Introduction

Diabetic nephropathy (DN) is the leading cause of end-stage renal disease, accounting for 44% of new cases each year [1]. Hyperglycaemia and hypertension are established causes of DN; however, elevated plasma lipids are also linked with progression to microalbuminuria [2]. Several investigators have hypothesised that lipids play a causal role in the progression of chronic kidney disease, noting the similarities between the pathogenesis of atherosclerosis and glomerulosclerosis. These include: influx of monocytes, build-up of lipid-laden cells, and the presence of cholesterol and cholesteryl esters [3–5]. Several cross-sectional studies of patients with type 1 diabetes have indicated that elevated apolipoprotein B or LDL-cholesterol and/or decreased levels of HDL-cholesterol are associated with progression of albuminuria [6–10]. Moreover, experiments in animal models of diabetes fed a Western diet have also supported the notion that sustained lipid abnormalities accelerate the progression of DN [11–14].

The liver X receptors (LXRs) α and β are nuclear hormone receptors. They are activated by endogenous oxysterols, which are natural products derived from the metabolism of cholesterol. LXRs are considered to be intracellular ‘cholesterol sensors’ because they activate transcription in response to cholesterol metabolites [15]. LXR activation by a synthetic ligand was atheroprotective in animal models of dyslipidaemia [16]. Mechanistically, this atheroprotective effect has been attributed to enhanced reverse cholesterol transport and inhibition of inflammatory responses upon LXR activation [16, 17]. Direct LXR target genes include those encoding the ATP-binding cassette (ABC) transporter A1 (ABCA1) and the ABC transporter G1 (ABCG1), which govern cholesterol efflux from macrophages and peripheral organs, and the inducible degrader of LDL receptor (Idol [also known as Mylip]), which decreases LDL receptor protein levels and peripheral cholesterol uptake [18, 19]. In macrophages, LXR activation inhibits the expression of inflammatory cytokines (i.e. IL-1β, IL-6 and TNFα) [20, 21], monocyte chemoattractant protein and osteopontin [22]. Unfortunately, first-generation LXR agonists (T0901317, GW3965) caused hypertriacylglycerolaemia through upregulation of hepatic sterol regulatory element-binding protein 1c (SREBP1c) [23]. N,N-Dimethyl-hydroxycholenamide (DMHCA) is a new-generation LXR gene-selective agonist, which reduces plaque formation in animal models of atherosclerosis without inducing hypertriacylglycerolaemia [24, 25].

Human biopsies and experimental rodent models have revealed a consistent downregulation of LXRs in DN compared with control [26, 27], and experiments with first-generation LXR ligands have shown anti-inflammatory and anti-albuminuric actions in diabetic mice [28, 29]. While these observations are promising for the development of DN therapies targeting LXRs, several fundamental questions remain. First, whereas LXR agonism may slow renal decline, the effects of LXR downregulation (as observed in patients) on the natural progression of renal decline are unknown. Second, although LXRs appear to prevent inflammation within the renal tubulointerstitium, the role of these receptors in preserving glomerular homeostasis is incompletely defined, considering the importance of glomerular injury in the pathogenesis of DN and the broad range of LXR target genes. Finally, while first-generation LXR agonists have provided proof of concept for the biological importance of this pathway in DN, they are not clinical drug candidates because of their undesirable effects on plasma and liver triacylglycerol levels. To address these deficiencies, we performed experiments in Lxrα/β −/− (Lxrα, also known as Nr1h3; Lxrβ, also known as Nr1h2) mice and in mice treated with a new-generation LXR ligand, DMHCA. We hypothesised that Lxrα/β −/− mice (which lack the ability to regulate intracellular cholesterol) would have hastened progression of DN and, conversely, that activation of LXR would protect against DN through improved renal cholesterol homeostasis and inhibition of inflammatory activators. Here, we present our data outlining the important role of LXR in modifying intraglomerular lipid accumulation in the context of DN.

