Pre-clinical model of severe glutathione peroxidase-3 deficiency and chronic kidney disease results in coronary artery thrombosis and depressed left ventricular function

ABSTRACT

Background

Chronic kidney disease (CKD) patients have deficient levels of glutathione peroxidase-3 (GPx3). We hypothesized that GPx3 deficiency may lead to cardiovascular disease in the presence of chronic kidney disease due to an accumulation of reactive oxygen species and decreased microvascular perfusion of the myocardium.

Methods. To isolate the exclusive effect of GPx3 deficiency in kidney disease–induced cardiac disease, we studied the GPx3 knockout mouse strain (GPx3^−/−) in the setting of surgery-induced CKD.

Results. Ribonucleic acid (RNA) microarray screening of non-stimulated GPx3^−/− heart tissue show increased expression of genes associated with cardiomyopathy including myh7, plac9, serpine1 and cd74 compared with wild-type (WT) controls. GPx3^−/− mice underwent surgically induced renal mass reduction to generate a model of CKD. GPx3^−/− + CKD mice underwent echocardiography 4 weeks after injury. Fractional shortening (FS) was decreased to 32.9 ± 5.8% in GPx3^−/− + CKD compared to 62.0% ± 10.3 in WT + CKD (P < 0.001). Platelet aggregates were increased in the myocardium of GPx3^−/− + CKD. Asymmetric dimethylarginine (ADMA) levels were increased in both GPx3^−/− + CKD and WT+ CKD. ADMA stimulated spontaneous platelet aggregation more quickly in washed platelets from GPx3^−/−. In vitro platelet aggregation was enhanced in samples from GPx3^−/− + CKD. Platelet aggregation in GPx3^−/− + CKD samples was mitigated after in vivo administration of ebselen, a glutathione peroxidase mimetic. FS improved in GPx3^−/− + CKD mice after ebselen treatment.

Conclusion

These results suggest GPx3 deficiency is a substantive contributing factor to the development of kidney disease–induced cardiac disease.

INTRODUCTION

End-stage renal disease (ESRD) patients have an 8.8–10-fold higher incidence of cardiovascular disease compared with the general populations in the USA and Europe [1, 2]. Similar in magnitude to diabetes mellitus, chronic kidney disease (CKD) is recognized as an independent risk factor for cardiovascular disease [3]. The etiology of cardiovascular disease in ESRD patients is unclear, potentially resulting from the activation of several cell-signaling cascades [4–10].

Our previous work focused on the mechanism of the development of left ventricular hypertrophy (LVH) in CKD involving extracellular signal-regulated kinase (ERK) phosphorylation [6, 11]. In this current report we study the potential mechanism by which glutathione peroxidase-3 (GPx3) deficiency causes acute cardiac events in CKD patients. GPx3, previously known as extracellular glutathione peroxidase (Gpx), is a selenoprotein antioxidant enzyme that is synthesized in the basolateral compartment of the kidney [12]. It is transported to the systemic circulation with functions in the plasma, the vessel wall and circulating platelets [13, 14]. GPx3 is the only known GPx subtype with primary activity in the extracellular compartment. Patients with CKD are deficient in this enzyme [15–17]; however, the clinical consequences of this deficiency in CKD are unclear.

We observed that mice deficient in GPx3 due to deletion of the gpx3 gene were more susceptible to myocardial microthrombi, impaired ventricular function and fibrosis in the setting of surgically induced CKD. Extreme GPx3 deficiency was also associated with myocardial platelet activation in the setting of asymmetric dimethylarginine (ADMA) excess. Ebselen, a GPx mimetic, attenuated the pathophysiological effects of GPx3 deficiency, emphasizing the significance of heightened platelet activity in kidney disease-induced cardiac disease. These findings suggest that GPx replacement therapy (or its pharmacological equivalent) has the potential to offset the deleterious effect of GPx3 deficiency in CKD.

MATERIALS AND METHODS

(See supplementary materials for detailed methods.) on ndt online.

Statisical analysis

SPSS (version 22; IBM, Armonk, NY, USA) was used for all statistical analysis.

Genetic mouse model

Genotyping of GPx3^−/− mice was performed as previously reported [13].

Gene expression analysis

Gene expression profiles were constructed using the GeneChip Mouse Genome 430 A 2.0 (Affymetrix, Southern Oaks, CA, USA) high-density olignouncleotide microarray technique. Human ventricular myocardium was obtained from the National Disease Research Interchange and was performed under the supervision of the Partners Healthcare Human Research Committee.

