The major manifestations of nephrotic syndrome include proteinuria, hypoalbuminemia, edema, hyperlipidemia and lipiduria. Common causes of nephrotic syndrome are diabetic nephropathy, minimal change disease (MCD), focal and segmental glomerulosclerosis (FSGS), and membranous nephropathy (MN). Among the primary glomerular diseases, MCD is usually glucocorticoid sensitive, whereas the response in other diseases is variable (1). Despite recent identification of key structural proteins in the glomerular capillary loop, many disease mechanisms in nephrotic syndrome remain unresolved. In this study, we show that the glomerular expression of Angiopoietin-like 4 (Angptl4), a secreted glycoprotein, is glucocorticoid sensitive, and is highly upregulated in podocytes in experimental and human MCD. Podocyte-specific transgenic overexpression (NPHS2-Angptl4) in rats induces nephrotic range selective proteinuria (500 - fold increase in albuminuria), loss of GBM charge and foot process effacement, whereas adipose tissue-specific transgenic expression (aP2-Angptl4) results in increased circulating Angptl4 levels, but no proteinuria. Angptl4−/− mice injected with lipopolysaccharide or nephritogenic antisera develop significantly lower proteinuria than controls. Angptl4 secreted from podocytes in some forms of nephrotic syndrome lacks normal sialylation, and feeding sialic acid precursor N-Acety-D-mannosamine (ManNAc) to NPHS2-Angptl4 transgenic rats increases sialylation of Angptl4 and reduces albuminuria by over 40%. These studies suggest a key role of podocyte secreted Angptl4 in nephrotic syndrome.
As part of our studies to identify novel mechanisms of proteinuria, we injected rats with the γ2 fraction of nephtotoxic serum (γ2-NTS), an antiserum reactive to multiple podocyte proteins that induces complement- and leukocyte-independent glomerular injury by direct antibody binding (2). A panel of differentially expressed glomerular genes was assembled (3), and two genes not previously known to be involved in the pathogenesis of proteinuria investigated in detail. One of these genes encodes for the transcriptional factor zinc fingers and homeoboxes 3 (Zhx3) expressed in podocytes, and has now been shown to play a significant role in the pathogenesis of primary glomerular disease (3). The other gene, Angptl4, was highly upregulated in the podocyte, and is the focus of investigation in this paper.
Angiopoietin-like proteins have been implicated in the development of hypertriglyceridemia (4) and tumor metastasis (5), and have some functional properties that are different from the angiopoietins. Angptl4 is a Pparγ (6) and Pparα (7) target gene highly expressed in the liver and adipose tissue, strongly induced by fasting in white adipose tissue and liver, and is an apoptosis survival factor for vascular endothelial cells under normoxic conditions (8). Angptl4 is a potent inhibitor of lipoprotein lipase (LPL) (4), and induces significant hypertriglyceridemia following intravenous injection or adenovirus-mediated expression (9, 10). Other studies show lesser expression in cardiomyocytes and skeletal muscle, and low level expression in whole kidney on Northern blot analysis (6). Recent population based studies of the ANGPTL4 gene reveal variants that affect triglyceride levels in humans (11, 12). Most normal circulating Angptl4 in rodents is secreted from the liver as a cleaved protein that binds to HDL particles (13). A role of Angptl4 in proteinuria has not been previously reported.
We noted severe upregulation (70.45 ± 4.14, mean ± SEM) of Angptl4 mRNA in rat glomeruli at the peak of complement- and leukocyte-independent heterologous phase proteinuria 24 hours after injection of γ2-NTS (Fig. 1a, b, Supplementary Fig. 1). Injection of Angptl4−/− and Angptl4+/+ mice with LPS (Fig. 1c) and γ2-NTS (Supplementary Fig. 2a,b) induced significantly less proteinuria (Fig. 1c, Supplementary Fig. 2a) and foot process effacement (Supplementary Fig. 2b) in Angptl4−/− mice. Normal rat glomeruli express Angptl4 in a capillary loop pattern that co-localizes with podocyte protein CD2 adapter protein (CD2AP) (Fig. 1d). We noted early (Day 3, before the onset of proteinuria) and progressive upregulation of Angptl4 mRNA expression in young rats following intravenous injection of a single dose of puromycin aminonucleoside (PAN model, peak up to 80-fold increase in different studies) (Fig. 1e), and in situ hybridization confirmed upregulation in a peripheral capillary loop pattern (Supplementary Fig. 2c). In passive Heymann nephritis, we saw a smaller increase in Angptl4 expression, starting after the onset of proteinuria (Fig. 1e). Angptl4 mRNA expression did not change in anti-Thy1.1 nephritis, or in collapsing focal and segmental glomerulosclerosis (FSGS) (Fig. 1e) (14). Angptl4 protein expression increased dramatically in podocytes (Fig. 1f) following induction of PAN, with substantial additional overlap with the GBM, and this was confirmed by immunogold electron microscopy (EM) (Fig. 1g). We also noted reduced expression of GBM heparan sulfate proteoglycans (HSPG) in PAN glomeruli (Fig. 1f).