Methods

Animals

All animal procedures were performed in compliance with the Principles of Laboratory Animal Care (NIH publication number 85–23, revised 1985; http://grants1.nih.gov/grants/olaw/references/phspol.htm) and were approved by the Institutional Animal Care Committees at the Universities of Toronto and Colorado. To avoid the confounding effects of insulin on lipogenesis, we explored the role of LXR in insulinopenic models of DN induced by streptozotocin (STZ). For the first animal model, based on Lxrα/β −/− mice with STZ-induced diabetes, we used male Lxrα/β −/− and matching wild-type mice (C57Bl/6:129SvEv) [30], which were maintained on chow (2016 Teklad Global rodent diet; Harlan Teklad, Mississauga, ON, Canada). At 8 weeks, mice were given daily i.p. injections of STZ (Sigma-Aldrich, Oakville, ON, Canada) or vehicle (citrate) (Lxrα/β +/+ 50 mg/kg; Lxrα/β −/− 35 mg/kg) for 5 consecutive days. All mice were switched to a Western diet (21% wt/wt milk fat, 0.2% wt/wt cholesterol) (TD88137; Harlan Teklad) at 1 week after injections had started, and followed for 14 weeks. The second animal model was based on DBA/2J mice with STZ-induced diabetes. We chose DBA/2J male mice (Jackson Laboratories, Bar Harbor, ME, USA) because they are more prone to developing DN and do so at an earlier age than C57Bl/6 mice [31, 32]. At 8 weeks, these mice were treated with 40 mg/kg STZ or vehicle (citrate) for 5 consecutive days and then placed on a Western diet (TD88137) containing vehicle or the LXR agonist DMHCA (80 mg/kg daily) for 10 additional weeks (see electronic supplementary material [ESM] Methods).

Urine and plasma analyses

Blood was collected into EDTA-coated tubes and plasma was analysed using commercially available assays (see ESM Methods for further details).

Histology and immunofluorescence

One kidney was sectioned for paraffin embedding (fixed in formalin), optimal cutting temperature compound (OCT) embedding (fixed in formalin then sucrose-protected in a 10% wt/vol. and 30% wt/vol. sucrose solution) and electron microscopy (fixed in glutaraldehyde). See ESM Methods for details on staining and quantification.

Kidney lipid analysis

Tissues were digested in chloroform : methanol (2:1) with a 20:1 volume to tissue ratio, and processed as described previously [30].

RNA isolation and gene expression analysis

RNA was isolated and cDNA synthesised as described previously [30]. Gene expression levels were measured by high-throughput real-time quantitative PCR (7900HT; Applied Biosystems, Foster City, CA, USA) using the \( {2}^{-\Delta \Delta {\mathrm{C}}_{\mathrm{t}}} \) method and normalised to cyclophilin. Primer sequences are given in ESM Table 1.

DMHCA liquid chromatography and mass spectrometry analysis

Dried kidney lipid extracts were re-suspended in 250 μl HPLC-grade methanol and analysed by liquid chromatography (LC) and mass spectrometry (MS) using a device (6410 Triple Quad LC/MS; Agilent Technologies, Santa Clara, CA, USA) with the electrospray ionisation source in positive ion mode (details, see ESM Methods).

Western blot analysis

Protein extracts (80 μg) from kidney cortex were separated by SDS-PAGE and transferred to polyvinylidene difluoride membranes or nitrocellulose membrane (ABCA1 only) (see ESM Methods for further details).

Statistical analysis

ANOVA was used for comparisons between all groups, followed by Student–Newman–Keuls tests. The sample size for each group represents the number of animals that survived until the end of the treatment period, unless otherwise stated. All values are expressed as mean ± SEM; a value of p < 0.05 was considered statistically significant (see ESM Methods and ESM Table 2 for details).

Results

LXR deficiency increased the severity of diabetic kidney disease

To assess the role of LXR and cholesterol accumulation in the progression of DN, we induced diabetes in Lxrα/β +/+ and Lxrα/β −/− mice using STZ and a Western diet. Both STZ-treated groups were hyperglycaemic, with Lxrα/β +/+ mice achieving 15% higher glucose levels than Lxrα/β −/− mice (Table 1). After 14 weeks, diabetic Lxrα/β +/+ mice had twofold higher levels of albuminuria than non-diabetic controls (Fig. 1a). Remarkably, non-diabetic Lxrα/β −/− mice had a tenfold higher albumin : creatinine ratio (ACR) than Lxrα/β +/+ mice, with no further increase observed following STZ administration in Lxrα/β −/− mice (Fig. 1a). In contrast, urinary nephrin excretion was significantly increased in STZ Lxrα/β −/− mice compared with matching normoglycaemic mice, suggesting that structural features of renal pathology were worsened with hyperglycaemia in Lxrα/β −/− mice (Fig. 1b). Synaptopodin showed decreased staining in Lxrα/β +/+ STZ-treated animals and Lxrα/β −/− mice compared with the wild-type control group (Fig. 1c). These data suggest that Lxrα/β −/− mice on a Western diet have an impaired glomerular filtration barrier compared with Lxrα/β +/+ mice, and that this impairment is further exacerbated by diabetes.

