Asymmetric Dimethylarginine Contributes to the Impaired Response to Erythropoietin in CKD-Anemia

Abstract

Erythropoietin-resistant anemia is associated with adverse cardiovascular events in patients with ESRD, but the underlying mechanism remains unclear. Here, we evaluated the role of the endogenous nitric oxide synthase inhibitor asymmetric dimethylarginine (ADMA). In 54 patients with advanced CKD, erythrocyte but not plasma ADMA levels independently associated with low hemoglobin values, although levels of both types of ADMA were elevated compared with those in healthy volunteers. Furthermore, erythrocyte ADMA level associated with the erythropoietin resistance index in patients receiving a weekly injected dose of erythropoiesis-stimulating agents standardized for hemoglobin levels and body weight, whereas it correlated with the erythropoietin demand index (plasma erythropoietin units divided by the hemoglobin value) in patients not receiving erythropoiesis-stimulating agents. Compared with sham-operated controls, wild-type mice with 5/6 subtotal nephrectomy (Nx), a remnant kidney model with advanced CKD, had decreased hemoglobin, hematocrit, and mean corpuscular volume values but increased erythrocyte and plasma ADMA and plasma erythropoietin levels. In comparison, dimethylarginine dimethlaminohydrolase-1 transgenic (DDAH-1 Tg) mice, which efficiently metabolized ADMA, had significant improvements in all of the values except those for erythropoietin after 5/6 Nx. Additionally, wild-type Nx mice, but not DDAH-1 Tg Nx mice, had reduced splenic gene expression of erythropoietin receptor and erythroferrone, which regulates iron metabolism in response to erythropoietin. This study suggests that erythrocyte ADMA accumulation contributes to impaired response to erythropoietin in predialysis patients and advanced CKD mice via suppression of erythropoietin receptor expression.

Anemia is a common complication in patients with CKD and an independent risk factor for cardiovascular events and death.^1,2 The most well known cause of anemia in CKD is insufficient synthesis of erythropoietin (Epo) owing to reduced renal mass and function. Accordingly, erythropoiesis-stimulating agents (ESA) containing recombinant human erythropoietin proteins have been widely used for the management of anemia in patients with CKD in an effort to compensate for their relative erythropoietin deficiency.³ Nevertheless, it is reported that up to 10% of patients undergoing ESA treatment have exhibited poor erythropoietic response and require a large amount of the agents.⁴ Anemia with hyporesponse to ESA therapy is termed Epo-resistant anemia. Previous studies have demonstrated that Epo-resistant anemia is associated with increased risk of ESRD, cardiovascular events, and death.^5,6 However, the mechanism underlying Epo-resistant anemia remains unknown. One possible cause might be chronic inflammation.⁷ Some researchers have reported the association between Epo-resistant anemia and increased levels of inflammation markers, such as C-reactive protein, IL-6, IFN-γ, and TNF-α in patients with CKD.^8,9 However, Solomon et al. noted that increased inflammatory reactions were not enough to explain the exaggerated risk for Epo-resistant anemia in patients with CKD.⁶

Nitric oxide (NO), a gaseous signal mediator, has various physiologic functions including vasodilation,^10,11 anti-inflammatory,¹² and antithrombogenic reactions.^13,14 Recently, NO has been shown to contribute to erythropoiesis as well.^15–18 Under hypoxic conditions, NO derived from neuronal and endothelial cells stimulates expression of the erythropoietin receptor (EpoR) in these cell types.^15,16 In addition, L-arginine, which is a substrate of nitric oxide synthase (NOS), is required for erythroid cell differentiation.^17,18 Consistent with these findings, severe anemia is observed in the endothelial NOS–knockout mice.¹⁹

Asymmetric dimethylarginine (ADMA), an endogenous NOS inhibitor, has been found to accumulate in the erythrocytes of critically ill patients in surgical intensive care units, approximately 32% of whom had renal fialure.²⁰ ADMA reduces NO production not only via competitive inhibition of NOS but also by suppression of intracellular L-arginine uptake through a cationic amino acid transporter.^21,22 ADMA is generated from post-translational methylation of arginine residues by protein arginine methyltransferases and subsequent protein breakdown. Its elimination is mainly accomplished through hydrolysis into citrulline and dimethylamine by dimethylarginine dimethylaminohydrolase–1 (DDAH-1).²³ It has also been demonstrated that an ADMA-metabolizing system was present in the circulating erythrocytes.^24,25 However, its pathologic role in impaired response to Epo remains unclear.

