|Home | About | Journals | Submit | Contact Us | Français|
Patients with chronic kidney disease (CKD) have the highest risk for atherosclerotic cardiovascular disease (CVD). Current interventions have been insufficiently effective in lessening excess incidence and mortality from CVD in CKD patients versus other high-risk groups. The mechanisms underlying the heightened risk remain obscure but may relate to differences in CKD-induced atherogenesis, including perturbation of macrophage cholesterol trafficking.
We examined the impact of renal dysfunction on macrophage cholesterol homeostasis in the apoE-/- mouse model of atherosclerosis. Renal impairment induced by uninephrectomy dramatically increased macrophage cholesterol content, linked to striking impairment of macrophage cholesterol efflux. This blunted efflux was associated with downregulation of the cholesterol transporter ATP-binding cassette transporter A1 (ABCA1) and activation of the nuclear factor-kappa B (NF-κB). Treatment with the angiotensin receptor blocker (ARB) losartan decreased NF-κB and restored cholesterol efflux.
Our findings show that mild renal dysfunction perturbs macrophage lipid homeostasis by inhibiting cholesterol efflux, mediated by decreased ABCA1 transporter and activation of NF-κB, and that ARB can restore cholesterol efflux.
Patients with chronic kidney disease (CKD), including those with mild renal dysfunction, have very high risk for cardiovascular disease (CVD) (1-3). Our limited understanding of the underlying mechanisms of CKD-driven vasculopathy constrains development of specific therapeutic approaches. While risk reductions, through current therapeutic interventions, have halved the overall CVD morbidity and mortality in the general population (4), poor CV outcomes continue to escalate in CKD patients (1). Further confounding this issue is variability in the response of CKD patients to therapeutic interventions on lipids. Although lipid-lowering agents prevent CV events in patients with mild to moderate CKD (5-10), their effectiveness appears more equivocal as renal damage progresses to end stage renal disease (5,11,12). These observations suggest that renal dysfunction may exert a unique influence over the process of atherogenesis, and may influence response to standard lipid therapy.
Although renal damage can cause dyslipidemia, the increased incidence of CVD in CKD cannot be attributed to higher serum cholesterol per se (13). Recent studies indicate that multiple local vascular perturbations influence cholesterol trafficking and foam cell formation, a key process in atherogenesis (14-16). There is currently little information about foam cell formation in the setting of renal dysfunction. The formation of macrophage foam cells reflects a failure of cholesterol export mechanisms to keep pace with internalization of cholesterol from lipoproteins and cellular debris (14). Thus, macrophage cholesterol homeostasis is critically dependent on lipid efflux, which involves mobilization of excess cholesterol from intracellular pools to the plasma membrane and transfer to suitable external cholesterol acceptors (14-16). The key pathway for cholesterol movement out of macrophages involves an energy-dependent efflux linked to the ATP-binding cassette transporter A1 (ABCA1) (16,17). However, little is known about the status and regulation of this critical pathway in CKD.
Although we previously showed that renal impairment induced by uninephrectomy (UNx) in apoE-/- mice dramatically amplifies atherogenesis that was responsive to inhibition of angiotensin II actions, the underlying mechanism remained unclear (18). Notably, unlike other atherosclerosis-inducing conditions such as diabetes (19,20), where specific metabolites (i.e., glucose or advanced glycation end products) can influence macrophage lipid homeostasis in vitro, renal impairment does not have such a specific metabolite. We therefore used in vivo as well as ex vivo approaches to study the effects on macrophage lipid homeostasis in this model (11,21). Since angiotensin II (AII) inhibition is a mainstay in treatment of atherosclerotic heart disease (22,23) and CKD is an AII-responsive state (24) we also evaluated the effects of AII inhibition on macrophage lipid homeostasis and regulation of the efflux pathway in this setting.