Table 1 Physiological variables for non-diabetic and diabetic Lxrα/β +/+ and Lxrα/β −/− mice
Full size table
Fig. 1
figure 1

Diabetic Lxrα/β −/−mice have a compromised glomerular filtration barrier. (a) Urinary ACR and (b) urinary nephrin:creatinine ratio after 14 weeks on a Western diet. V, vehicle. (c) Representative immunofluorescence staining of kidney sections for the podocyte marker synaptopodin (red) at ×200 magnification. Veh, vehicle. (d) Oil Red O staining of kidney sections showing significant accumulation of neutral lipids in glomeruli of diabetic Lxrα/β −/− mice compared with diabetic Lxrα/β +/+mice at ×400 magnification. (e) Glomerular lipid accumulation was quantified by normalising the Oil Red O intensity to total glomerular surface area. The bars represent the average of kidney samples from two mice (individual values are shown as white circles). A total of 25 glomeruli were quantified for each animal. Data (a, b) are means ± SEM; n = 6 (a), except for Lxrα/β −/− (n = 3); n = 5 (b), except for Lxrα/β −/− (n = 3), please note that sample sizes were smaller because spot urine was not sufficient from all mice. *p < 0.05 compared with vehicle control of the respective genotype and † p < 0.05 compared with Lxrα/β +/+ of the same treatment group, tests by one-way ANOVA and Student–Newman–Keuls

Full size image

Diabetic Lxrα/β−/− mice showed increased neutral lipid accumulation in glomeruli compared with wild-type mice despite their similar circulating cholesterol levels

To determine whether the severity of DN correlated with renal lipid accumulation, we analysed kidney sections by histology and measured bulk lipids after organic extraction. Oil Red O staining for neutral lipids was primarily in the glomeruli, with sporadic staining in the tubulointerstitial region of STZ Lxrα/β −/− mice (Fig. 1d, ESM Fig. 1). Image analysis showed a 41-fold higher level of staining in non-diabetic Lxrα/β −/− mice than in Lxrα/β +/+ mice, and this was further increased by twofold after STZ treatment (Fig. 1e). Total kidney cholesterol was higher in Lxrα/β −/− mice, while triacylglycerol tended to be lower, but there were no significant differences between groups (Table 1). Plasma cholesterol levels remained constant in all treatment groups and plasma triacylglycerol levels were lower in Lxrα/β −/− than in Lxrα/β +/+ mice (Table 1).

To determine whether decreased cholesterol export contributed to the increase in glomerular lipids in Lxrα/β −/− mice, we examined the LXR target genes Abca1 and Abcg1 (cholesterol efflux transporters). There was no change in the basal expression of Abcg1, while Lxrα/β −/− mice had decreased mRNA levels of Abca1 (ESM Fig. 2a). Western blot analysis of ABCA1, however, showed no significant difference between groups (ESM Fig. 2c). Likewise, no significant difference was observed between groups in genes important to the cholesterol biosynthetic pathway, e.g. Srebp2 (also known as Srebf2) and Hmgcr (ESM Fig. 2a). To establish whether triacylglycerol synthesis was altered, we measured the LXR target genes Srebp1c (also known as Srebf1) and Scd1. While Srebp1c expression was significantly reduced in vehicle-treated Lxrα/β −/− kidney, this baseline reduction was blunted in STZ-treated Lxrα/β −/− mice, although its relative expression still remained lower than in STZ-treated Lxrα/β +/+ mice. Scd1 also showed a trend towards a baseline decrease in non-diabetic Lxrα/β −/− mice compared with Lxrα/β +/+ mice; this decrease was significantly blunted in response to STZ treatment (ESM Fig. 2a). The expression of fatty acid oxidation genes (Ppara, Acox1, Cpt1a) was unchanged between genotypes (ESM Fig. 2a). Taken together, these data suggest that the accumulation of neutral lipids in the glomeruli of STZ Lxrα/β −/− mice compared with STZ Lxrα/β +/+ mice cannot be explained by decreased cholesterol efflux, decreased fatty acid oxidation, or increased cholesterol or triacylglycerol synthesis.