Renal injury model

All research involving the use of mice was performed in strict accordance with protocols approved by the Animal Studies Committee of Harvard Medical School. Surgically induced CKD was performed in two stages as previously described [18].

GPx activity assay

GPx3 activity was measured indirectly using blood plasma samples. Preparation of samples was performed in accordance with the manufacturer’s instructions (Cayman Chemical, Ann Arbor, MI, USA).

Murine echocardiography

Transthoracic echocardiography was performed in mice using a Vevo 2100 high-resolution in vivo micro-imaging system (VisualSonics, Toronto, ON, Canada) and a 30-MHz real-time microvisualization scan head as previously described [11].

Murine hemodynamic monitoring

Cardiac catheterization was performed on anesthetized mice as previously described [11, 19].

Cardiac CD41 quantification

Cardiac tissue was prepared in Optimal Cutting Temperature media. Anti-mouse CD41 antibody (Abcam, Cambridge, MA, USA) was applied (1:200). Using cy3 secondary (1:300), CD41⁺ aggregates were quantified per total cells per high power field (hpf). Aggregates were then averaged among 10 hpfs per animal.

Collagen quantification

Tissue was prepared as per the manufacturer’s recommendations (Chrondrex, Redmond, WA, USA) and Sirius Red staining was performed.

Serum creatinine measurement

Serum creatinine measurements were determined by liquid chromatography–tandem mass spectrometry at the University of Alabama in Birmingham core facility for Acute Kidney Injury Research as previously described [20].

Tissue morphometry

Heart tissue was procured after animals were euthanized. Tissue was dissected, rinsed in cold normal saline, patted dry, then weighed to obtain left ventricular mass.

ADMA serum measurement

ADMA levels were obtained using a mouse/rat ADMA direct ELISA kit (PromoCell, Heidelberg, Germany) according to the manufacturer’s instructions.

L-arginine plasma measurement

L-argine levels were obtained using a competitive arginine ELISA assay kit (Eagle Biosceinces, Nashua, NH, USA). Mouse ethylenediaminetetraacetic acid plasma was prepared according to the manufacturer’s instructions.

Platelet aggregation

After animals were sedated with vaporized isoflurane (1.5% distributed by 1.5 L/min oxygen), blood was drawn by mandibular vein access. Whole blood was anticoagulated using buffer A (65 mM trisodium citrate, 70 mM citric acid, 100 mM dextrose, pH 6.4). Following platelet isolation, platelet aggregation was monitored in a 70-μL volume at 37°C by optical density (OD) after 4 min of agitation.

Ebselen treatment

On postoperative day (POD) 0, animals were treated with ebselen (30 mg/kg) (Sigma Aldrich, St. Louis, MO, USA) by gavage. Doses were given every 3 days, i.e. PODs 0, 3, 6, 9, 12, 15, 18, 21, 24 and 27.

RESULTS

RNA was extracted from GPx3^−/− cardiac tissue and compared with wild-type (WT) controls by microarray analysis. Of 14 000 differentially expressed genes, myh7 was the most highly upregulated gene in GPx3^−/− heart (2.39-fold, P = 0.0029) (Figure 1A, Supplementary Table S1) and has been associated with pathologic hypertrophic cardiomyopathy [21]. Additionally plac9, associated with familial dilated cardiomyopathy (CMD1C) [22]; serpine1, a prothrombotic factor [23]; and cd74, an adenosine monophosphate–activated protein kinase activator in cardiac ischemia [24] were also upregulated in GPx3^−/− heart (Supplementary Figure S1, Supplementary Table S1). We next compared MYH7 expression in the myocardium of 17 patients receiving hemodialysis to that in the myocardium of 13 patients with no kidney disease. MYH7 was increased 11.3-fold in renal disease patients compared with healthy controls (Figure 1B). From these initial results, we inferred that GPx3^−/− animals were susceptible to cardiac dysfunction that may have clinical relevance. GPx3^−/− with normal renal function showed no evidence of cardiac dysfunction (data not shown). We utilized a surgical kidney mass reduction model to stimulate CKD in GPx3^−/− and study cardiac function in a uremic environment with severe GPx3 deficiency.

FIGURE 1.