Therefore, in this study, we hypothesized that increased erythrocyte ADMA levels were associated with the development of Epo-resistant anemia by suppressing the NO-dependent EpoR expression in erythroid progenitor cells. To address the issues, we first examined whether erythrocyte ADMA levels were correlated with impaired response to Epo in a cross-sectional study of 54 predialysis patients. Then, we investigated whether impaired response to erythropoietin was attenuated in DDAH-1–overexpressing transgenic (DDAH-1 Tg) mice with 5/6 subtotal nephrectomy (Nx), a remnant kidney model with advanced CKD.

Results

Characteristics of Predialysis Patients

A total of 54 patients who were predialysis and admitted into Kurume University Hospital were enrolled in this study. Characteristics of the patients are shown in Table 1. Overall, 38.9% of the patients were women and mean age was 66.2 years. Of all patients, 27 (50.0%) and 22 (40.7%) had diabetes mellitus and dyslipidemia, respectively. Most patients (96.3%) had hypertension, of whom 50 (92.5%) took antihypertensive medications, including renin-angiotensin system inhibitors and calcium channel blockers. Nine patients (16.7%) were taking oral iron. Further, no patients received parenteral iron therapy. The median eGFR was 13.7 ml/min per 1.73 m² and 64.5% had an eGFR<15 ml/min per 1.73 m². The mean hemoglobin concentration was 10.1 g/dl. As shown in Figure 1, A and B, compared with healthy volunteers, plasma and erythrocyte ADMA levels were significantly elevated in predialysis patients (plasma ADMA levels: 0.59±0.23 μM in healthy volunteers versus 1.03±0.28 μM in patients with CKD, P<0.01; erythrocyte ADMA levels: 6.00±1.42 nmol/g protein in healthy volunteers versus 6.92±2.18 nmol/g protein in patients with CKD, P<0.05).

Table 1.

Univariate and stepwise multiple regression analyses of hemoglobin levels in predialysis patients

Variable	Patients (n=54)	Univariate Regression Analysis of Hb Levels		Stepwise Multiple Regression Analysis of Hb Levels
		r	P Value	Adjusted β (95% CI; Lower to Upper)	P Value
Men, n (%)	33 (61.1)
Age, yr	66.2±13.4	−0.19a	0.17
Diabetes mellitus, n (%)	27 (50)
Hypertension, n (%)	52 (96.3)
Dyslipidemia, n (%)	22 (40.7)
Smoking history, n (%)	21 (38.9)
Medical therapy, n (%)
RAS inhibitors	44 (81.5)
CCBs	47 (87.0)
RAS inhibitors + CCBs	50 (92.6)
Oral iron supplement	9 (16.7)
ESA	26 (48.1)b
BMI, kg/m2	23.8±4.0	0.04a	0.79
Systolic BP, mmHg	147.5±26.5	0.04a	0.78
Diastolic BP, mmHg	77.9±19.0	0.33a,c	0.02c	0.314 (0.007 to 0.050)c	0.01c
Hb, g/dl	10.1±1.7	——	——
Total protein, g/dl	6.3±0.8	0.27a	0.05	0.334 (0.197 to 1.167)c	0.01c
Albumin, g/dl	3.2±0.7	0.21a	0.13
Total cholesterol, mg/dl	178.0±45.2	0.33a,c	0.01c	0.280 (0.002 to 0.020)c	0.02c
BUN, mg/dl	55.6±22.9	−0.17a	0.21
Serum Creatinine, mg/dl	4.99 (0.96–10.5)	−0.18d	0.19
eGFR, ml/min per 1.73 m2	13.7 (3.8–43.5)	0.28c,d	0.04c
30–59, n (%)	5 (9.3)
15–29, n (%)	14 (25.9)
<15, n (%)	35 (64.8)
LDH, IU/l	223.2±58.0	0.04a	0.76
MOF, %	0.37±0.04	−0.15a	0.37
CRP, mg/dl	0.31 (0.01–1.55)e	−0.03d	0.85
Ferritin, ng/ml	135.7 (11.0–590.0)	−0.01d	0.96
TSAT, %	27.0±10.6	0.12a	0.40
NT-proBNP, pg/ml	4558±8984 (54.9–56479)	–0.30c,d	0.03c
Plasma ADMA, μM	1.03±0.28	0.03a	0.81
Erythrocyte ADMA, nmol/g protein	6.92±2.18	–0.41a,c	0.002c	–0.299 (–0.419 to –0.055)c	0.01c