Lipoprotein uptake in the kidney is primarily mediated through the LDL receptor with some contribution from scavenger receptors [33, 34]. The scavenger receptor encoded by Cd36 was moderately decreased in Lxrα/β −/− compared with Lxrα/β +/+ mice, but not significantly altered by diabetes (ESM Fig. 2b). In contrast, mRNA and protein levels of the oxidised-LDL (ox-LDL) receptor (OLR1) were increased in diabetic Lxrα/β −/− mice (Fig. 2a, b). With a focus on lipid uptake, we found that the expression of Idol, which regulates LDL receptor expression and is an LXR target gene, was significantly decreased in Lxrα/β −/− mice (Fig. 2c) and correlated with increased LDL receptor levels in Lxrα/β −/− mice (Fig. 2d). Electron microscopy of glomeruli from Lxrα/β −/− mice showed increased lipid droplets after STZ treatment, these being localised to the mesangial matrix (Fig. 2e). Together, these data suggest that, although STZ Lxrα/β −/− mice have total circulating cholesterol levels similar to mice in other groups, the extensive accumulation of lipids in their glomeruli may be due to increased uptake and/or increased retention of circulating lipids.

Fig. 2
figure 2

Diabetic Lxrα/β −/− glomeruli have increased lipid accumulation in the mesangium. (a) Kidney gene expression and (b) protein levels of OLR1, an oxidised cholesterol uptake receptor. (c) Kidney expression of Idol (inducible degrader of the LDL receptor [LDLR]) and (d) kidney protein levels of the LDL receptor. Total protein (b, d) was normalised to β-actin. V, vehicle. (e) Electron micrographs of glomeruli at ×800 magnification. Normal glomeruli without lipid droplets were found in Lxrα/β +/+ Veh and Lxrα/β +/+ STZ mice. Increasing numbers of lipid droplets were observed in Lxrα/β −/− Veh and Lxrα/β −/− STZ mice (changes highlighted by arrows). Veh, vehicle. Data (a–d) are means ± SEM; n = 6, except for Lxrα/β −/− (n = 3); *p < 0.05 compared with vehicle control of the respective genotype and † p < 0.05 compared with Lxrα/β +/+ of the same treatment group, tests by one-way ANOVA and Student–Newman–Keuls

Full size image

Diabetic Lxrα/β−/− mice had increased markers of extracellular matrix and profibrotic growth factors compared with diabetic Lxrα/β+/+ mice

To determine whether Lxrα/β −/− mice exhibited features of early DN, including tubulointerstitial fibrosis and mesangial matrix protein accumulation, we analysed markers of each by histological and gene expression techniques. Periodic acid–Schiff's reagent (PAS) stained samples were blinded and analysed by a renal pathologist. Control and diabetic Lxrα/β −/− mice had areas of concentrated PAS staining indicative of mesangial matrix accumulation (Fig. 3a). Diabetic Lxrα/β −/− samples had the most severe pathological changes and were the only group with hyalinotic lesions (Fig. 3b). Semi-quantitative scoring of sections revealed an 18-fold increase in mesangial matrix expansion in the diabetic Lxrα/β −/− samples compared with diabetic Lxrα/β +/+ mice and a sevenfold increase in diabetic compared with non-diabetic Lxrα/β −/− mice (Fig. 3c). Gene expression analyses showed significant increases in markers of renal fibrosis, including genes contributing to the extracellular matrix, e.g. Fn1 and Col1a2 (Fig. 3d, e), as well as those encoding profibrotic growth factors, e.g. Tgfβ (also known as Tgfb1) and Ctgf (Fig. 3f, g). The TGFβ target genes, Col1a2 and Ctgf, were only significantly increased in STZ-treated Lxrα/β −/− mice, remaining unchanged in all other groups. No evidence of interstitial fibrosis was noted by histological analysis and no changes in the expression of α-smooth muscle actin, a marker of collagen-producing myofibroblasts, were observed (data not shown).