Gene expression in nonstimulated GPx3^−/−. (A) Volcano plot of relative RNA from cardiac tissue identified with an RNA microarray screening platform comparing nonstimulated wild-type (n = 3) and GPx3^−/− animals (n = 3). (B) Gene expression by quantitative polymerase chain reaction using comparative cycle time (2^−ΔΔCT) normalized to actin for myh7 in mice (n = 4 per group) [nonstimulated WT and GPx3^−/− animals (1.00 ± 0.34 versus 2.86 ± 0.93), respectively; *P = 0.01] and humans [healthy control (n = 13), ESRD (n = 17), 0.63 ± 0.76 versus 7.14 ± 4.11, respectively].

Renal-induced cardiac disease in GPx3-deficient animals

In WT C57/B6 animals, GPx3 localized to the intima of the renal microvasculature and colocalized with endothelial cells (CD31) (Figure 2). GPx3 deficiency was assessed in a C57/B6 mouse line that underwent homologous deletion of the GPx3 allele (GPx3^−/−) [13]. GPx3^−/− and WT mice were subject to a surgical form of severe CKD (see ‘Materials and Methods’ section) [18, 25]. Two weeks after injury, GPx3 activity levels in plasma were reduced by 34.2% in WT littermate controls undergoing surgically induced CKD compared with sham-operated WT mice (Figure 3). Levels in GPx3^−/− mice were 96.3% lower than those measured in sham-operated WT mice (58.4 ± 14.7 nmol/min/mL), overlapped with the assay’s negative control values (50.6 ± 2.6) and were not decreased further by CKD surgery. More severe CKD surgeries were performed to further decrease GPx3 levels in WT animals; however, this resulted in animal death within 5 days (data not shown). Steady-state renal function was observed 4 weeks after surgery with equivalent function 8 weeks after surgery. Similar elevations of serum creatinine were maintained in both GPx3^−/− and WT mice over this time frame (Figure 4A) using a model we previously reported [18]. ADMA, an endogenous inhibitor of nitric oxide (NO) synthase known to accumulate in CKD [26], was also elevated after 4 and 8 weeks of surgically induced CKD compared to L-arginine plasma levels, which was not significantly altered (Figure 4B, C).

FIGURE 2.

Localized GPx3 expression in the kidney. GPx3 is concentrated in the interstitium of the kidney of WT mice and colocalizes with cells that display CD31: (A) low power (scale bar = 100 μm), (B) high power (scale bar = 10 μm).

FIGURE 3.

Plasma GPx3 activity in GPx3^−/− and littermate controls 2 weeks after surgically induced CKD. WT + sham (266.1 ± 44.2) versus WT + CKD (175.2 ± 29.0) (*P < 0.001), WT + CKD versus GPx3^−/− + CKD (56.2 ± 2.6) (**P < 0.001) and WT + sham versus GPx3^−/− + CKD ^(†P < 0.001).

FIGURE 4.

Serum creatinine and ADMA levels 4 and 8 weeks after surgically induced CKD in GPx3^−/− from nonserial blood draws. (A) Serum creatinine (mg/dL) at 4 weeks WT + sham (0.11 ± 0.02) (n = 6) versus WT + CKD (0.21 ± 0.02) (n = 6) (*P < 0.001) and 8 weeks 0.11 ± 0.02 (n = 6) versus 0.20 ± 0.02 (n = 6) (**P < 0.001), respectively; GPx3^−/− + sham (0.13 ± 0.02) (n = 6) versus GPx3^−/− + CKD (0.22 ± 0.02) (n = 6) (^†P < 0.001) and 8 weeks [0.12 ± 0.02 (n = 6) versus 0.21 ± 0.02 (n = 6)] (^‡P < 0.001), respectively. (B) Serum ADMA (μM) at 4 weeks WT + sham (0.47 ± 0.47) (n = 4) versus WT + CKD (5.00 ± 0.80) (n = 4) (*P < 0.001) and GPx3^−/− + sham (0.39 ± 0.39) (n = 6) versus GPx3^−/− + CKD (7.08 ± 0.66) (n = 7) (**P = 0.019) and 8 weeks WT + sham [0.39 ± 0.38 (n = 6)] versus WT + CKD [5.40 ± 4.07 (n = 7)] (^†P = 0.030) and GPx3^−/− + sham [0.59 ± 0.57 (n = 6)] versus GPx3^−/− + CKD [5.34 ± 4.04 (n = 7)] (^‡P = 0.024), respectively. (C) Plasma L-arginine (μM) at 4 weeks WT + sham (44 ± 16) (n = 4) versus WT + CKD (41 ± 30) (n = 5) (P = NS) and GPx3^−/− + sham (41 ± 19) (n = 4) versus GPx3^−/− + CKD (51 ± 30) (n = 4)] (P = NS) and 8 weeks WT + sham [38 ± 13 (n = 4)] versus WT + CKD [23 ± 10 (n = 5)] (P = NS) and GPx3^−/− + sham [35 ± 17 (n = 4)] versus GPx3^−/− + CKD [21 ± 8 (n = 4)] (P = NS), respectively.