To study the biological role of Angptl4 upregulation in nephrotic syndrome, we first assessed Angptl4 expression in the podocyte in a previously published Angptl4 transgenic mouse model (13) (Fig. 2a–2d, Supplementary Fig. 2d,e). Glomeruli appeared normal on light microscopy (Fig. 2a). We noted increased glomerular expression of Angptl4, but not of another podocyte expressed protein Zona Occludens 1 (ZO-1) (Fig. 2a) in 3 month old transgenic mice. Angptl4 expression co-localized with podocyte expressed CD2AP (Supplementary Fig. 2d). Electron microscopy revealed 50% effacement or broadening of foot processes (Fig. 2b). immunogold electron microscopy revealed a large number of gold particles in podocyte foot processes, GBM and close to the endothelial cell surface (Fig. 2c), especially in areas opposite to foot process effacement (Fig. 2c, right panel). These mice had mild proteinuria (Fig. 2d).
To study higher levels of Angptl4 upregulation noted in disease states, we generated transgenic rat models, one each for podocyte specific (NPHS2-Angptl4) and adipose tissue specific (aP2-Angptl4) transgenic over expression (Fig. 2e). Analysis of mRNA expression in organs that normally express Angptl4 confirmed specificity of expression (Fig. 2f, Supplementary Fig. 2f). In keeping with the normal expression pattern of the NPHS2 promoter, we noted significant expression in the pancreas. Histological assessment of 3 month old heterozygous male NPHS2-Angptl4 transgenic rats revealed normal appearing glomeruli with prominent podocytes on light microscopy (Fig. 2g) and increased glomerular Angptl4 expression by confocal imaging (Fig. 2h) that overlaps with podocyte nephrin and GBM heparan sulfate proteoglycans (HSPG). Electron microscopy of 5 month old homozygous TG rats revealed diffuse foot process effacement (Fig. 2i, Supplementary Fig. 3a). Using immunogold electron microscopy in homozygous NPHS2-Angptl4 transgenic rats, we observed a correlation between gold particle clustering in the GBM and the development of foot process effacement, which is first noted around age 3 months (Fig. 2j). In situ hybridization for Angptl4 (Supplementary Fig. 3b) revealed selective increase in the signal in glomeruli in a peripheral podocyte distribution. The aP2-Angptl4 transgenic rats had morphologically normal glomeruli by light (Fig. 2g) and electron microscopy (not shown).
We noted significant albuminuria in both NPHS2-Angptl4 TG rat founder lines. Female homozygous and male heterozygous rats developed albuminuria at age 1 month (Fig. 3a–3c, Supplementary Fig. 3c). Homozygous females developed up to 100-fold, heterozygous males up to 20-fold and homozygous males over 500-fold increase in albuminuria over time. Heterozygous females were not albuminuric. Over 90% of the urinary protein comprises of intact albumin (Fig. 3d, Supplementary Fig. 3d), thereby making these rats the first model of selective proteinuria. Blood pressure was significantly lower in proteinuric heterozygous NPHS2-Angptl4 transgenic rats compared to wild type controls (Supplementary Fig. 3e). By contrast, aP2-Angptl4 transgenic rats did not develop albuminuria (Fig. 3d), despite high circulating Angptl4 levels (Supplementary Fig. 4a–4d). Despite upregulation of Angptl4 in pancreas in NPHS2-Angptl4 transgenic rats, circulating Angptl4 levels were not elevated, and were in fact marginally reduced (Supplementary Fig. 4a–4d) due to presence of proteinuria in these rats. In keeping with the proteinuric effects of podocyte secreted Angptl4, NPHS2-Angptl4 transgenic rats developed more proteinuria (Fig. 3e) than wild type littermates after induction of PAN. Serum electrolytes, creatinine and glucose levels were indistinguishable from wild type rats.
As also previously published (15), proteinuria in PAN is partially glucocorticoid sensitive on day 6 (Fig. 3f). mRNA expression studies for Angptl4 revealed significant reduction in Angptl4 expression in glucocorticoid treated rats on day 6 (Fig. 3g), thereby suggesting that Angptl4 was a glucocorticoid sensitive gene in vivo. Early increase in Angptl4 mRNA expression noted on PAN Day 3 (before the onset of proteinuria) was also reversed in glucocorticoid treated rats. In addition, Angptl4 oligomers were also detectable in the urine in experimental models of proteinuria (Supplementary Fig. 4e,f).