Fig. 3
figure 3

Diabetic Lxrα/β −/−mice have increased mesangial matrix expansion compared with diabetic Lxrα/β +/+ mice. (a) PAS staining at ×400 magnification showing normal glomeruli (Lxrα/β +/+ vehicle-treated [Veh] and Lxrα/β +/+ STZ-treated). Lxrα/β −/− vehicle-treated mice had an increase in loose-appearing mesangium and occasional PAS-positive mesangial matrix; Lxrα/β −/− STZ-treated mice had prominently increased matrix. Arrowheads indicate changes. (b) Additional PAS staining of Lxrα/β −/− STZ glomeruli showing strong matrix staining with areas of hyalinotic lesions (arrow). (c) Semi-quantitative analysis of mesangial matrix expansion in ∼100 glomeruli per mouse (n = 3–4). (d) Kidney expression of genes encoding extracellular matrix proteins, i.e. Fn1 and (e) Col1a2, and (f) of genes encoding the profibrotic growth factors TGFβ and connective tissue growth factor (CTGF). Data (c–g) are means ± SEM: n = 6, except for Lxrα/β −/− (n = 3) and quantification of mesangial matrix expansion in (c); *p < 0.05 compared with vehicle (V) control of the respective genotype and † p < 0.05 compared with Lxrα/β +/+ of the same treatment group, tests by one-way ANOVA and Student–Newman–Keuls

Full size image

Diabetic Lxrα/β−/− mice had elevated proinflammatory and oxidative stress markers

LXRs are important determinants of inflammatory signalling. To assess local kidney inflammation, we analysed kidney sections for CD45 leucocyte staining and measured the gene expression of the inflammatory markers Tlr2, Il1b, Icam1 and Opn (also known as Spp1), all of which are known to be dysregulated in DN. The number of CD45 cells in the glomeruli tended to increase in STZ-treated animals compared with controls, as well as basally in Lxrα/β −/− mice compared with Lxrα/β +/+ mice (Fig. 4a, b). Consistent with these data, all inflammatory genes showed their highest expression levels in diabetic Lxrα/β −/− mice, including the macrophage marker Cd68 (Fig. 4c–g).

Fig. 4
figure 4

Diabetic Lxrα/β −/−mice have elevated proinflammatory and oxidative stress markers. (a) Immunostaining of infiltrating leucocytes with CD45 at ×400 magnification. Veh, vehicle; WT, wild-type. (b) Quantification of CD45-positive cells per glomerulus in 25 glomeruli per mouse; n = 3–4. (c) Kidney expression of genes encoding proinflammatory markers that are dysregulated in DN, i.e. Tlr2, (d) Il1b, (e) Icam-1 and (f) Opn, as well as (g) those expressing a macrophage marker (Cd68) and (h, i) oxidative stress markers (Nox2 and p47 phox). (j) Urinary MDA levels normalised to urinary creatinine. Data (b–j) are means ± SEM; n = 6, except for Lxrα/β −/− (n = 3) and CD45 quantification above (b); *p < 0.05 compared with vehicle (V) control of the respective genotype and † p < 0.05 compared with Lxrα/β +/+ of the same treatment group, tests by one-way ANOVA and Student–Newman–Keuls

Full size image

Oxidative stress and inflammation are highly intertwined in pathological states. A key mediator of oxidative stress is the multiprotein NADPH oxidase (NOX) system. NOX2 and its functional partner p47phox are critical for the progression of DN. Consistent with an increased inflammatory response, Nox2 (also known as Cybb) and p47 phox (also known as Ncf1) were elevated in Lxrα/β −/− mice compared with their Lxrα/β +/+ counterparts (Fig. 4h, i). Physiologically, oxidative stress can result in increased lipid peroxidation. As a surrogate marker, we measured malondialdehyde (MDA) and found a significant increase in urinary MDA in STZ Lxrα/β +/+ and Lxrα/β −/− animals compared with non-diabetic controls. Furthermore, non-diabetic Lxrα/β −/− mice had a trend towards increased urinary MDA compared with Lxrα/β +/+ mice (Fig. 4j), suggesting that Lxrα/β −/− mice have an increased basal level of oxidative stress.

Treatment with an LXR agonist attenuated proteinuria and glomerular injury in diabetic DBA/2J mice