Echocardiographic and hemodynamic measures of heart failure

Animals underwent echocardiography 4 weeks after surgery. Left ventricular mass increased in both WT and GPx3^−/− mice undergoing CKD surgery; however, left ventricular fractional shortening (FS) was decreased to 32.9 ± 5.8% in GPx3^−/− + CKD compared to 62.0 ± 10.3% (normal) in WT + CKD mice (P < 0.05) (Figure 5, Table 1). Left ventricular cardiac catheterization was also performed 4 weeks after surgery. Although there was no significant difference in left ventricular end-diastolic pressure (LVEDP), the maximum left ventricular pressure (LVP_max) during systole and mean systemic arterial pressure were significantly decreased in GPx3^−/− + CKD compared with WT + CKD mice (Figure 6). Gram weight of animals at the time of cardiac catheterization was equivalent to that of animals prior to injury, inferring that volume overload did not accompany decreased LVP_max. Eight weeks after surgery, a significantly different proportion of GPx3^−/− + CKD mice died (50.0%) compared with WT + CKD mice (10.0%; Figure 7) (P = 0.047).

FIGURE 5.

Echocardiography after CKD. Representative left ventricular wall motion per unit time (milliseconds) in WT + CKD versus GPx3^−/− + CKD mice 4 weeks after surgically induced CKD.

Table 1.

Morphologic and echocardiographic analysis of GPx3^−/− mice 4 weeks after surgically induced CKD

Morphologic and echocardiographic analysis of GPx3−/− mice 4 weeks after CKD surgery
	WT +sham	GPx3−/− + sham	WT + CKD	GPx3−/− + CKD
	(n = 5)	(n = 5)	(n = 6)	(n = 6)
BW (g)	24.7 ± 2.8	24.8 ± 0.5	25.3 ± 0.9	25.8 ± 0.9
HR (bpm)	593 ± 35	566 ± 26	580 ± 14*	542 ± 24**
LVM (mg)	86.1 ± 0.4	88.2 ± 8.7	99.2 ± 3.4*	96.4 ± 15.1**
FS (%)	68.8 ± 2.4	73.1 ± 12.3	62 ± 10.3	32.9 ± 5.8***
LVM/BW	3.4 ± 0.5	3.6 ± 0.3	4 ± 0.2*	3.7 ± 0.4

BW, body weight; bpm, beats per minute; HR, heart rate; FS, fractional shortening; LVM, left ventricular mass by dry weight.

^*P < 0.05, WT + sham versus WT + CKD, solid line.

^**P < 0.05, GPx3^−/− + sham versus GPx3^−/− + CKD, dotted line.

^***P < 0.05, WT + CKD versus GPx3^−/− + CKD, red line.

FIGURE 6.

Aortic and left ventricular cardiac catheterization in GPx3^−/− 4 weeks after surgically induced CKD. Continuous blood pressure measurements were taken in the aortic root and left ventricle in (A) WT + CKD and (B) GPx3^−/− + CKD mice. (C) Mean arterial pressure (MAP) in WT + CKD (64.85 ± 0.48) (n = 6) versus GPx3^−/− + CKD (55.78 ± 1.89) (n = 6) (P = 0.001) mice, LVP_max in WT and CKD (90.86 ± 1.04) (n = 6) versus GPx3^−/− + CKD (75.72 ± 1.50) (n = 6) (P < 0.001) mice and LVEDP in WT + CKD (10.34 ± 0.58) (n = 6) versus GPx3^−/− + CKD (10.23 ± 0.55) (n = 6) (P = NS) mice.

FIGURE 7.

