Abstract
In chronic kidney disease, vascular calcification, renal osteodystrophy, and phosphate contribute substantially to cardiovascular risk and are components of CKD–mineral and bone disorder (CKD-MBD). The cause of this syndrome is unknown. Additionally, no therapy addresses cardiovascular risk in CKD. In its inception, CKD-MBD is characterized by osteodystrophy, vascular calcification, and stimulation of osteocyte secretion. We tested the hypothesis that increased production of circulating factors by diseased kidneys causes the CKD-MBD in diabetic mice subjected to renal injury to induce stage 2 CKD (CKD-2 mice). Compared with non-CKD diabetic controls, CKD-2 mice showed increased renal production of Wnt inhibitor family members and higher levels of circulating Dickkopf-1 (Dkk1), sclerostin, and secreted klotho. Neutralization of Dkk1 in CKD-2 mice by administration of a monoclonal antibody after renal injury stimulated bone formation rates, corrected the osteodystrophy, and prevented CKD-stimulated vascular calcification. Mechanistically, neutralization of Dkk1 suppressed aortic expression of the osteoblastic transcription factor Runx2, increased expression of vascular smooth muscle protein 22-α, and restored aortic expression of klotho. Neutralization of Dkk1 did not affect the elevated plasma levels of osteocytic fibroblast growth factor 23 but decreased the elevated levels of sclerostin. Phosphate binder therapy restored plasma fibroblast growth factor 23 levels but had no effect on vascular calcification or osteodystrophy. The combination of the Dkk1 antibody and phosphate binder therapy completely treated the CKD-MBD. These results show that circulating Wnt inhibitors are involved in the pathogenesis of CKD-MBD and that the combination of Dkk1 neutralization and phosphate binding may have therapeutic potential for this disorder.
CKD is a pandemic affecting 26 million Americans in its early stages.1 CKD is associated with high rates of cardiovascular mortality, making it much more likely that affected individuals will sustain cardiovascular morbidity, including death, than reach end stage kidney disease that requires RRT.2,3 The established cardiovascular risk factors do not account for the heightened risk associated with CKD. Novel risk factors markedly intensified by CKD, vascular calcification,4 phosphorus,5 and fibroblast growth factor 23 (FGF23)6 have been identified and await definitive prospective outcome studies to fully validate their risk factor status. Hyperphosphatemia stimulates vascular osteoblastic transition and vascular calcification,7,8 linking the two cardiovascular risk factors together in CKD. A new cardiovascular risk factor discovered in CKD, FGF23, seems strongly associated with left ventricular hypertrophy,6,9 and it is regulated by hyperphosphatemia. Thus, all of the CKD-associated specific cardiovascular risk is linked in the CKD–mineral bone disorder (CKD-MBD) syndrome. The CKD-MBD is a new syndrome named in 2006 in recognition that the skeletal (renal osteodystrophy) and mineral (disordered Ca and Pi homeostasis) disorders, elevated FGF23, hyperparathyroidism, and calcitriol deficiency caused by CKD are critical contributors to the high mortality rates attached to kidney disease.10
We and others have shown that the CKD-MBD begins early in the course of kidney disease and is clinically detectable in stage 2 CKD.11–15 We have characterized the CKD-MBD in early CKD as stimulation of vascular calcification, production of renal osteodystrophy, and stimulation of osteocyte secretion of FGF23, sclerostin, and other factors.11 Studies have also shown that reductions in the fraction of filtered phosphorus being reabsorbed were evident before changes in serum Ca, Pi, parathyroid hormone (PTH), or calcitriol.16,17 We had shown in translational studies that increased FGF23, reduction in bone formation rates, and vascular calcification began in early CKD in our model when GFR was reduced equivalent to stage 2 CKD.12 Pereira et al.13 showed that FGF23 and dentin matrix protein1 expression in skeletal osteocytes was stimulated in stage 2 CKD along with increased circulating FGF23 before Ca, Pi, PTH, or calcitriol levels were abnormal. Increased osteocyte expression of sclerostin was also shown along with decreased portmanteau of Wingless and Integration1 (Wnt) signaling.14,15,18 Thus, the early CKD-MBD is characterized as increased levels of bone turnover inhibitory signals, elevated production of osteocytic proteins FGF23 and sclerostin, the onset of renal osteodystrophy, dedifferentiation in the vascular smooth muscle, and the stimulation of vascular calcification.11,13,15,19 This new characterization of the early CKD-MBD when Ca, Pi, PTH, and calcitriol levels are normal raises the question of its pathogenesis.
Here, we test the hypothesis that the pathogenesis of the CKD-MBD is the release of factors upregulated during kidney disease into the circulation, including the Wnt inhibitor family, which are known humoral substances.20,21 We show increased renal production of Dickkopf1 (Dkk1), sclerostin, and sclerostin domain containing 1 (also known as Wise and uterine sensitization antigen 1) and increased circulating Dkk1 in a model of CKD produced by incomplete recovery from AKI in type 2 diabetes and the onset of the CKD-MBD. We then show that Dkk1 neutralization is sufficient to prevent vascular dedifferentiation, vascular calcification, and renal osteodystrophy and decrease circulating sclerostin levels. Dkk1 neutralization did not affect FGF23 levels, which were normalized by a phosphate binder added to the diet. In a treatment protocol, the combination of Dkk1 antibody treatment and phosphate binding completely reversed the CKD-MBD. Dkk1 neutralization did not affect GFR, BUN level, or renal pathology, indicating the absence of critical functions of Dkk1 in the diseased remnant kidney.