To explore whether LXR activation could delay the onset of DN, we treated STZ-injected DBA/2J mice that were on a Western diet with vehicle or an LXR agonist for 10 weeks. We used DMHCA (instead of GW3965 or T091317) to avoid upregulation of hepatic SREBP1c and thus hypertriacylglycerolaemia and hepatic steatosis [25]. As anticipated, diabetic DBA/2J mice had twofold higher urinary ACR compared with non-diabetic mice, this increase being abolished with DMHCA (Fig. 5a). We measured urinary nephrin to assess podocyte integrity. There was a sixfold increase in nephrin excretion in STZ-treated DBA/2J mice compared with non-diabetic mice; this increase was reduced by 50% with DMHCA (Fig. 5b). To ensure that DMHCA was not activating other pharmacological targets, we studied STZ-injected Lxrα/β −/− mice on a Western diet (with or without DMHCA) for 14 weeks. As anticipated, the ACR was unchanged in Lxrα/β −/− mice in response to DMHCA (Fig. 5c). In STZ DBA/2J mice, DMHCA improved the glomerular structure, decreasing mesangial expansion and reducing foam cell formation on PAS-stained sections (Fig. 5d). Although other LXR ligands have been implicated in repression of the renin–angiotensin system (RAS), we found no differences in renin or Ace expression in response to DMHCA (ESM Fig. 3b) or in STZ-treated Lxrα/β −/− mice compared with STZ Lxrα/β +/+ mice (ESM Fig. 2b). Neither DMHCA nor loss of LXRα and LXRβ affected systolic blood pressure (ESM Fig. 4). Animals treated with DMHCA showed no significant change in glucose levels (Table 2), suggesting that the improvement in renal function was independent of hyperglycaemia.

Fig. 5
figure 5

LXR activation with DMHCA attenuates proteinuria and glomerular injury in diabetic DBA/2J mice. (a) Urinary ACR and (b) urinary nephrin:creatinine ratio after 10 weeks of control or STZ treatment with or without DMHCA (D) added to the diet. Veh, vehicle. (a) n = 4–8, except for control DMHCA (n = 3); (b) n = 4–5, except for control DMHCA (n = 3). Sample sizes were smaller for nephrin (b) and for control DMHCA because insufficient spot urine was available from all mice. (c) Urinary ACR in Lxrα/β −/− STZ mice on a diet with or without DMHCA for 14 weeks (n = 7–8). (d) PAS staining of representative kidney sections from DBA/2J mice at ×400 magnification. Arrows point to foam-like cells. Data (a–c) represent means ± SEM; *p < 0.05 for DMHCA vs the respective vehicle group and † p < 0.05 compared with the non-diabetic control group receiving the same treatment, tests by one-way ANOVA, Student–Newman–Keuls and (c) t test

Full size image
Table 2 Physiological variables for non-diabetic and STZ-induced diabetic DBA/2J mice treated with the LXR agonist DMHCA
Full size table

Treatment with an LXR agonist decreased inflammation and lipid accumulation in kidneys of diabetic DBA/2J mice

To assess inflammation in the DBA/2J model of diabetes, we examined inflammatory markers known to be dysregulated in DN. CD68 infiltration was markedly induced with diabetes and then attenuated in diabetic mice treated with DMHCA (Fig. 6a). Consistent with these results, the mRNA expression levels of Tlr2, Icam1, Opn and Cd68 were increased with diabetes and returned to non-diabetic levels upon DMHCA treatment (Fig. 6b–e). Furthermore, in diabetic mice treated with DMHCA, we found decreased Nox2 mRNA (Fig. 6f) and a significant decrease in levels of urinary MDA (Fig. 6g).

Fig. 6
figure 6

LXR activation decreases inflammation in kidneys of diabetic DBA/2J mice. (a) Immunofluorescence staining of kidney sections for synaptopodin (red) and CD68 (green) at ×200 magnification. (b) Gene expression analysis, in the kidney, of proinflammatory markers that are dysregulated in DN, e.g. Tlr2, (c) Icam1 and (d) Opn, as well as of (e) a macrophage marker (Cd68) and (f) an oxidative stress marker (Nox2). (g) Urinary MDA levels normalised to creatinine content. D, DMHCA; Veh, vehicle. (b–g) Data are means ± SEM; n = 4–8; *p < 0.05 for DMHCA vs the respective vehicle group and † p < 0.05 compared with the non-diabetic control group receiving the same treatment, tests by one-way ANOVA and Student–Newman–Keuls