Survival analysis after surgically induced CKD in GPx3^−/− mice. Groups were compared by log-rank (Mantel–Cox) test. Six of 12 GPx3^−/− animals (50.0%) survived 8 weeks after injury compared to 9 of 10 littermate control animals (90.0%) undergoing injury (P = 0.047).

Platelet dysfunction in the setting of CKD and GPx3 deficiency

To understand better the cause of early mortality in GPx3^−/− + CKD mice, cardiac tissue was evaluated histologically. Using a murine-specific monoclonal CD41 antibody (to integrin alpha-IIb), intracardiac platelet microthrombi were detected 4 weeks after CKD in GPx3^−/− mice at levels significantly greater than controls (P < 0.001) and WT + CKD mice (P < 0.001) (Figure 8A). Collagen type I–V staining was increased after 8 weeks in GPx3^−/− + CKD mice compared with GPx3^−/− + sham mice (Figure 8B). WT + CKD mice also developed collagen deposition, but it was less extensive than that detected in GPx3^−/− + CKD mice.

FIGURE 8.

CD41 and collagen content in GPx3^−/− myocardium after surgically induced CKD. (A) four weeks after CKD, CD41⁺ thrombi per total cell number per hpf in GPx3^−/− + sham (0.03 ± 0.01) (n = 9) versus GPx3^−/− + CKD (0.17 ± 0.05) (n = 9) mice (*P < 0.001) and WT + CKD (0.03 ± 0.01) (n = 9) versus GPx3^−/− + CKD mice (**P < 0.001). (B) Collagen content per total pixel area after 8 weeks: WT + sham (0.19 ± 0.04) (n = 10) mice versus WT + CKD (0.38 ± 0.07) (n = 10) mice (*P < 0.001), GPx3^−/− + sham (0.21 ± 0.05) (n = 10) versus GPx3^−/− + CKD (0.51 ± 0.08) (n = 10) mice (**P < 0.001) and WT + CKD versus GPx3^−/− + CKD mice (^†P = 0.002).

GPx3^−/− animals have previously shown to be susceptible to thrombosis following platelet activation; this effect was shown to be a consequence of impaired metabolism of reactive oxygen species leading to enhanced oxidative inactivation of NO [13, 27, 28]. As the increased concentration of microthrombi in GPx3^−/− mice was observed only after kidney injury and not in sham-treated mice, we concluded that this was due to the presence of an additional prothrombotic determinant in the uremic milieu that triggered platelet activation. As described above, levels of ADMA, an NO synthase inhibitor, were significantly elevated in WT + CKD and GPx3^−/− + CKD mice compared with age-matched sham controls at 4 weeks and 8 weeks after surgery (Figure 4B). Washed platelet studies showed increased platelet aggregation, based on decreased OD, when platelets derived from GPx3^−/− sham controls compared with WT sham-treated mice were exposed to 1 μM ADMA (0.62 ± 0.05 versus 0.70 ± 0.06 OD; P = 0.028) and 10 μM ADMA (0.47 ± 0.04 versus 0.57 ± 0.04 OD; P = 0.004), respectively (Figure 9A). ADMA concentrations used for the in vitro studies were consistent with levels measured in the animal models.

FIGURE 9.

Platelet aggregation in GPx3^−/− in the setting of chronic kidney injury. (A) OD of washed platelets derived from WT + sham versus GPx3^−/− + sham following treatment with 1 μM ADMA (0.70 ± 0.06) (n = 6 animals) versus (0.62 ± 0.05) (n = 6 animals) (*P = 0.028) and 10 μM ADMA (0.57 ± 0.04) (n = 6 animals) versus (0.47 ± 0.04) (n = 6 animals) (**P = 0.004). (B) OD after treatment with vehicle, WT + CKD (1.10 ± 0.04) (n = 6 animals) versus GPx3^−/−+  CKD (0.63 ± 0.05) mice (*P < 0.001), GPx3^−/− + CKD + vehicle versus GPx3^−/− + CKD + ebselen (0.84 ± 0.03) mice (**P < 0.001) after 0.1 μM ADP; WT + CKD (0.89 ± 0.03) (n = 7 animals) versus GPx3^−/− + CKD (0.52 ± 0.03) mice (***P < 0.001), GPx3^−/− + CKD + vehicle versus GPx3^−/− + CKD + ebselen (0.70 ± 0.03) mice (^†P < 0.001) after 10 μM ADP, WT + CKD (0.73 ± 0.05) (n = 7 animals) versus GPx3^−/− + CKD + vehicle (0.48 ± 0.03) mice (^††P < 0.001), GPx3^−/− + CKD + vehicle versus GPx3^−/− + CKD + ebselen (0.67 ± 0.03) mice (^{^}P < 0.001) and after 100 μM ADP WT + CKD (0.57 ± 0.03) (n = 7 animals) versus GPx3^−/− + CKD + vehicle (0.39 ± 0.07) (^†††P < 0.001), GPx3^−/− + CKD + vehicle versus GPx3^−/− + CKD + ebselen (0.65 ± 0.02) (^{^^}P < 0.001).