Results
We have reported staging the severity of renal injury in our model analogous to human CKD.11 In the experimental design used here (Figure 1), the mice with CKD had reductions in GFRs measured by inulin clearance that are equal to those reductions in patients with stage 2 CKD (CKD-2). The serum Pi and Ca of the CKD-2 mice were normal, whereas the BUN and PTH levels were insignificantly elevated (Table 1). The chronic renal injury-induced activation of the Wnt pathway led to increased expression of Wnt inhibitors, including Dkk1 (Figure 2). Tissue levels of Dkk1 were increased in our CKD-2 mice at 15 weeks and returned to the elevated levels equal to sham-operated, high fat-fed LDL receptor-deficient (ldlr−/−) mice at 22 weeks. Circulating Dkk1 levels were massively increased in CKD-2 mice at 15 weeks and remained 4-fold increased at 22 weeks. Circulating Dkk1 levels were not increased in the sham-operated mice, despite the increased Dkk1 expression compared with wild type, suggesting that decreased renal catabolism contributed to the increased circulating levels in the CKD-2 mice (Supplemental Figure 1). To examine the effects of the elevations of Dkk1 on production of the CKD-MBD, we undertook their neutralization using a monoclonal antibody to Dkk1 administered intraperitoneally (30 mg/kg) two times per week (Dkk1 mAb) from 15 to 22 weeks.
Experimental design. (A) Schematic drawing of the experimental design defining the various animal groups. Wild-type C57B6J mice (WT) were used to establish the normal ranges of the parameters being studied. CKD-2 was induced in ldlr−/− mice fed high-fat diets (d) at 10 weeks of life by electrocautery (EC) injury of one kidney at 12 weeks and contralateral nephrectomy (Nx) at 14 weeks. The mice received vehicle injections from 14 weeks to euthanasia at 22 weeks. The Dkk1 mAb-treated mice (CKD-2/Dkk1 mAb) were ldlr−/− high fat-fed mice undergoing the same protocol, except for receiving the Dkk1 mAb (30 mg/kg intraperitoneally twice weekly) instead of vehicle. The control mice for the effects of early CKD were ldlr−/− mice in the C57B6J background fed a high-fat diet undergoing sham operations (SOs). (B) GFRs by inulin clearance in Sham, CKD-2, and CKD-2/Dkk1 mAb-treated mice. Inulin clearances and BUN levels (Table 1) were used to establish that the mild ablative CKD was equivalent in GFR reduction to human stage 2 CKD. Inulin clearances were at 20 weeks of age. The reduction in GFR was equivalent to the reduction in eGFR considered as stage 2 CKD clinically. Dkk1 mAb treatment did not affect GFR.
Serum chemistries in the various groups of mice
Kidney homogenate and plasma Dkk1 levels. (A) Dkk1 levels in kidney homogenates were increased at 15 weeks in CKD-2 mice. By 22 weeks, kidney Dkk1 levels in the CKD-2 and CKD-2+Dkk1 mAb mice were similar to sham mice. The increased Dkk1 expression in sham-operated mice compared with wild-type mice is related to diabetes. IB, immunoblot. (B) Plasma Dkk1 levels were massively increased in the AKI phase at 15 weeks and remained 4-fold elevated at 22 weeks. Compared with wild-type mice, the sham mice did not have increased Dkk1 in the plasma, despite the increased protein levels in renal homogenates. The Dkk1 mAb affected the ELISA assay for Dkk1, preventing measurement of plasma levels in the Dkk1 mAb-treated group.
Effects of Dkk1 mAb on the Skeleton
The ldlr−/− high fat-fed mouse is associated with a low-turnover osteodystrophy shown in our sham-operated mice (Figure 3). The sham-operated mice had decreased femoral trabecular osteoblast and osteoclast numbers and decreased bone formation rates compared with wild-type C57B6J mice. Compared with the sham-operated mice, CKD-2 did not add significantly to the depressed osteoblast and osteoclast numbers, but it suppressed osteoid volume and was additive to the decreased osteoblast and osteoclast numbers and bone formation rates (Figure 3, A and B). Treatment with the antibody to Dkk1 increased bone volume and osteoblast and osteoclast numbers, and it significantly stimulated bone formation rates. Analysis of the femoral metaphysis by microcomputed tomography (μCT) showed that the Dkk1 antibody increased trabecular bone volume and trabecular numbers and decreased trabecular separation (Figure 3C). The effects of CKD-2 on our animal model were most apparent in cortical bone, where they produced a significant reduction in cortical bone area and mineral density (increased porosity) as shown by three-dimensional reconstruction of the distal femurs of sham, CKD-2, and Dkk1 antibody-treated mice (Figure 3D, Supplemental Figure 2).
Effect of the Dkk1 neutralizing antibody on skeletal histomorphometry and μCT imaging of femoral metaphyseal trabeculae and midshaft femur cortices. (A) Histomorphometry: Dkk1 antibody treatment increased trabecular bone volume (BV/TV), osteoblast number (ObN/BL), and osteoclast number (OcN/BL) compared with CKD-2 vehicle-treated mice (OV/BV). (B) Dkk1 antibody treatment increased bone formation rates (BFRs). (C) μCT of femoral metaphyseal trabecular bone revealed increased bone volume (BV/TV), trabecular number (Tb.N), and decreased trabecular separation (Tb.Sp) in the Dkk1 mAb-treated mice without changing trabecular thickness (Tb.Th). There were no significant differences between sham-operated and CKD-2 mice. wild-type (WT), n=11; sham, n=9; CKD-2, n=11; Dkk1 Ab, n=12. *P<0.05; **P<0.005. (D) Three-dimensional reconstruction of the distal femoral cortical bone showing loss of bone volume in CKD-2 mice prevented by treatment with the Dkk1 mAb. Data are shown as mean±SEM.