Full size image

We next determined whether changes in neutral lipids correlated with the severity of DN in this animal model. Diabetic DBA/2J mice had increased glomerular Oil Red O staining, which was significantly attenuated with DMHCA (Fig. 7a). We explored pathways important to cholesterol efflux, synthesis and uptake, our aim being to determine whether these could explain changes in renal cholesterol handling. Surprisingly, mRNA levels of Abca1 and Abcg1 were not significantly altered by DMHCA, but protein levels of ABCA1 tended to increase in DMHCA STZ mice (1.3-fold compared with vehicle-treated STZ) (ESM Fig. 3a, Fig. 7b). No significant differences in Hmgcr or Srebp2 mRNA were observed (ESM Fig. 3a), while a trend towards increased Srebp1c and Scd1 (in the kidney) was observed only in non-diabetic animals (ESM Fig. 3a). Expression of Cd36 (mRNA) and Olr1 (mRNA and protein), which are both involved in ox-LDL uptake, were not changed during DMHCA treatment in STZ-treated mice (ESM Fig. 3b, c). To assess whether DMHCA acted directly at the kidney, we measured tissue drug levels by LC/MS/MS. DMHCA was detectable in STZ-treated mice, but not in non-diabetic mice, suggesting that DMHCA does indeed reach the kidney and that renal clearance of DMHCA is impaired with diabetes (Fig. 7c). The expression of Lxrα was decreased with STZ and normalised to non-diabetic levels with DMHCA, with a similar trend being observed for Lxrβ (Fig. 7d, e). Likewise, the LXR target gene Idol was increased with DMHCA in the kidney of STZ-injected mice (Fig. 7f), suggesting that lipoprotein uptake in the kidney may be decreased.

Fig. 7
figure 7

LXR activation decreases glomerular lipid accumulation in kidneys of diabetic DBA/2J mice. (a) Oil Red O staining of kidney sections showing significant accumulation of neutral lipids in glomeruli of STZ mice, but not in those of DMHCA-treated STZ animals (×100 magnification). (b) Kidney protein abundance of ABCA1, a cholesterol efflux transporter. D, DMHCA; Veh, vehicle. (c) Intra-kidney levels of DMHCA as measured by LC/MS/MS. (d) Kidney gene expression of Lxrα, (e) Lxrβ and (f) Idol. (g) Liver cholesterol and (h) triacylglycerol content was assayed after organic extraction and normalised to tissue weight. (i) Hepatic expression of Cyp7a1, which encodes the rate-limiting enzyme involved in the conversion of cholesterol to bile acids. Values determined by quantitative PCR. (j) Hepatic expression of Srebp1c, which encodes a regulator of triacylglycerol synthesis. Values measured by quantitative PCR. Data (b–j) are means ± SEM; n = 4–8; *p < 0.05 for DMHCA vs the respective vehicle group and † p < 0.05 compared with the non-diabetic control group receiving the same treatment, tests by one-way ANOVA and Student–Newman–Keuls

Full size image

DMHCA significantly improved systemic lipid homeostasis

Liver cholesterol content decreased 40% in mice treated with DMHCA compared with vehicle (Fig. 7g). In contrast to other LXR ligands, DMHCA also decreased liver triacylglycerol (Fig. 7h). These changes were supported by liver gene expression, with an increase in Cyp7a1 (which encodes a cholesterol metabolising enzyme) (Fig. 7i) and a decrease in the lipogenic gene Srebp1c (Fig. 7j), consistent with DMHCA being a gene-selective regulator of LXR. Remarkably, we observed a 48% decrease in plasma cholesterol and a 70% decrease in plasma triacylglycerol levels in DMHCA-treated vs vehicle-treated diabetic mice (Table 2). Together these data support a favourable role for LXR activation in mitigating the progression of DN through several mechanisms, including reduction of circulating plasma lipids, inhibition of glomerular lipid accumulation and suppression of inflammatory signalling and oxidative stress.

Discussion

New pharmacological targets are urgently needed to counteract progressive renal decline in patients with DN. In the present study, we demonstrate that LXRs play an essential role in preserving renal structural and functional integrity under normal conditions and when challenged with diabetes. Experiments either in Lxrα/β −/− double knockout mice or using a novel next-generation LXR agonist revealed that LXRs ordinarily function to prevent intraglomerular lipid accumulation through differential uptake and/or retention of lipids. Our data specifically implicate the glomerulus (not tubular epithelial cells) as a primary target for dysfunction, which can be reversed with LXR activation. This is the first report on the in vivo significance of loss of LXR expression for renal filtration and underscores the protective role of LXRs in the prevention of DN.

LXRα and LXRβ are involved in normal renal function with known actions on the RAS and high expression in the kidney [35, 36], including mesangial cells [37], glomerular endothelial cells and proximal tubular cells [38]. Previous studies have shown that activation of LXR with T091317 resulted in inhibition of isoprenaline-induced activation of RAS (without affecting blood pressure) [39], while treatment of angiotensin II-infused mice with GW3965 decreased expression of At1r (also known as Agtrap), Ace and renin, with a concomitant decrease in systolic blood pressure [40]. In our studies, we found that DMHCA treatment did not change renin or Ace expression, and LXR agonism or genetic knockout did not affect systolic BP. Therefore, the association we observed between glomerular lipid accumulation and other indices of diabetic renal injury are unlikely to be a consequence of LXR-mediated RAS-dependent actions, although these analyses need to be repeated with a larger cohort of mice.