Treatment with the GPx mimetic ebselen

The rate of platelet aggregation with adenosine diphosphate (ADP) stimulation of P2Y₁₂ receptor was increased for platelets isolated from GPx3^−/− + CKD compared with WT +  CKD mice (Figure 9B). Recognizing that ebselen is a GPx mimetic, GPx3^−/− + CKD mice were next treated with ebselen for 4 weeks. Platelets isolated from GPx3^−/− + CKD + ebselen mice showed decreased aggregation at each ADP dose compared with platelets isolated from vehicle-treated GPx3^−/− + CKD mice. GPx3^−/− + CKD + ebselen mice also showed an increase in left ventricular FS compared with GPx3^−/− + CKD + vehicle mice 4 weeks after injury (55.0 ± 13.0% versus 30.6 ± 9.5%, respectively; P = 0.003) (Table 2). Histologic analysis of hearts (Figure 10) showed a decreased number of CD41 platelet aggregates per total cells per hpf in ebselen-treated GPx3^−/− + CKD mice compared with vehicle-treated GPx3^−/− + CKD mice after 4 weeks (0.171 ± 0.017 versus 0.058 ± 0.005; P < 0.001). Myh7 expression was not decreased in treated animals, however, Col3a1 and SOD3, markers of fibrosis and oxidative stress, respectively, were both downregulated with ebselen treatment. Eight weeks after therapy, absolute collagen content was decreased compared with controls; however, this change was not statistically significant.

Table 2.

Morphologic and echocardiographic analysis of GPx3^-/- mice 4 weeks after surgically induced CKD and 4 weeks of ebselen treatment

	GPx3−/− +CKD + vehicle	GPx3−/− +CKD + ebselen
	(n = 7)	(n = 7)
BW (g)	24.8 ± 2.3	24.2 ± 1
HR (bpm)	549 ± 22	545.6 ± 19.3
LVM (mg)	98.2 ± 7.4	97.4 ± 5.2
FS (%)	30.1 ± 3.8	49.4 ± 9.8*
LVM/BW	4.1 ± 0.2	3.9 ± 0.3

BW, body weight; bpm, beats per minute; HR, heart rate; LVM, left ventricular mass by dry weight (mg).

^*P < 0.05

FIGURE 10.

Cardiac phenotype of GPx3^−/− after surgically induced CKD in animals treated with ebselen. (A) CD41 content (platelet aggregates/total cells) in GPx3^−/− + CKD + vehicle (n = 7) versus GPx3^−/− + CKD + ebselen (n = 6), (0.171 ± 0.017 versus 0.058 ± 0.005; *P < 0.001), respectively. (B) mRNA expression 4 weeks after surgery in GPx3^−/− + CKD + vehicle (n = 5) versus GPx3^−/− + CKD + ebselen (n = 5) mice, myh7 [1.15 ± 0.58 (2^−ΔΔCT) versus 1.86 ± 1.11] (P = NS), col3a1 (1.00 ± 0.25 versus 0.68 ± 0.16) (P = 0.042), SOD3 (1.07 ± 0.32 versus 0.69 ± 0.13) (P = 0.039), respectively. (C) Collagen content per total pixel area 8 weeks after surgery in GPx3^−/− + CKD + vehicle (n = 7) versus GPx3^−/− + CKD + ebselen (n = 7) (0.54 ± 0.09 versus 0.47 ± 0.05; P = NS), respectively.