We investigated loss of GBM charge in transgenic rats and mice (Supplementary Fig. 5). NPHS2-Angptl4 transgenic rats had significant loss of GBM charge by alcian blue and polyethyleneimine methods (Supplementary Fig. 5a,b,d). As also seen in PAN rats (Fig. 1f), GBM heparan sulfate proteoglycan is also reduced in NPHS2-Angptl4 transgenic rats (Supplementary Fig. 5c).
Immunoblotting of glomerular proteins typically underestimates Angptl4 production, since the protein is rapidly secreted. On 2D gel electrophoresis, most Angptl4 in normal glomeruli migrated as glycosylated low order oligomers at neutral pI (Fig. 4a), though less prominent spots for both glycosylated intact 70 KDa and cleaved / non glycosylated 45 KDa monomeric forms were also noted (Supplementary Fig. 6a,b). The 45 kDa protein, but not the oligomers, were threonine phosphorylated (Supplementary Fig. 6a). In PAN, both neutral pI oligomers, that were reactive with sialic acid binding lectin Maackia amurensis (MAA), and high pI oligomers, that were not MAA reactive, were increased (Fig. 4a,b). This increase was blunted in PAN treated with glucocorticoids, though disproportionately higher amounts of high pI oligomers were noted (Fig. 4a,b).
Next, we studied the sialylation of Angptl4 (Fig. 4, Supplementary Fig. 6) in vitro (Fig. 4c, d, Supplementary Fig. 6c,d) using the Angptl4-HEK293 (Supplementary Fig. 7a–7c) and Angptl4-GEC stable cell lines (Supplementary Fig. 7d,e). Incubation of both stable cell lines with sialic acid precursor ManNAc resulted in a shift of the secreted Angptl4 protein towards neutral pI, with increased reactivity to MAA, thereby suggesting that sialylation played a significant role in the differential electrophoretic migration of Angptl4 (Fig. 4c,d).
Since an increase in hyposialylated Angptl4 was noted in PAN, a potential role of this lack of sialylation in the pathogenesis of proteinuria was studied. Over half of glomerular Angptl4 in NPHS2-Angptl4 transgenic rats was hyposialylated and migrated at a high pI (Fig. 4e,f, Supplementary Fig. 6e). NPHS2-Angptl4 transgenic rats that received ManNAc in tap water had reduced albuminuria (Fig. 4g, Supplementary Fig. 8a,b), with 40.6 ± 3.3% (mean ± SEM) reduction in albuminuria over a 12 day period compared to baseline, and returned to levels comparable with control untreated transgenic rats after 24 days of washout. Densitometry analysis of Western blots for glomerular Angptl4 from rats euthanized on Day 12 (Fig. 4f) showed an increase in the neutral pI Angptl4 fraction from 43.3 ± 4.2% (mean ± SEM) in control rats to 72.9 ± 1.4% (mean ± SEM) in ManNAc treated rats. The neutral pI fraction was also reactive with the sialic acid binding lectin Sambucus nigra (SNA I) (Fig. 4e), confirming increased sialylation of Angptl4 in ManNAc treated rats. We did not note any changes in the quantitative expression of podocalyxin, another sialylated glomerular glycoprotein important in podocyte structural integrity, or significant shifts towards a lower pI (below 7) in ManNAc treated TG rats (Supplementary Fig. 6f–h).
In normal rodents, circulating Angptl4 is mostly HDL bound cleaved protein secreted by the liver (13). Analysis of oligomer formation by Angptl4 in transgenic rats showed that most glomerular Angptl4 was either monomeric or formed low order oligomers, whereas circulating Angptl4 formed mostly middle order and high order oligomers (Supplementary Fig. 8c). This may be one reason why the circulating form does not enter the GBM in aP2-Angptl4 transgenic rats. Most circulating Angptl4 in these rats migrates between pI 5 and 7 (Supplementary Fig. 4a,b), and is therefore unlikely to avidly bind anionic GBM proteins in a charge dependant manner.
Of the various forms of primary glomerular disease examined, Angptl4 expression was most significantly upregulated in experimental MCD, less prominently and later in the disease course in experimental membranous nephropathy, and was unchanged in experimental FSGS. Therefore, Angptl4 expression was studied in biopsies from individuals with MCD by confocal imaging (Supplementary Fig. 9). Increased expression in a podocyte distribution with additional GBM overlap was noted in all 5 biopsies. In addition, Angptl4 oligomers were detectable in the urine of an additional four individuals with MCD (Supplementary Fig. 10a,b, Supplementary Table 1), and distinctive patterns of circulating Angptl4 expression were noted in their plasma (Supplementary Fig. 10c–10e).