Female apoE-/- or wild type (WT) mice on C57BL/6 background (Jackson Laboratories) were maintained on normal mouse chow. Animal care and procedures were carried out in accordance with National Institutes of Health and Vanderbilt University animal care facility guidelines. At 9 weeks of age, apoE-/- and WT mice underwent uninephrectomy (UNx) or sham operation performed under isoflurane inhalation anesthesia. After one week, UNx and sham mice were begun on HFD (19,20,25) or continued on the regular chow and assigned to groups as follows: apoE-/- sham with no further treatment (sham on chow, n=10 or sham+HFD, n=5); apoE-/- UNx with no treatment (UNx on chow, n=15 or UNx+HFD, n=6); apoE-/- UNx treated with angiotensin receptor blocker (ARB, losartan, 100 mg/L in drinking water) (UNx on chow+L, n=11 or UNx+HFD+L, n=6) and WT sham on chow (WT sham, n=4) or WT UNx on chow (WT UNx, n=8). All mice were sacrificed at 18 weeks and serum cholesterol and triglyceride levels determined and peritoneal macrophages harvested as described (19,21,25).
At termination of the study, total cholesterol (TC) and free cholesterol (FC) content was assessed in freshly isolated peritoneal macrophages, harvested by peritoneal lavage 3 days after peritoneal injection of 3% thioglycollate. Cells from each mouse were seeded and after 4 hour incubation, baseline cholesterol mass determined by gas-liquid chromatography (GLC) (18,26-28). Esterified cholesterol mass, cholesteryl ester (CE), was taken as the difference between total and free cholesterol. Cellular protein was assayed using the Lowry procedure and bovine serum albumin as a standard. For efflux studies, macrophages from mice in each group were pooled, washed, incubated for 24hr with each of the following: apoAI at 20μg/ml in DMEM, HDL at 150μg/ml in DMEM or DMEM alone. Cholesterol efflux was determined as the difference between cells incubated with DMEM alone and DMEM with one of cholesterol acceptors, apoAI or HDL, and expressed as the percent decrease in cell cholesterol (19,20,27-29).
Uninephrectomy in high-fat fed apoE-/- mice caused a dramatic expansion in macrophage cholesterol, including a 2.5-fold increase in total cholesterol (TC), a 1.8-fold increase in free cholesterol (FC) and a 3.0-fold increase in cholesterol ester (CE) (Table). Treatment with losartan reduced cellular cholesterol by 76%. The cellular cholesterol differences were not directly explained by serum lipid levels (Table). To evaluate whether reduction in renal mass affects cellular cholesterol in the absence of in vivo cholesterol loading, we also assessed the cellular lipid characteristics in macrophages of apoE-/- mice with UNx or sham operation maintained on regular chow diet. ApoE-/- UNx mice on chow also showed a marked increase (3-fold) in macrophage cholesterol, reflecting largely an increase in FC (Table). Again, the cellular cholesterol differences in UNx vs. sham on chow were not explained by serum lipid levels (Table). ARB treatment significantly reduced cellular TC, FC and CE (Table), in the absence of any reduction in serum cholesterol.
Uninephrectomy resulted in blunted macrophage cholesterol efflux, regardless of diet. In the presence of apolipoprotein AI (apoAI) as an extracellular cholesterol acceptor, cellular cholesterol levels in macrophages from apoE-/- UNx+HFD remained elevated compared to cells harvested from apoE-/- sham+HFD (Figure 1A) due to strikingly repressed efflux (5.4±0.9% in UNx+HFD versus 31.6±3.5 in sham+HFD, p<0.05). ARB treatment in these UNx mice partially restored efflux (to 16.1±2.1%, p<0.05 vs. UNx+HFD). Similarly, repressed efflux was seen in macrophage of UNx maintained on chow, (UNx, 7.9±2.7% vs. sham, 17.7±2.9; p<0.05) (Figure 1B). In vivo treatment of apoE-/- UNx mice with losartan also significantly restored efflux, (to 12.0±2.1%, p<0.05 vs. UNx). Similar trends were observed when the macrophages were incubated with HDL as cholesterol acceptor, but the differences did not reach statistical significance (data not shown).