DISCUSSION

GPx3 deficiency has previously been investigated in hemodialysis-dependent patients and those with CKD, but studies of its association with cardiovascular disease in these patients has thus far been limited [15, 29, 30]. CKD promotes significant oxidative stress [31, 32] that leads to endothelial dysfunction [33], reduced levels of NO [34], vascular disease and hence a parallel tendency toward thrombosis [35]. Furthermore, it is recognized that the frequency of thromboembolic events in patients with CKD is increased compared with age-matched controls, occurring 2.34 times more frequently than in non-CKD patients [36]. In this context, we report an association between GPx3 deficiency and cardiac microthrombosis in the setting of CKD. We demonstrate for the first time the significance of severe GPx3 deficiency in a preclinical GPx3^−/− rodent model of CKD.

GPx proteins are selenoproteins capable of reducing hydrogen peroxide and organic peroxides [37]. This family of enzymes is typified by glutathione peroxidase 1 (GPx1) and GPx3. Widely distributed in the cytosolic compartment of most cells, decreased GPx1 activity has been linked to cardiovascular disease burden [38, 39]. It has been described as a cardiovascular risk factor in CKD; however, the mechanism of systemic deficiency in CKD remains unclear, as GPx1 synthesis is not exclusive to the kidney [40]. In contrast, GPx3 is a selenoprotein primarily synthesized by epithelial cells in the proximal renal tubule, then secreted and distributed to the plasma and circulating platelet and endothelial microenvironment [12, 41–44]. GPx3 is the primary selenoprotein found in the circulation, and its deficiency results in the accumulation of circulating reactive oxygen species, including hydrogen peroxide and organic hydroperoxides [13].

In this study we have overcome the challenge of recapitulating GPx3 levels <175 nmol/min/mL in preclinical CKD models. Until now, such GPx3 deficiency has only been achievable in models of bilateral nephrectomy. In WT mice, aggressive renal ablation techniques are needed to achieve this level of reduction, which at the same time, incurs significant mortality. This limitation was obviated in our studies by utilizing the GPx3^−/− mouse strain, where GPx3 activity levels overlapped with the assay’s negative control (see Figure 3 and the ‘Materials and Methods’ section).

Prior to this report, thrombosis of the coronary microvasculature had not been described as a phenotype of the GPx3^−/− mouse strain. GPx3^−/− animals were first found to be susceptible to thrombotic stroke following cerebral ischemia/reperfusion injury [13]. In our report we showed that microthrombi in the myocardium were increased in GPx3^−/− mice in the context of CKD, a phenotype not found in WT mice with CKD. This implicated the additive effect of a uremic factor exacerbating the prothrombotic state of GPx3 deficiency in GPx3^−/− mice.

In the environment of GPx3 deficiency, we observed an accumulation of plasma ADMA after CKD similar to the levels found in WT mice with CKD. However, we also showed in washed platelet aggregation studies that GPx3^−/− platelets isolated from mice not undergoing CKD were more susceptible to spontaneous aggregation in the presence of ADMA than WT platelets. Further, washed platelets from GPx3^−/− + CKD were already susceptible to aggregation prior to ADP stimulation (Figure 9B). Thus platelet aggregation studies suggest that ADMA (an inhibitor of NO synthase) had a direct effect on platelet aggregation by inhibiting intraplatelet NO synthase and promoting platelet activation in the setting of severe GPx3 deficiency [45, 46]. These findings implicate GPx3 activity <175 nmol/min/mL in combination with ADMA levels ≥5 pg/mL as factors necessary to initiate microthrombosis in vivo.

ADMA is an inhibitor of NO synthase, and patients with declining glomerular and tubular function are known to have elevated plasma ADMA levels [47–49]. The proven clinical relevance of ADMA in patients with CKD influenced our decision to study this NO synthase inhibitor rather than other uremic toxins (such as phenyl acetic acid, indoxyl sulfate, p-cresyl sulfate and homocysteine) in platelet activation studies [50–55]. Reduced concentrations of NO perpetuate platelet activation through a phosphoinositide kinase 3 (PI3K)-dependent pathway. NO prevents the formation of a complex between intracellular PI3K and the src family protein lyn. In a NO-depleted state, the extracellular thrombin receptor can then transmit an intracellular signal for platelet activation through PI3K while at the same time inhibiting guanylyl cyclase–dependent platelet homeostasis [56, 57]. In the composite setting of elevated ADMA and severely depressed GPx3, we identified an increased concentration of microthrombi in the cardiac vasculature that could be explained by impaired NO synthase activity and an increased pool of oxidants, respectively.