Effects of Dkk1 mAb on the Vasculature
Early CKD produces vascular calcification that was apparent in our CKD-2 mice (Figure 4A). Treatment with the antibody to Dkk1 prevented the stimulation of aortic calcium levels by CKD. We examined established mechanisms of CKD-stimulated vascular calcification to explain the effect of Dkk1 antibody treatment. CKD-2 causes vascular smooth muscle dedifferentiation, which is shown by loss of message and protein levels for smooth muscle cell contractile proteins (i.e., sm22α) (Figure 4B), and increased aortic message levels for runt-related transcription factor 2 (Runx2), a biomarker of osteoblastic transition, and its transcriptional target alkaline phosphatase (ALP) (Figure 4B). The Dkk1 mAb caused a marked increase in sm22α expression and inhibited Runx2 and ALP expression. In addition, FGF23 expression in the aorta, which is stimulated in our sham mice and decreased by CKD-2, was stimulated by the Dkk1 mAb treatment (Figure 4B). Aortic protein levels of the genes studied revealed increased smooth muscle cell actin, another smooth muscle cell contractile protein, and sm22α (not shown), loss of Runx2, and increased FGF23 levels (Figure 4C).
Aortic calcification and osteoblastic transition in the various groups of mice. (A) Aortic calcium was increased in the CKD-2 mice compared with sham and wild-type mice. The increase in aortic Ca was prevented by treatment with the Dkk1 antibody. (B) Relative mRNA levels of smooth muscle 22α (SM22α), ALP, Runx2, and FGF23. In the two bars on the left of the graphs, the effect of CKD-2 on the message levels is compared with levels in aortas of sham-operated mice, which were set to equal to one. Then, in the bars on the right, the aortic message levels in mice treated with the Dkk1 antibody are compared with levels in vehicle-treated CKD-2 mice, which were set to equal to one. (C) Immunoblots of aortic α-smooth muscle actin, Runx2, and FGF23; α-tubulin levels served as loading controls. IB, immunoblot. *P<0.05.
A recently discovered mechanism of vascular calcification in CKD, CKD-stimulated depression of vascular klotho22,23 (Figure 5), was reversed at the message and protein levels in the aorta by Dkk1 antibody treatment. By immunohistochemistry, CKD-2 eliminated klotho expression in the vascular media, and the Dkk1 antibody treatment produced significant gain of klotho expression in the media (Figure 5B). Klotho is mainly expressed in the distal tubule of the nephron, the parathyroid gland, and the brain. In the distal tubule, the single-pass transmembrane protein is cleaved, releasing the extracellular domain, cklotho, into the circulation, where it functions as a hormone regulating osteocyte FGF23 secretion.24 We have shown that cklotho levels are increased in our CKD-2 mice,11 and here, we show that the Dkk1 antibody treatment failed to further affect cklotho levels (Figure 5C). We also show that the CKD-2 of our mice is a state wherein there are fewer distal tubules, but the remnant tubules express klotho similar to the sham-operated control mice (Supplemental Figure 3).
Effect of Dkk1 neutralizing antibody on aortic and circulating klotho levels. (A) Aortic mRNA and immunoblots for αklotho. (B) Aortic immunohisotochemistry for αklotho. (C) Plasma cklotho levels in the various groups of mice at euthanasia. IB, immunoblot; WT, wild-type.
Effects of Dkk1 mAb on Osteocyte-Secreted Proteins
Secretion of the hormone, FGF23, and sclerostin by osteocytes is stimulated in early CKD in humans (see below).13,15 In our CKD-2 mice, plasma FGF23 levels were significantly elevated, and Dkk1 neutralization did not affect FGF23 levels (Figure 6A). However, the Dkk1 antibody treatment did affect the levels of another osteocytic protein in the circulation, sclerostin, compared with CKD-2 vehicle-treated mice (Figure 6B).
Effect of Dkk1 neutralizing antibody on osteocyte-secreted proteins FGF23 and sclerostin (Scl). (A) Plasma FGF23 levels in the various groups of mice. (B) The change in Scl levels between weeks 15 and 22 of life in plasma of CKD-2 vehicle-treated mice and CKD-2 mice treated with Dkk1 mAb.