The effects of podocyte secreted Angptl4 on the glomerular capillary loop in nephrotic syndrome are illustrated in Supplementary Fig. 11. Transgenic expression of Angptl4 from the podocyte reproduced some of the key features of human MCD, including the classic morphologic changes, nephrotic range selective proteinuria, and loss of GBM charge. One feature that is prominently missing is the explosive onset of proteinuria. It is possible that this feature is contributed to by other genes / proteins concomitantly dysregulated with Angptl4 in human MCD, and that these putative proteins may accelerate the interaction or transport of Angptl4 across the glomerular capillary loop. These results demonstrate that Angptl4 may be involved in the pathogenesis of MCD. Development of albuminuria in NPHS2-Angptl4 transgenic rats at age 1 month, when podocyte foot processes are intact and scattered immunogold particles are noted in the GBM, suggests that podocyte secreted Angptl4 induces a filtration defect in the GBM. Clustering of immunogold particles in the GBM coincides with the development of foot process effacement and a sharp increase in proteinuria around age 3 months, suggesting a complex interaction between Angptl4, GBM proteins and the podocyte – GBM interface.
Glucocorticoid sensitivity of Angptl4 in PAN favors a significant role for this gene in glucocorticoid sensitive nephrotic syndrome. Other podocyte expressed glucocorticoid sensitive genes have been described in literature (15). Mutations in one of these genes, NPHS2, most often results in the development of glucocorticoid resistant FSGS (16). Of proteins secreted by the podocyte, published studies suggest a role of VEGFA in human thrombotic microangiopathy (17), whereas experimental overexpression of Angpt2 in mice results in a 2 – 2.5 fold increase in albuminuria (18). Neither of these genes is known to be directly involved in the pathogenesis of glucocorticoid sensitive nephrotic syndrome.
Absence of proteinuria in aP2-Angptl4 transgenic rats, despite high circulating Angptl4 levels, highlights the importance of production of Angptl4 in the podocyte in proteinuric disease. Also, the commonly observed 60 to 80 fold peak upregulation of glomerular Angptl4 expression in PAN is close to the 120-fold increase in Angptl4 expression noted in proteinuric heterozygous male NPHS2-Angptl4 transgenic rats, thereby making these rats a suitable model for future studies on nephrotic syndrome.
The secretion of high pI Angptl4 by the podocyte in PAN is likely to facilitate the tethering of Angptl4 to the GBM, and prior studies have shown the binding of Angptl4 to heparan sulfate proteoglycans (19). This correlates well with reduced GBM charge in rodents with transgenic expression of Angptl4 from the podocyte. A relationship of GBM charge with proteinuria is hard to conclude from these studies, since despite mild proteinuria in the transgenic mouse model, the reduction in GBM charge is comparable between the Angptl4 transgenic mouse and NPHS2-Angptl4 transgenic rat models. It is likely that charge facilitates the transit of high pI Angptl4 across the GBM against the direction of fluid flow.
Changes in sialylation may explain some of the variations in pI, since increasing the sialylation of Angptl4 in vitro induces migration from high towards neutral pI. There are at least two potential O-glycosylation sites in rat and human Angptl4 where sialic acid residues could be incorporated. The reduction in albuminuria in NPHS2-Angptl4 transgenic rats fed with ManNAc, and increase in the sialylation of glomerular Angptl4 suggest that hyposialylation may be one mechanism by which Angptl4 overexpression contributes to proteinuria. Therefore, treatment with sialic acid precursors constitute a potential therapeutic tool to reduce proteinuria in some forms of nephrotic syndrome. ManNAc therapy has been used recently to improve muscle weakness (20) and the severe glomerular phenotype (21) in mice harboring mutations in the GNE/MNK gene seen in patients with hereditary inclusion body myopathy. The kidney phenotype noted (21) however does not have a counterpart in humans with this disease.
The introduction of Angptl4 in the field of proteinuria investigation is likely to stimulate additional studies of its role in other aspects of nephrotic syndrome. A recent publication (22) suggests that glucocorticoids increase Angptl4 expression in adipose tissue, and could potentially increase levels in the circulation. Circulating Angptl4 increases plasma triglyceride levels by inhibiting tissue uptake via inactivation of endothelial bound LPL activity in target organs (4). Future studies will address whether some of the hypertriglyceridemia in nephrotic syndrome is contributed to by Angptl4 leaking from glomeruli into the circulation. It would be equally important to study the interaction of Angptl4 with specific GBM proteins, and identify DNA motifs in the Angptl4 gene that mediate glucocorticoid sensitivity.