In addition to intracardiac microthrombi, we found that GPx3 deficiency was associated with decreased myocardial FS by echocardiography in GPx3^−/− + CKD mice and increased collagen deposition. Microthrombi, which were not present in sham-operated animals or WT animals, suggested a mechanism for systolic dysfunction. GPx3^−/− mice undergoing the sham procedure manifest no excess microthrombi or systolic dysfunction. Similar to the lack of effect of GPx3 deficiency on cerebral blood flow in unstressed animals [13], we did not observe a baseline decrease in left ventricular systolic function in knockout animals or animals undergoing sham surgery.

GPx3^−/− + CKD rapidly developed a decline in FS over 4 weeks along with cardiac fibrosis, hypotension and a high mortality rate, pathological changes consistent with severe systolic heart failure [58–60]. It has been recognized by several groups that CKD-induced cardiac disease can be characterized by cardiac fibrosis in the absence of hypertension [11, 61–63]. Our model recapitulates these findings while adding the additional characteristics of severe systolic dysfunction and microvascular coagulopathy often seen in clinical presentation.

Ebselen proved to be an effective treatment for the model of CKD and severe GPx3 deficiency. Ritz et al. [61] inferred this mechanism several years ago and showed a reduction in medial thickness of cardiac arterioles following antioxidant therapy in a model of uninephrectomy in ApoE^−/− mice. In contrast to these earlier studies, our investigations focused on the intravascular effect of ebselen on platelet activity. We showed that washed platelets that were collected from GPx3^−/− + CKD + ebselen mice were less activated following ADP stimulation at multiple concentrations than vehicle-treated controls. In vivo, the cardiac pathology observed in GPx3^−/− + CKD mice was partially reversed by ebselen with an improvement in FS. Cardiac tissue from these animals revealed a decrease in col3a1 and SOD3 gene expression, consistent with a system under decreased oxidative stress [64, 65]. Histologically, the same tissue showed fewer platelet aggregates in the ebselen-treated group.

These findings suggest that ebselen overcomes the effects of severe GPx3 deficiency, attenuating the initiation of platelet activation in the context of elevated plasma ADMA levels. In contrast, these findings also suggest that persistently low GPx3 levels promote NO consumption while ADMA inhibits NO synthase, promoting endothelial dysfunction [13], and thereby enhancing platelet susceptibility to agonist-induced activation. The combined effect of GPx3 gene disruption and ADMA excess in our preclinical model shows a remarkable display of cardiac pathology that cannot otherwise be generated in a single-mammalian model system and often takes several years to develop in the clinical setting of progressive renal failure.

The limitation of our study centers on the persistence of LVH following ebselen administration. Reduced thrombotic burden in the cardiac microvasculature by administration of a GPx mimetic was anticipated, given the previous reports that GPx3 deficiency promoted injury in the context of acute cerebral ischemia [13]; however, it is unclear how GPx3 deficiency may contribute to fibrosis and LVH. We previously found that LVH due to CKD can be treated with mammalian target of rapamycin (mTOR) blockade and inhibition of ERK phosphorylation. Overall, the mechanism for the development of normotensive LVH remains unknown. Our current working model thus depicts two pathologic processes that separately involve microthrombosis due to GPx3 deficiency and cardiac hypertrophy due to mTOR pathway activation. New models of kidney disease–induced cardiac disease may benefit from incorporating both processes in phenotypic analyses.

In summary, patients with CKD have deficient levels of circulating GPx3, which may contribute to cardiovascular disease risk due to the accumulation of reactive oxygen species that exacerbate inflammatory signaling and platelet activation in the setting of ADMA-impaired NO generation in the microvasculature of the myocardium (Figure 11). This prothrombotic state can be counteracted by oral GPx mimetic supplementation. Such a condition warrants future clinical studies that prospectively study the incidence of cardiac events in CKD patients with relative GPx3 deficiency.

FIGURE 11.

GPx3 converts hydrogen or lipid peroxides to their corresponding alcohols (i.e. water and lipid alcohols), inhibiting the cascade of platelet activation. ADMA inhibits NO synthase, which inhibits synthesis of NO (dotted line) from L-arginine. Platelet activation is greatest in an ADMA-rich, GPx3-deficient state consistent with the plasma environment of patients with severe CKD. In this environment, platelet inhibitory NO is decreased by reduced synthesis via ADMA inhibition of NO synthase and by increased oxidative consumption via reaction with peroxyl derivatives of hydrogen and lipid peroxides.

SUPPLEMENTARY DATA

Supplementary data are available at ndt online.