Variables are expressed as mean±SD, or median (min–max) as appropriate. Stepwise multiple regression analysis was performed to input age and sex in addition to significantly correlate variables by univariate correlation after logarithmic transformation of NT-proBNP levels. Adjusted r² for this model is 0.36. Hb, hemoglobin; 95% CI, 95% confidence interval; RAS, renin-angiotensin system; RAS inhibitor, angiotensin-converting enzyme inhibitor or/and angiotensin II receptor blocker; CCB, calcium channel blocker; BMI, body mass index; LDH, lactate dehydrogenase; CRP, C-reactive protein; MOF, mean of erythrocyte fragility; TSAT, transferrin saturation.

^aPearson correlation coefficient.

^bEight patients received ESA therapy during <1 mo.

^cStatistically significant values (P<0.05).

^dSpearman rank correlation coefficient.

^eValues of 17 patients were below the detection limit (P<0.01 mg/dl).

Figure 1.

Plasma and erythrocyte ADMA levels were significantly elevated in predialysis patients. (A) Plasma and (B) erythrocyte ADMA levels were elevated in predialysis patients. ADMA levels in plasma and erythrocyte lysate were measured by LC/MS as described in the Concise Methods (healthy volunteers: n=31; predialysis patients: n=54). Columns expressed as mean±SD (parametric). Statistical significance was determined using an unpaired t test. *P<0.01, **P<0.05.

Erythrocyte ADMA Levels as One of the Determinants of Anemia in Predialysis Patients

Univariate linear regression analyses were performed on laboratory measures to determine their correlation factors with hemoglobin levels (Table 1). Hemoglobin levels were associated with diastolic BP (r=0.33, P=0.02), nutritional state (total protein levels: r=0.27, P=0.05; total cholesterol levels: r=0.33, P=0.01), renal function (eGFR: r=0.28, P=0.04), and N-terminal pro-B-type natriuretic peptide (NT-proBNP), which is a surrogate marker for heart failure (r=−0.30, P=0.03). Erythrocyte ADMA levels were also significantly correlated with hemoglobin levels (r=−0.41, P=0.002), whereas plasma ADMA levels showed no association with hemoglobin levels. Hemolytic status and erythrocyte fragility also had no correlation with hemoglobin levels (lactate dehydrogenase: r=0.04, P=0.76; mean of erythrocyte fragility: r=−0.15, P=0.37).

A stepwise linear regression analysis was then used to examine the independent determinants of hemoglobin levels in predialysis patients. Diastolic BP (adjusted β=0.31, P=0.01), total protein levels (adjusted β=0.33, P=0.01), total cholesterol levels (adjusted β=0.28, P=0.02), and erythrocyte ADMA levels (adjusted β=−0.30, P=0.01) were independent determinants of hemoglobin levels. The variance accounted for by total protein, total cholesterol, and diastolic BP was 0.29. The correlation with ADMA added to the variance was 0.36 using multiple regression analysis of hemoglobin levels in predialysis patients (Table 2).

