Introduction
Diabetic nephropathy is characterised by damage and dysfunction of the microvasculature [
1]. A critical factor in maintaining the microvasculature is vascular endothelial growth factor (VEGF)-A, which regulates many aspects of vascular physiology, including vascular permeability and the migration, proliferation and survival of endothelial cells (for review, see Bartlett et al [
2]). Several studies in both human and animal models have indicated that proper glomerular function requires tight regulation of VEGF-A levels, as both upregulation and downregulation of VEGF-A can lead to kidney disease [
3].
Animal models of diabetic nephropathy develop increased levels of glomerular VEGF-A [
4,
5], possibly due to the effect of high glucose on VEGF-A production in podocytes [
6]. Therefore, inhibiting VEGF-A may be beneficial in treating renal complications. Consistent with this notion, antibodies directed against VEGF-A have been shown to prevent albuminuria [
7,
8] and glomerular hypertrophy [
9] in animal models of diabetes. However, in these studies, the inhibition of VEGF-A was initiated prior to the onset of diabetic kidney disease (i.e. prior to the development of albuminuria, glomerular hypertrophy and mesangial expansion/matrix production); thus, whether this strategy is feasible for treating diabetic people with existing kidney damage is currently unknown.
In addition to its role in maintaining vascular homeostasis, VEGF-A also facilitates the migration of monocytes and macrophages. Several studies have found that macrophages play a role in diabetic nephropathy [
10‐
12]. VEGF-A-induced migration of monocytes and macrophages is mediated by the binding of VEGF-A to VEGF receptor (VEGFR)-1 (also known as fms-related tyrosine kinase (FLT)-1) expressed on these cells [
13‐
15]. In addition, both VEGF-A [
16] and high glucose levels [
17] can activate endothelial cells, leading to increased levels of vascular cell adhesion molecule (VCAM)-1 and intercellular adhesion molecule (ICAM)-1, thereby promoting monocyte infiltration.
Here, we investigated whether the VEGF-A inhibitor soluble FLT-1 (sFLT-1; also known as soluble VEGFR-1) can reduce renal complications, including albuminuria and mesangial matrix expansion, in a mouse model of type 1 diabetes and pre-existing kidney damage. In addition, because diabetic nephropathy is accompanied by endothelial activation [
1] and macrophage infiltration [
11,
12,
18], both of which are mediated by VEGF-A, we also investigated the effect of inhibiting VEGF-A on these variables. Last, we investigated whether transfection with
sFlt-1 reduces glomerular TNF-α protein levels (a measure of inflammation) in diabetic mice.
Discussion
Here, we show that transfection with the VEGF-A inhibitor gene sFlt-1 in mice with diabetic nephropathy reverses pre-existing kidney damage by normalising albumin:creatinine levels and mesangial matrix content. Furthermore, transfection with sFlt-1 in diabetic mice also reduced endothelial activation (measured as VCAM-1, ICAM-1 and PECAM-1 protein levels), glomerular infiltration of macrophages and glomerular TNF-α protein levels. Finally, treating HUVECs with sFLT-1 decreased VEGF-A-induced endothelial activation in a dose-dependent manner. Taken together, these data suggest that treatment with sFLT-1 may be beneficial in individuals with diabetic nephropathy.
Animal models of diabetic nephropathy develop increased levels of glomerular VEGF-A [
4,
5], and inhibiting VEGF-A in diabetic animal models can prevent the development of albuminuria, glomerular hypertrophy and podocyte loss [
7‐
9,
23]. Consistent with these findings, podocyte-specific overexpression of
sFlt-1 has been reported to reduce mesangial expansion and glomerular basement membrane thickening in diabetic mice [
24]. However, that study did not investigate the effect of systemic sFLT-1 treatment, which will likely be required to treat individuals with diabetes. In contrast, other studies have found that anti-VEGF-A treatment has no effect on early renal pathology [
25], and that podocyte-specific deletion of
Vegfa in diabetic mice causes increased proteinuria and kidney damage [
21]. Moreover, although another study reported that treating diabetic mice with sFLT-1 decreased albuminuria, it did not reduce glomerular matrix deposition and led to an increase in tubular damage [
26]. These conflicting results could be due to a variety of factors, including the time at which treatment is initiated, and the dose and/or type of anti-VEGF-A treatment used (e.g. an anti-VEGF-A antibody, a VEGFR2 inhibitor or sFLT-1). For example, using a construct in which domain 2 of FLT-1 is linked to human IgG1Fc may lead to increased inflammation due to binding to Fc receptors on macrophages (for review, see Guilliams et al [
27]), increasing tubular damage [
26] independent of sFLT-1. In contrast, we used a full-length sFLT-1 construct without an Fc tag. In addition, VEGF-A inhibitors such as native sFLT-1 may have beneficial functions in addition to binding VEGF-A. For example, sFLT-1 has been reported to bind to lipid microdomains in podocytes, thereby affecting the actin cytoskeleton and the function of the glomerular barrier [
28]. Podocyte-specific deletion of
Flt-1 expression causes reorganisation of the cytoskeleton, leading to proteinuria and kidney damage; these effects are rescued by expressing a kinase-deficient mutant of
Flt-1, suggesting that physiological levels of sFLT-1 are necessary for the proper structure and function of podocytes [
28]. Therefore, with respect to kidney damage, treating individuals with sFLT-1 may provide improved outcomes compared with anti-VEGF-A antibodies and VEGFR2 inhibitors.