Effects of Lanthanum Carbonate on Urinary Phosphate, FGF23 Levels, Vascular Calcification, and Renal Osteodystrophy
The lack of effect of the Dkk1 mAb on plasma FGF23 levels might be expected if FGF23 secretion in early CKD (before any change in serum phosphate to hyperphosphatemia) was affected by changes in the renal hormone cklotho and if cklotho levels were dependent on tubular fluid phosphate regulation of a disintegrin and metalloproteinase domain (ADAM)-10 and -17 activity. Therefore, we examined the effects of phosphate binding to decrease urinary phosphate and affect FGF23 secretion using lanthanum carbonate (LaCO3) mixed in the high-fat diet of our CKD-2 mice. The experimental plan for these studies is shown in Figure 7A. The plan was modified to examine the actions of the Dkk1 antibody in a treatment strategy rather than the prevention strategy that was used in the studies presented above by starting treatment after the vascular calcification and renal osteodystrophy were established. In addition, the LaCO3 group and a LaCO3 plus the Dkk1 antibody group were added to the modified plan. Figure 7B shows the inulin clearances of the groups, establishing that they were CKD-2 in terms of the CKD. LaCO3 decreased the elevated FGF23 levels of the CKD-2 mice (Figure 7C). LaCO3 decreased the urine phosphate to creatinine levels of the CKD-2 mice (Figure 7D) and decreased the serum phosphate from 9.7 and 9.9 mg/dl in the other CKD groups to 8.8±1.0 mg/dl (Table 2). As shown in Figure 8, vascular calcification was not affected by LaCO3 therapy alone, whereas the combination of LaCO3 with the Dkk1 antibody decreased vascular calcification similar to Dkk1 mAb treatment alone. LaCO3 did not affect bone formation in the osteodystrophy of the CKD-2 mice (data not shown), and it had no effect on the stimulation of osteoblast and osteoclast numbers and bone formation by the Dkk1 mAb. Thus, at the level of vascular calcification and renal osteodystrophy, there was no additivity of LaCO3 to treatment with the Dkk1 antibody.
Combination Dkk1 antibody and phosphate binder therapy. (A) Schematic drawing of the modified experimental design defining the various animal groups and a switch from a prevention to a treatment strategy. Sham, ldlr−/− mice on a high-fat diet undergoing SOs are the control mice for the effects of CKD; CKD-2–28, ldlr−/− high fat-fed mice undergoing mild renal injury (EC) and contralateral NX with vehicle treatment beginning at 22 weeks and euthanasia at 28 weeks; CKD-2–28-Dkk1 mAb, same as CKD-2–28 but with Dkk1 mAb (30 mg/kg intraperitoneally twice weekly) beginning at 22 weeks; CKD-2–28-LaCO3, same as CKD-2–28 but 3% LaCO3 added to food beginning at 22 weeks; CKD-2–28-Dkk1 mAb+LaCO3, same as CKD-2–28 but both the Dkk1 mAb and LaCO3 added at 22 weeks. (B) Inulin clearances in the mice used in the modified experimental design. (C) Effect of LaCO3 on plasma FGF23 levels. (D) Urine phosphate/creatinine ratios at euthanasia in the various groups of mice. Urinary phosphorus/creatinine was decreased by LaCO3 treatment. EC, electrocautery; SOs, sham operations; WT, wild-type. *P<0.05 compared to CKD-2 (22 weeks).
Serum chemistries in the modified study design
Aortic calcium levels. Aortic calcium levels in the various groups of mice in the modified experimental design. WT, wild-type.
Plasma DKK1, Sclerostin, and FGF23 Levels in Human Normophosphatemic CKD
To relate our findings above to early CKD in humans, we analyzed plasma DKK1, sclerostin and FGF23 levels in 38 stable patients with normophosphatemic CKD stage 3a of various etiologies (40% diabetes) attending an outpatient nephrology practice. The creatinine levels, 24-hour urine-based creatinine clearances, BUN, Ca, Pi, and PTH levels are shown in Supplemental Table 1. Referent plasma levels of the measured factors were obtained by simultaneous determination of plasma samples from the Diabetes Heart Study.25 Plasma DKK1 levels were significantly elevated, and FGF23 and sclerostin levels were also elevated (Figure 9), indicating that early CKD is a state of elevated levels of circulating Wnt inhibitors, including DKK1 and sclerostin.
Plasma FGF23, sclerostin, and Dkk1 levels in normophosphatemic patients with CKD-3a. Plasma levels were determined at baseline before treatment in 38 patients enrolled in a pilot and feasibility study of phosphate binding in normophosphatemic patients with early CKD. The levels in the CKD-3a cohort were compared with those levels in 700 patients with normal kidney function in the Diabetes Heart Study (DHS) determined simultaneously in the same assays. *P<0.01 compared with the DHS study cohort.
Discussion
This paper reports the discovery that reactivation of the Wnt pathway in early kidney disease produces elevations of circulating Wnt inhibitors, which cause the vascular dedifferentiation, vascular calcification, and renal osteodystrophy of the CKD-MBD. In early CKD, the CKD-MBD is characterized as stimulation of vascular calcification, an osteodystrophy associated with inhibitory signals to bone formation rates, stimulation of osteocytic proteins in the circulation consisting of the hormone FGF23 and sclerostin, and normal serum phosphorus, Ca, and PTH.11,13,15,26 Neutralization of a critical renal Wnt inhibitor increased in the circulation by CKD, Dkk1, was sufficient to prevent and treat the vascular calcification (the osteodystrophy) and preserve vascular differentiation. In addition, early changes in phosphate homeostasis were linked to elevated FGF23 levels, thereby completing the pathogenesis of the CKD-MBD.