Interestingly, our study uncovered a striking renal phenotype in Lxrα/β −/− mice, even without STZ, suggesting a critical homeostatic role for LXRs in kidney function. This was highlighted by the tenfold higher ACR in non-diabetic Lxrα/β −/− mice compared with Lxrα/β +/+ mice. Likewise, glomerular lipid accumulation and mesangial matrix expansion was increased in non-diabetic Lxrα/β −/− mice compared with Lxrα/β +/+ mice. This new model of renal disease may be relevant to human disease, since the expression of LXRα and LXRβ is significantly decreased in patients with DN compared with controls [27] and renal biopsies from patients with DN show lipid deposition primarily in glomeruli [41–43]. Recent studies looking at LXR action in the kidney have shown improved albuminuria with other LXR ligands (T091317, GW3965) and different animal models of DN (STZ C57Bl/6, STZ Ldlr −/−) [28, 29]; however, these studies were unable to address the importance of LXR in baseline renal function. The authors attributed the improvement in renal function to inhibition of the proinflammatory gene osteopontin in renal tubular cells [29] or to enhanced activity of macrophage LXRα [28]. In contrast, our data, supported by loss of function and gain of function (using a gene-selective LXR ligand), suggest a more complex mechanism, which includes protection from lipid deposition in the glomerulus and inhibition of renal inflammation.

The reason for the preferential increase in glomerular lipids in diabetic Lxrα/β −/− is still not clear. Oxidised-LDL has been previously shown to have a five- to tenfold greater affinity for uptake into mesangial cells than LDL [44]. We considered the possibility that ox-LDL may be more abundant in STZ-treated Lxrα/β −/− mice, but found no significant difference in plasma ox-LDL between groups (Table 1). We also measured the uptake of fluorescently labelled ox-LDL in STZ Lxrα/β −/− and Lxrα/β +/+ kidneys, but found no difference at 30 min post infusion (data not shown). These data suggest that ox-LDL is not responsible for the glomerular lipid accumulation. Preferential uptake by the LDL receptor, a key lipoprotein carrier in the kidney [34], remains a plausible explanation. Alternatively, increased proteoglycan deposition as occurs in DN has also been recently shown to increase lipoprotein retention [45].

There are several limitations to our study. Gene expression analyses were performed on whole kidney, but greater changes might have been observed in isolated glomeruli, where the pathological state was primarily located. Our mice exhibited varying sensitivities to STZ, which left the number of animals in some groups lower than anticipated. We did not perform an exhaustive survey of the RAS; however, it is notable that Lxrα/β −/− mice have previously been shown to have a blunted response to isoprenaline, suggesting that Lxrα/β −/− mice would be protected against (not susceptible to) hypertension [46]. DMHCA was very effective at decreasing circulating cholesterol and triacylglycerol levels with additional beneficial effects on liver lipid levels. Therefore, despite the accumulation of DMHCA in the diabetic kidney, we cannot specifically define the contribution of renal vs systemic LXR activation to the improvement of DN. Nonetheless, cultured kidney mesangial cells treated with T0901317 showed increases in Abca1 promoter activity and cholesterol efflux, suggesting that LXR is functional in this cell type and could contribute to the clearance of glomerular lipids [37]. Finally, due to the assessment of renal histology at a single time point, it is not clear whether glomerular lipids accumulate before or after the recruitment of inflammatory cells. Inflammatory cytokines can repress the expression of LXR and its target genes [47], and induce lipid accumulation [48], whereas lipids can induce the expression of inflammatory cytokines [49, 50]. Thus, there is evidence supporting both scenarios. Regardless of the above, the activation of LXR is expected to inhibit both processes, suggesting it may be an ideal pharmacological target for DN.

In summary, our study highlights the importance of cholesterol homeostasis (independently of hyperglycaemia) in the development and severity of DN. In light of our findings, we propose that intraglomerular lipid accumulation will be more reflective of disease progression than plasma lipids. The data support the notion that (1) LXR plays an important basal role in the kidney and (2) pharmacological activation of LXR using a gene-selective agonist that does not cause hypertriacylglycerolaemia could be a novel approach to delaying the progression of DN.