Importantly, the studies discussed above investigated the prevention—rather than the treatment—of diabetes-induced kidney damage, as therapy was initiated before the onset of kidney damage. Therefore, it is difficult to estimate the effects of such treatments in diabetic individuals who have already developed kidney damage. Fioretto et al reported that kidney lesions in diabetic individuals were reversed by normalising glycaemia levels as a result of pancreatic transplantation [
29]. Therefore, we tested the effect of treating diabetic mice with the VEGF-A inhibitor sFLT-1 after the onset of kidney damage, including albuminuria and mesangial matrix accumulation. We found that even though transfection with
sFlt-1 did not normalise blood glucose levels in diabetic mice (ESM Fig.
2), kidney damage was reversed, as both albuminuria and mesangial matrix accumulation were reduced.
Several studies have reported that macrophages play a role in the development of diabetic nephropathy [
10‐
12]. Moreover, VEGF-A plays a role in the migration of monocytes and macrophages [
13] by binding the FLT-1 receptor on these cells [
14,
30]. In addition, incubating endothelial cells with either glucose [
17] or VEGF-A [
16] results in endothelial activation, a key event in the adhesion and migration of monocytes from the circulation into the tissue. Consistent with this finding, both animals and people with diabetes have increased levels of endothelial activation [
31‐
33]. Furthermore, we found that incubating HUVECs with VEGF-A increased endothelial activation, and that this effect was reversed by treating the cells with sFLT-1. We also found that transfection with
sFlt-1 normalised both the number of glomerular macrophages and the level of TNF-α in diabetic mice. Taken together, these findings suggest that the VEGF-A inhibitor sFLT-1 reduces endothelial activation and subsequent macrophage infiltration. Treatment with sFLT-1 has reported benefits in treating other diseases, including arthritis [
34,
35], vascular disease [
36,
37], sepsis [
38] and psoriasis [
39]; these clinical benefits are attributed primarily to reduced numbers of infiltrating macrophages and reduced inflammation. The current results indicate that sFLT-1 may be a valuable treatment for diabetic nephropathy, as well as other diseases in which inflammation plays an important role. Macrophages produce cytokines such as TNF-α and TGF-β, which increase the production of matrix proteins by mesangial cells [
40]. Thus, reducing the number of glomerular macrophages using sFLT-1 might also reduce mesangial matrix expansion in diabetic nephropathy.
As reviewed by Deeds et al [
41], techniques using STZ (such as dosage and administration) and consistency with respect to the resulting diabetes mellitus in small animal models have not been standardised. In our study, we used a moderate dosing regimen of three doses of 75 mg/kg STZ, for two reasons: (1) this regimen is less nephrotoxic than a single high dose; and (2) this regimen induces more diabetes-related histological damage compared with several low doses, which result in a relatively mild phenotype. In rodents, STZ can cause nephrotoxicity; however, Kraynak et al have reported that STZ-induced cellular and molecular damage resolves within 3 weeks [
42]. This suggests that the albuminuria seen in our diabetic mice at 5 weeks was probably related to diabetes rather than to STZ. This is supported by the histological characteristics typical of diabetic nephropathy seen in these diabetic mice (i.e. mesangial matrix expansion and glomerular hypertrophy). Although some groups have reported albuminuria and histological lesions at this time point [
43‐
45], other groups did not find albuminuria at this time point [
31,
46]; this discrepancy may be due to differences in the dose and/or route of administration of STZ. It is important to note that although present, the albuminuria in our STZ-injected diabetic mice was not exceedingly high, and we suggest that the importance of albuminuria in C57BL/6 mice must be considered in combination with the presence (or absence) of histological findings.
Importantly, we found a small, but significant, decrease in podocyte numbers in
sFlt-1-transfected diabetic mice; decreased numbers of podocytes have also been reported in pre-eclampsia, which is characterised by high circulating levels of sFLT-1 [
47]. Despite this decrease in podocyte numbers, albuminuria was significantly reduced in
sFlt-1-transfected diabetic mice. It is possible that the decrease in podocyte numbers in these
sFlt-1-transfected mice was too small to functionally affect the filtration barrier. This notion is supported by previous reports that a substantial decrease in podocyte numbers is required for increased albuminuria [
48,
49]. Nevertheless, we cannot exclude the possibility that longer treatment and/or higher levels of sFLT-1 expression could affect the glomerular filtration barrier. Thus, we hypothesise that sFLT-1 will likely have a beneficial effect in people with diabetes until the production of VEGF-A by podocytes drops below a certain threshold, given that decreased VEGF-A levels also result in kidney damage [
21]. In this respect, it is important to note that both VEGF-A and sFLT-1 levels should be adjusted with care, as both increased and decreased levels of VEGF-A can lead to renal pathology [
3,
50].
In conclusion, we report that normalising VEGF-A levels with sFLT-1 might be a viable approach for treating individuals with existing diabetic nephropathy by reducing endothelial activation, glomerular macrophage infiltration and glomerular inflammation, thereby reversing kidney damage.