The Wnt pathway is critical in the regulation of the uncommitted state of tissue stem cells and kidney development during nephrogenesis.27–29 Wnt9b, Wnt7b, and Wnt4 are indispensable in the inductive signals leading to the epithelia that form the nephron,30,31 and they are reactivated during renal repair and in renal fibrosis.32–34 In renal repair, Wnt4 stimulation contributes to cellular proliferation and epithelial reconstitution.35 Wnt stimulation of stem cell division produces daughter cells closest to the stimulus that remain as stem cells and a cell farther removed from the signal available for differentiation, allowing a planar tubular epithelium to develop.36 Wnt-stimulated target genes include negative feedback regulators (Wnt inhibitors such as Dkk1) that are secreted proteins that participate in cell differentiation and branching morphogenesis of the developing nephron.37 Reactivation of the Wnt pathway during renal injury and repair33,38 induces the Wnt inhibitor–secreted proteins that enter the circulation as shown here. Other than the induction of Dkk1 and the Wnt inhibitor family after unilateral ureteral obstruction that we have previously reported,33 here, we show that Dkk1, sclerostin, and uterine sensitization-associated antigen (also known as sclerostin domain containing 1) are upregulated in the tubular epithelium distal to the proximal tubule and circulate in the plasma. Neutralization of elevated Dkk1 by an mAb prevented vascular calcification, preserved a differentiated vascular smooth muscle phenotype, and stimulated bone formation.
In the examination of the potential mechanisms of action of the Dkk1 antibody in the vasculature, three established mechanisms of vascular calcification were found to be reversed by antibody treatment. First, the Dkk1 antibody increased vascular klotho levels. Reductions in vascular klotho caused by CKD have recently been reported as causes of vascular calcification, although high levels of klotho expression are limited and do not include vascular tissues.22,23 However, the function of klotho in the endothelium may provide clues to the mechanism of Dkk1 neutralization in protection against vascular calcification.39,40 Second, the Dkk1 antibody prevented the aortic expression of Runx2, a critical osteoblastic transcription factor associated with vascular osteoblastic transition and vascular calcification.41,42 Third, the antibody prevented the increase in aortic ALP observed in the CKD-2 mice. ALP has been shown to decrease levels of vascular pyrophosphate, which is a critical inhibitor of vascular calcification in CKD,43 and ALP is a transcriptional target of Runx2. Thus, the Dkk1 antibody treatment prevented osteoblastic transition in the aorta of our CKD-2 mice. These findings are surprising, because the Wnt pathway has been implicated in the pathogenesis of vascular calcification and osteoblastic transition in the vascular smooth muscle cells in the animal model represented by our sham-operated animals.44,45 However, analysis of vascular β-catenin expression in our studies revealed very low levels in aortic smooth muscle and higher levels in the endothelium (Supplemental Figure 4). Because of the quantitative insensitivity of immunohistochemistry for β-catenin, we analyzed the expression of axin2. Axin2 is rapidly induced during Wnt signaling and used as a quantitative measure of Wnt signaling.46 As shown in Supplemental Figure 5, Wnt signaling was mainly detected in the kidneys of mice that had early diabetic nephropathy. Induction of CKD-2 increased axin2 levels, and treatment with the Dkk1 antibody did not further affect renal axin2 levels. These findings show that the Wnt inhibitors have significant redundancy during the reactivation of the Wnt pathway in our CKD model and that neutralization of one, Dkk1, is not critical and did not increase Wnt signaling (Supplemental Material). Aortic axin2 levels were much lower than in the kidney, and Dkk1 neutralization increased aortic axin2, showing the stimulation of the pathway by the antibody used. Our results suggest that kidney disease-induced Wnt inhibitors cause a dramatic klotho-dependent shift in endothelial function in the vasculature and that the high levels of circulating Wnt inhibitors are causative of vascular calcification in CKD through a Wnt- and klotho-inhibited endothelial pathway.
The studies reported here were performed in a well characterized translational model of CKD in type 2 diabetes and CKD-stimulated atherosclerotic vascular calcification.7,47–50 We show that our sham-operated animals had decreased bone formation rates caused by decreased osteoblast numbers stemming from the increased adipogenesis of type 2 diabetes.51 Superimposition of CKD-2 produced decreased trabecular bone osteoid and cortical bone area and porosity. Dkk1 neutralization overcame both the decreased osteoblastogenesis and the effects of CKD by stimulation of osteoblast numbers and bone formation rates. Skeletal Wnt inhibition and increased sclerostin are found in early CKD,15 although remodeling rates may not be reduced in early CKD. The frequencies of decreased bone formation rates in early CKD stages 2–4 are controversial, ranging from 0% to 50%,52–54 and may be age-related. In animal models of early CKD caused by polycystic kidney disease and Alport’s disease, normal or elevated modeling rates were observed as osteodystrophy was detected, despite findings of decreased nuclear β-catenin in osteocytes and elevated sclerostin levels.15,55,56 Wnt and PTH are regulators of the skeletal stem cell niche,18,57,58 and Wnt inhibition produces PTH-mediated adaptation, which maintained remodeling rates in CKD. Here, Dkk1 neutralization prevented the development of renal osteodystrophy. One possible mechanism, in addition to the direct effects of decreasing Dkk1 inhibition of skeletal Wnt activity, is that Dkk1 neutralization affected osteocytic production of sclerostin, another Wnt inhibitor that is increased in early CKD, which is shown here in Figure 9 and the work by Sabbagh et al.15 Sclerostin is a powerful inhibitor of bone formation.59,60 Our data shown in Figure 6B, which show that Dkk1 neutralization decreases circulating sclerostin levels, are compatible with the demonstration that it protects from systemic bone loss during inflammation in the work by Heiland et al.,61 which also showed that it reduced sclerostin expression. Interestingly, the binding sites of Dkk1 and sclerostin on LDL receptor–related protein 5 and 6 and Kremen differ, perhaps explaining why the actions of these Wnt inhibitors may differ and not be totally redundant.62