To elucidate mechanisms for the reduced cholesterol efflux in macrophages of UNx mice, we examined the expression of pivotal transporters involved in cholesterol efflux. As noted above, regardless of diet, UNx increased cellular cholesterol level which is expected to increase ABCA1 expression (30). Notably, however, the increase is stunted. Thus, although the level of cellular cholesterol in UNx on chow is similar to that seen in mice with intact kidneys on HFD, the level of ABCA1 expression is strikingly less in macrophages from UNx mice (Table and Figure 3). Greater cellular cholesterol loading with HFD feeding further emphasizes this point as comparison of UNx on HFD had dramatically lower level of ABCA1 expression than shams on HFD. Notably, increased efflux with ARB was linked to increased macrophage ABCA1 levels (Figure 2). By contrast, ABCG1 protein levels were not affected by ARB in apoE-/- sham vs UNx (ABCG1/β actin ratio 0.14±0.02 in sham vs 0.14±0.01 in UNx and 0.26±0.06 in UNx+L, p=NS among the groups).
To further evaluate the mechanisms by which UNx modulates the level of abundance of the transporters, we evaluated mRNA expression. ABCA1 gene expression in apoE-/- UNx on chow was significantly lower when compared with apoE-/- sham-operated mice on chow (by 36%, p<0.05). Examination of degradation of ABCA1 protein revealed numerically more degradation of ABCA1 in UNx compared with ABCA1 in shams (47%±5% in macrophages from UNx versus 30%±9% in macrophages from shams, p=NS). ABCA1 degradation was determined as the difference between ABCA1 protein before and after addition of the protease inhibitor, ALLN (-8%±9% in macrophages from UNx versus 28%±34% in macrophages from shams, p=NS). Notably, even in non-hyperlipidemic wild types mice (WT), ABCA1 mRNA expression was also lower in UNx than in sham-operated WT (by 38%, p<0.05).
The NF-κB pathway is involved in the downregulation of ABCA1 in cholesterol-loaded cells (30,31), and can be activated in renal disease (32). We found significantly increased macrophage nuclear NF-kB activity assessed by electrophoretic mobility-shift assay (EMSA) in apoE-/- UNx versus apoE-/- sham macrophages, with almost two-fold increase in the protein levels of RelA and p50 DNA binding subunits of NF-kB (p<0.05 UNx vs. sham) (Figure 4A and B). Ex vivo exposure of UNx macrophages to an NF-κB antagonist, ammonium pyrrolidinedithiocarbamate (PDTC), significantly increased ABCA1 protein levels (Figure 4C). The up-regulation of both p65 and p50 NF-κB subunits (Figure 4B) was inhibited by treatment with ARB. Taken together, these data suggest that UNx suppresses ABCA1 through activation of NF-κB, and that ARB restores ABCA1 by decreasing activation of NF-κB.
The present study makes the novel observation that in vivo, renal impairment induced by reduction in renal mass markedly increases macrophage lipid content. We further show that the mechanisms for this involve reduction in cholesterol efflux from the macrophages. The underlying molecular mechanism involves decreased macrophage ABCA1 transporter and activated NF-κB. Antagonizing activated macrophage NF-κB with ARB or a specific NF-κB inhibitor reversed these events. These data indicate that the dramatic acceleration in atherosclerosis observed in the setting of renal dysfunction may be related to abnormal macrophage cholesterol homeostasis due, at least in part, to repression of ABCA1 caused by activated NF-κB, which can be ameliorated by a novel effect of ARB treatment.
Since foam cell formation depends on perturbations in macrophage cholesterol homeostasis, we investigated whether renal dysfunction affects macrophage lipid metabolism in this setting. We show that uninephrectomy in apoE-/-mice on high fat Western diet causes significant accumulation of macrophage cholesterol, compared with mice with intact kidneys. The cellular lipid expansion was not simply a reflection of the in vivo plasma lipid environment, as the plasma cholesterol in UNx was only 30% higher than sham while the macrophage cholesterol in UNx was increased by 250% (Table). A similar step-up between plasma and cellular cholesterol was seen in UNx versus shams with intact kidneys maintained on regular chow. These data indicate that foam cell formation and atherosclerosis do not necessarily parallel plasma lipid levels, particularly in the presence of renal dysfunction (5, 33-35).
