³ and in models of glomerulonephritis ^74,75.

TGF-β and podocytes

There is a growing body of evidence that podocytes occupy a significant role in the pathogenesis of diabetic nephropathy, diabetic glomerular proteinuria and matrix accumulation ^76-78. A further discussion of the effect of TGF–β on podocytes is discussed in another chapter. The role of the podocytes as a cell type producing TGF–β in diabetic nephropathy has not been clarified. Human diabetic nephropathy kidney sections have been demonstrated to have increased glomerular mRNA and protein for TGF–β1 in podocytes as well as mesangial cells ⁵¹. It remains to be established whether the high-glucose condition stimulates TGF-β expression in podocytes. In human podocytes, high glucose induced increased TGF-β expression in both protein and mRNA levels ⁷⁹. However, Iglesias-de la Cruz et al. ⁸⁰ reported that high glucose did not stimulate the production of TGF-β1 in cultured mouse podocytes, although high glucose did stimulate the expression of TGF-β type II receptor. In addition, Chen et al. demonstrated that AngII could induce podocyte dysfunction via TGF-β and VEGF activation, and these factors could lead to GBM thickness and albuminuria in diabetic nephropathy ^81,82. Of note, it is intriguing that anti-TGF–β approaches are associated with reduction of albuminuria in some experimental studies of diabetic nephropathy ⁸³ but not in others ^84,85.

Implications for Anti-TGF–β Approaches

Based on the convincing body of work implicating stimulation of TGF–β in the diabetic milieu and its critical role in progressive nephropathy ⁸⁶, a new theoretical basis for therapy emerges ^2,87. Emerging from the hemodynamic and metabolic control paradigms, the next decade or so may be characterized by an anti-fibrotic and cell based therapeutic approach. The multiple pathways by which TGF–β may be stimulated in the diabetic condition leads to consideration of a multi-pronged strategy to improve metabolic control, decrease hemodynamic stress, and reduce local angII effects. However, with the understanding of reactive oxygen species and glucose-induced intracellular signaling pathways to stimulate TGF–β1, it is clear that even sub-optimal glycemic control would lead to ongoing production of TGF–β in diabetes. Accumulating evidence suggests that blood glucose fluctuations are correlated with complications of diabetes ⁸⁸. Future studies that determine the metabolic pathways activated by repeated glycemic fluctuations are likely to be important in understanding the basis for renal TGF–β production and serve as targets for therapeutic interventions. Furthermore, there will also likely be heterogeneity in the pathways involved among various ethnic groups and genders. Translational studies in humans will facilitate future personalized approaches for interventions to block TGF–β production in the kidney.