Dkk1 neutralization failed to affect plasma FGF23, a critical component of the early CKD-MBD.11 FGF23 is a new hormone regulating phosphate homeostasis discovered as the basis for autosomal dominant hypophosphatemic rickets,63 and it is found to be elevated in CKD.64 Subsequent studies have shown that FGF23 levels are a powerful biomarker of cardiovascular mortality in CKD,6 possibly related to direct actions of high levels on cardiac myocytes.65 Recent studies have shown that elevations of FGF23 are the earliest biomarker of developing CKD-MBD.13,66 Other than Dkk1 neutralization correcting vascular calcification and renal osteodystrophy without affecting FGF23, correction of elevated FGF23 by phosphate binding had no effect on vascular calcification and renal osteodystrophy. The pathogenesis of increased osteocyte secretion of FGF23 in CKD before hyperphosphatemia is incompletely resolved. One hypothesis has been that skeletal inhibition produced by kidney disease activates FGF23 secretion. This hypothesis seems proven wrong by the studies reported here. Dkk1 neutralization increased bone formation and corrected the osteodystrophy of our model, but it did not affect urinary phosphorus, serum phosphorus, or FGF23 levels. However, reducing urinary phosphorus concentration with a phosphate binder, as shown here in Figure 7, normalized FGF23 levels. These data are compatible with the recent clinical studies combining dietary restriction and phosphate binding in prospectively decreasing FGF23 levels in early CKD.67 The connection between renal tubular phosphate, which is the stimulus to the early increase in FGF23 secretion, and the osteocyte may have been discovered with the hormonal function of circulating klotho.24 In the studies reported here, addition of LaCO3, a Pi binder, to the diet of mice being treated with the Dkk1 neutralizing antibody normalized FGF23 in addition to the effects of Dkk1, producing complete elimination of the CKD-MBD as defined in our CKD-2 mice.
In summary, CKD-2 stimulates vascular dedifferentiation and calcification, causes renal osteodystrophy, and increases circulating FGF23 and sclerostin levels. CKD-2 increased Dkk1 levels in the kidney and the circulation, and Dkk1 neutralization was sufficient to prevent vascular calcification and renal osteodystrophy. These effects are strongly indicative of a pathogenic role of Wnt inhibition in the vascular calcification and the inhibition of the skeleton producing the osteodystrophy of CKD. The combination of Dkk1 neutralization and reduction of urinary phosphate by a phosphate binder was sufficient in a treatment protocol to decrease vascular calcification, correct the osteodystrophy, and decrease FGF23 and sclerostin levels to baseline. Thus, the CKD-MBD of the CKD-2 ldlr−/− high fat-fed mice was reversed by the Dkk1 mAb treatment combined with phosphate binder therapy. What remains to be shown now, with the feasibility of CKD without the CKD-MBD, are the effects on improving cardiovascular outcomes.
Concise Methods
Production of Animal Models
The animal protocols were approved by the Washington University animal studies committee. The atherosclerotic ldlr−/− mice fed a high-fat diet (42% calories from fat; Teklad) beginning at 10 weeks of age were bred or purchased from The Jackson Laboratory. The mice are obese, insulin-resistant at 22 weeks of age, diabetic at 28 weeks of age, and hypercholesterolemic.
A two-step procedure was used to create CKD as described previously.47,48 Electrocautery was applied to the right kidney through a 2-cm flank incision at 12 weeks postnatal followed by left total nephrectomy at 14 weeks of age. The intensity of the cautery was varied to produce mild (CKD-2) renal injury that was confirmed by inulin clearances at age 20 weeks. Control animals received sham operations, in which the appropriate kidney was exposed and mobilized but not treated in any other way. After the surgical procedures, 14-week-old mice were randomized into groups for two sets of studies. The first study was a prevention protocol with four groups of mice (Figure 1A). The first group was wild-type mice fed a regular diet, with normal renal function and diet. The second group was ldlr−/− mice that were fed a high-fat diet and sham-operated. This group had normal renal function. This group served as the control group to determine the effect of high-fat diet in the face of normal renal function. The third group was ldlr−/− mice with GFR reduced equivalent to human CKD stage 2 fed a high-fat diet with euthanasia at 22 weeks (the baseline control group; 22 weeks). The fourth group was ldlr−/− mice with CKD-2 receiving intraperitoneal injections of an mAb to Dkk1 (Amgen) two times per week beginning at 15 weeks until euthanasia at 22 weeks. The dose used was previously shown in pharmacokinetic/pharmacodynamic studies to be the optimal dose for inhibition of inflammation-induced bone loss,61 and it was the dose used by Diarra et al.21 in their study of inflammation induced skeletal inhibition.