Our studies further show that in vivo treatment with an ARB, losartan, reduced the cholesterol content in macrophages of UNx mice on either diet. Epidemiological studies have documented fewer cardiovascular events and increased survival with ARB in the general population, and in patients with early renal damage (22, 36). The benefits of ARB in various studies have occurred in the absence of any plasma lipid-lowering effects and have previously been ascribed to modulation of macrophage infiltration, endothelial cell activation, or vascular oxidant stress (18, 37). We now show that ARB directly modulates macrophage cholesterol handling, thereby providing a mechanism for our previous observation (18). In this previous study with the UNx model, we showed that ARB treatment lessened atherosclerosis while decreasing blood pressure, whereas a non-specific vasodilator (hydralazine) reduced blood pressure without affecting atherosclerosis. This suggests that the vascular effects of an ARB are not dependent on systemic hemodynamics (18). Nonetheless, in the current study we cannot exclude possible hemodynamic effects on macrophage cholesterol accumulation.
Patients with renal damage have increased activity of scavenger receptors, which promote increased cellular uptake of cholesterol, however, once upregulated, macrophages do not down-regulate scavenger receptors or inflow of cholesterol by this pathway (38). Cholesterol efflux in this setting is therefore a pivotal step in determining whether intracellular lipid homeostasis is maintained or whether the macrophage will turn into a foam cell (39). Our data show that a modest reduction in renal mass decreases macrophage cholesterol efflux. This effect is linked to repression of the macrophage ABCA1 transporter (Figure 3). UNx also reduced ABCA1 gene expression in normolipemic wild type mice. This effect is also apparent in humans, in that plasma from CKD patients downregulates ABCA1 in cultured endothelial cells (40). Our study makes the novel observation that repression of the ABCA1 transporter underlies the decreased efflux and expanded cholesterol content in macrophages in renal dysfunction. UNx did not significantly change ABCG1 levels despite its established effects on cholesterol efflux. These findings reiterate the knowledge that changes in ABCA1 vs ABCG1 are not necessarily coordinated in a parallel fashion (19, 29, 41). Our studies also show that in vivo treatment with ARB restored macrophage ABCA1 expression in renal dysfunction. While ARB increased macrophage ABCA1 expression, macrophage cholesterol efflux was not completely normalized. These results suggest possible involvement of other mechanisms, including oxidative stress or perturbation in intracellular lipid trafficking. However, in cultured human macrophages exposed to the ARB telmisartan cholesterol efflux was increased via PPARγ-dependent pathway together with upregulation of ABCA1, suggesting a direct ARB effect on macrophages (42). These results complement previous in vivo and in vitro findings that exogenous AII downregulates ABCA1 (29, 42, 43). Taken together, these observations suggest a pivotal importance of the AII-ABCA1 interaction for macrophage cholesterol homeostasis in the setting of renal dysfunction, providing a basis for possible modulation of excess CVD risk in this setting.
We then explored potential mechanisms of interactions between AII and ABCA1. ABCA1 is markedly upregulated by oxysterols via activation of liver X receptor (LXR) (44), and is downregulated by inflammatory stimuli, including lipopolysaccharide and interleukin-1β through the NF-κB signaling pathway (30, 31, 45). The relationship between the NF-κB system, ABCA1, AII and lipid accumulation in the setting of renal damage remains unclear. We show that UNx macrophages have significantly elevated NF-κB activity (Figure 4) together with a corresponding increase in protein expression of NF-κB DNA binding subunits, p65 (RelA) and p50 (Figure 4A and B). We also show that specific antagonism of the NF-κB activation pathway in macrophages lessens the repression of the ABCA1 transporter (Figure 4C). Further, ARB significantly decreased up-regulation of RelA and p50, suggesting this as a potential key regulatory step (Figure 4B) (46).
In summary, our study shows for the first time that loss of the renal parenchyma disrupts macrophage cholesterol homeostasis by repressing ABCA1 through activation of the NF-κB pathway. Further, ARB downregulates NF-κB subunits, upregulates ABCA1 and thus lessens macrophage cholesterol burden. These advantageous effects of ARB may decrease atherogenesis and provide further support for the use of ARB to decrease CV risk in CKD patients.
The authors acknowledge the expert technical assistance of Cathy Xu.
Sources of Funding This work was supported in part by grants from NIH DK44757 and HL087061 (V.K.), DK37868 (I.I.), HL53989 and HL65405 (M.F.L), HL65709 and 57986 (S.F.), and the Lipid, Lipoprotein and Atherosclerosis Core of the Vanderbilt Mouse Metabolic Phenotyping Center (NIH DK59637-01).