In the second study (Figure 7A), the protocol was modified into a treatment protocol by instituting treatment after the CKD-MBD including vascular calcification was established. The first group in this second study set was ldlr−/− mice that were fed a high-fat diet and sham-operated with euthanasia at 28 weeks of age. This group had diabetes, incipient nephropathy, and normal GFR. This group served as the control group for the effect of CKD. The second group was ldlr−/− mice with GFR reduced equivalent to human CKD stage 2 fed a high-fat diet and receiving vehicle from 22 to 28 weeks (CKD-2–28), with euthanasia at 28 weeks. The third group was ldlr−/− mice with CKD-2 receiving intraperitoneal injections of an mAb to Dkk1 (Amgen) three times per week beginning at 22 weeks until euthanasia at 28 weeks (CKD-2–28-Dkk1 mAb). The fourth group was ldlr−/− mice with GFR reduced equivalent to human CKD stage 2 fed a high-fat diet, treated with LaCO3 (3% wt/wt) mixed in the diet beginning at 22 weeks and euthanized at 28 weeks (CKD-2–28-LaCO3). The fifth group was ldlr−/− mice with CKD-2 receiving the Dkk1 mAb injections from 22 to 28 weeks plus having the LaCO3 (3% mixed in the diet) and euthanasia at 28 weeks. Euthanasia was under anesthesia. Intraperitoneal anesthesia (13 mg/kg xylazine and 87 mg/kg ketamine) was used for all procedures. Saphenous vein blood samples were taken 1 week after the second surgery to assess baseline postsurgical renal function. At the time of euthanization, blood was taken by intracardiac stab, and the heart and aorta were dissected en bloc.
Inulin Clearances
Inulin clearances at 20 or 26 weeks, if euthanasia was at 28 weeks, were performed according to the manufacturer’s instructions.
Chemical Calcification Quantitation
Aortas and hearts were dissected after euthanization, and all extraneous tissue was removed by blunt dissection under a dissecting microscope. Tissues were desiccated for 20–24 hours at 60°C, weighed, and crushed to a powder with a pestle and mortar. Calcium was eluted in 1N HCl for 24 hours at 4°C. Calcium content of eluate was assayed using a cresolphthalein complexone method (Sigma-Aldrich, St. Louis, MO) according to the manufacturer’s instructions, and results were corrected for dry tissue weight.
Blood Tests
Serum was analyzed on the day of blood draw for BUN, calcium, and phosphate by standard autoanalyzer laboratory methods performed by our animal facility. Serum Dkk1 levels were analyzed using a commercial mouse Dkk1 ELISA kit (catalog number DY1765; R&D Systems, Inc., Minneapolis, MN). PTH levels in plasma were measured using a PTH (mouse intact) ELISA kit (ALPCO Diagnostics, Salem, NH). This two-site ELISA was performed according to the manufacturer’s recommendations. The absorbance was read at 450 nm using a microplate reader (VERSA max; Molecular Devices). FGF23 levels were measured by the Kainos intact hormone assay (Kainos, Tokyo, Japan). The circulating cklotho levels were measured by an ELISA (MBS702609; MyBioSource.COM, San Diego, CA). The sclerostin levels were analyzed using a sandwich enzyme immunoassay kit (ABIN426039; Antibodies Online, Inc., Atlanta, GA) for mouse plasma. For the ELISA assays, blood was drawn by saphenous vein or cardiac puncture at the time of euthanasia. All blood samples were placed on ice at collection. Platelet-poor EDTA plasma samples were made by a two-step centrifugation at 6000 rpm for 5 minutes and 14,000 rpm for 2 minutes (both at 4°C). Samples were stored frozen at −20°C or below until being used.
μCT
The femur of each mouse was removed after euthanasia and placed in 70% alcohol; after several days, the bone was subject to scan and analysis using μCT (model μCT 40; SCANCO). The distal femoral metaphysic proximal to the growth plate was selected, and a 624-μm volume (40 slices at 16 um/slice) of bone was analyzed. The data of three- and two-dimensional analyses in trabecular and cortical bone were documented.
Bone Histology and Histomorphometry
Bone formation rate/bone surface was determined at the time of euthanasia. All mice received intraperitoneal tetracycline (5 mg/kg) 7 and 2 days before being euthanized. Both femurs were dissected at the time of euthanasia and placed in 70% ethanol. All bone samples were dehydrated and embedded in methylmethacrylate as previously described.68 Serial sections of 4- and 7-μm thickness were cut with a Microm microtome (model HM360; Carl Zeiss, Thornwood, NY). Sections were stained with the modified Masson–Goldner trichrome stain.69 Unstained sections were prepared for phase contrast and fluorescent light microscopy. Parameters of bone structure, formation, and resorption were evaluated at standardized sites in cancellous bone using the semiautomatic method (Osteoplan II; Kontron, Munich, Germany) at ×200 magnification.70,71 The histomorphometric parameters comply with the guidelines of the nomenclature committee of the American Society of Bone and Mineral Research.72 Osteoplan II software has been programmed to transfer data automatically to statistical software (SPSS for Windows; Chicago, IL).
Histology and Immunohistochemistry
Aortic tissues were fixed in 10% neutral buffered formalin overnight, transferred to 70% ethanol at 4°C, and embedded in paraffin; 5-μm sections were prepared. Alizarin red staining was used to detect calcification according to a standard protocol.73 For immunohistochemical staining, all slides were deparaffinized in xylene, dehydrated in a graded ethanol series, and then, rehydrated. Endogenous peroxidase was blocked with 3% H2O2 in methanol for 15 minutes. Nonspecific binding was blocked with avidin and biotin blocking agents (Vector Laboratories, Burlingame, CA) for 30 minutes and then, 5% normal donkey serum for 10 minutes. After washing with PBS, the slides were incubated with primary antibody at 4°C overnight followed by incubation with biotinylated secondary antibody (Vector Laboratories) at room temperature for 1 hour. Then, the strepatividin-conjugated perioxidase staining was performed using the DAB Kit (SK-4100; Vector Laboratories) and the ImmPACT AEC Kit (SK-4205; Vector Laboratories). The species-specific total Ig (Vector Laboratories), monoclonal anti-mouse FGF-23 antibody (MAB26291), polyclonal anti-mouse klotho antibody (AF1819), and monoclonal anti-human/mouse RUNX2/CBFA1 antibody (MAB2006; R&D Systems, Inc.) were used in this study.
RT-PCR
RNA was extracted from aortas and cell cultures using RNeasy Mini Kits (Qiagen, Valencia, CA); 1 μg total RNA was reverse-transcribed using the iScript cDNA Synthesis Kit from Bio-Rad (Hercules, CA) according to the manufacturer’s instructions. Primers were designed using Vector NTI software (Invitrogen, Grand Island, NY), and optimal conditions for each prime pair were determined. A Perkin–Elmer DNA Thermal Cycler was used to perform the reaction. After reverse transcription was performed as above, real time was performed using the MX 4000 (Strategene, La Jolla, CA), SYBR Green from Sigma-Aldrich (St. Louis), and the PCR kit from Invitrogen. Each reaction was performed in triplicate at 95°C for 45 seconds, 60°C for 30 seconds, and 72°C for 60 seconds for 40 cycles. This process was followed by a melt cycle, which consisted of stepwise increase in temperature from 72°C to 99°C. A single predominant peak was observed in the dissociation curve of each gene, supporting the specificity of the PCR product. Ct numbers (threshold values) were set within the exponential phase of PCR and used to calculate the expression levels of the genes of interest. Glyceraldehyde 3-phosphate dehydrogenase was used as an internal standard and used to normalize the values. A standard curve consisting of the cT versus log cDNA dilutions (corresponding to the log copy numbers) was generated by amplifying serial dilutions of cDNA corresponding to an unknown amount of amplicon. Negative controls were performed by inactivating the reverse transcription by boiling for 5 minutes before RT-PCR to insure that genomic DNA was not amplified.
Human Studies
The study protocol was approved by the Human Research Protection Office at Washington University in St. Louis, Missouri. Subjects were enrolled after giving informed consent in accordance with guidelines from the Declaration of Helsinki. Subjects were eligible if more than 18 years of age with stage 3 CKD (eGFR=30–59 ml/min per 1.73 m2) using the Modification of Diet in Renal Disease study equation.74 Exclusion criteria included pregnancy, bone disease, myocardial infarction, congestive heart failure, diastolic dysfunction, and severe hypertension. Members of the research team identified their clinic patients who satisfied the inclusion criteria and were interested in the study. Those patients who provided informed consent to our research nurse allowed a review of their medical records to determine final eligibility before enrollment. As subjects were enrolled, they were stratified for age, sex, race, and diabetes status and then randomized into two groups (allocated 1:1 to receive either 1000 mg LaCO3 or a matching placebo with meals three times per day for 12 months). The randomization and double-blind strategy were designed and maintained by our research pharmacist and statistician. Because practitioners often do not prescribe dietary phosphate restriction in normophosphatemic CKD, dietary phosphate intake was not regulated. All participants were encouraged to track phosphate sources in the diet, but these data were not collected. The purpose of the study was to determine the cohort sizes needed to adequately power primary and secondary outcomes in definitive prospective trials. The baseline values of serum chemistries and CKD-MBD factors are reported here.
Blood samples were obtained from each subject at the baseline visit and after 12 months of treatment. There was intersubject variation in the time of day that blood samples were obtained. Aliquots of plasma were generated from each sample and used immediately for biochemical testing or frozen at −80°C. Plasma levels of FGF23, DKK1, and sclerostin were measured in duplicate using commercially available ELISA kits according to the manufacturer’s instructions (FGF23; Kainos Laboratories, Tokyo, Japan; DKK1; R&D Systems, Inc.; sclerostin; Teco Medical Group, Sissach, Switzerland).
Statistical Analyses
Statistical analysis was performed using ANOVA. Differences between groups were assessed post hoc using Fisher’s least significant difference method and considered significant at P<0.05. Data are presented as mean±SEM. Analyses were performed using Sigma Stat statistical software (Point Richmond, CA). Data for all groups represent an n of 7–10. Unpaired t test was used to compare CKD-2 mice treated with Dkk1 antibody versus vehicle-treated CKD-2 mice or sham-operated mice.
Disclosures
H.M. and K.A.H. have served on advisory boards for Shire and Fresenius, and K.A.H. reports investigator-stimulated trial support from Amgen, Shire, and Fresenius. The other authors report no disclosures.
Acknowledgments
These studies were supported by National Institutes of Health Grants KL2-TR000450 (to M.S.), UL1-TR000448 (to Washington University), DK071891 (to B.I.F.), DK080770 (to H.M.), DK070790 (to K.A.H.), DK089137 (to K.A.H.), and M01-RR07122 (to the Wake Forest School of Medicine General Clinical Research Center). Investigator-stimulated trial support was obtained from Shire and Amgen. Additional support was from the Kentucky Nephrology Research Trust (to H.M.).
Footnotes
Published online ahead of print. Publication date available at www.jasn.org.
This article contains supplemental material online at http://jasn.asnjournals.org/lookup/suppl/doi:10.1681/ASN.2013080818/-/DCSupplemental.
- Copyright © 2014 by the American Society of Nephrology