Table 2.

Multiple regression analyses of hemoglobin levels in predialysis patients

Variable	Model 1		Model 2
	Adjusted β (95% CI; Lower to Upper)	P Value	Adjusted β (95% CI; Lower to Upper)	P Value
Total protein, g/dl	0.41 (0.34 to 1.34)	0.001	0.33 (0.20 to 1.17)	0.01
Total cholesterol, mg/dl	0.33 (0.003 to 0.02)	0.01	0.28 (0.002 to 0.02)	0.02
Diastolic BP, mmHg	0.32 (0.01 to 0.05)	0.01	0.31 (0.01 to 0.05)	0.01
Erythrocyte ADMA, nmol/g protein			−0.30 (–0.42 to –0.06)	0.01
	Adjusted R2=0.29		Adjusted R2=0.36

Multiple regression analyses used forced entry models. 95% CI, 95% confidence interval.

Erythrocyte ADMA Levels Were Associated with ESA Resistance Index

We next investigated whether erythrocyte ADMA levels contributed to impaired response to Epo in predialysis patients. Univariate regression analyses were performed in two separate groups according to the presence or absence of ESA treatment. Degrees of the resistance to ESA therapy were evaluated using the ESA resistance index (ERI), which was calculated from weekly injected doses of ESA (micrograms) per kilogram of body weight divided by blood hemoglobin (grams per deciliter).^9,26,27 Among 26 patients receiving ESA, ERI was evaluated in only 14 patients who received ESA for >1 month and were not taking oral iron. In the group of predialysis patients who received ESA treatment, the ERI was found to be significantly correlated with erythrocyte ADMA levels (r=0.54, P=0.05; Figure 2A). In addition, erythrocyte ADMA levels were also positively associated with plasma Epo levels per hemoglobin value (Epo demand index) in patients untreated with ESA (r=0.45, P=0.02; Figure 2B). Erythrocyte ADMA was correlated with Epo demand index (Pearson r=0.61, P=0.01) in ESA-treated patients as well. The slopes of two lines were similar in non-ESA and ESA-treated patients. Further, we performed a sensitivity analysis. However, erythrocyte ADMA was not correlated to ERI without body weight (Pearson r=0.45, P=0.11). On the other hand,, plasma ADMA levels showed no correlation with either of them (Supplemental Figure 1, A and B).

Figure 2.

Increased erythrocyte ADMA levels are associated with erythropoietin resistance in predialysis patients. (A) Erythrocyte ADMA levels show a significant direct correlation with ERI, which was calculated from weekly doses of ESA per kilogram of body weight divided by blood hemoglobin, in patients undergoing ESA therapy during >1 month without medications via oral iron supplementation (n=14). (B) Erythrocyte ADMA levels were significantly correlated with log-transformed Epo demand index, which was evaluated as the plasma Epo units divided by blood hemoglobin concentration, in patients who were not receiving ESA (n=28). Statistical significance was determined using Spearman correlation coefficient (A) or Pearson correlation coefficient (B).

Erythrocyte ADMA Accumulation and Anemia Were Improved in DDAH-1Tg Nx Mice

To examine whether erythrocyte ADMA accumulation is involved in impaired response to Epo in a mouse model of advanced CKD, we used wild-type (WT) C57BL/6 mice and DDAH-1 Tg mice and rendered them advanced CKD with 5/6 Nx. We confirmed that DDAH-1 proteins were overexpressed in the erythrocytes of DDAH-1 Tg mice by western blot (Figure 3A). Red blood cell count, hemoglobin, hematocrit, mean corpuscular volume, and mean corpuscular hemoglobin values were decreased in WT Nx mice compared with those in WT mice (Table 3), but these red blood cell (RBC) indices, although lower, remained within normal limits.²⁸ Therefore, anemia was not hypochromic or microcytic anemia in our Nx mice. These parameters, except for red blood cell count, were significantly improved in DDAH-1 Tg Nx mice. Plasma and erythrocyte ADMA levels in WT Nx mice were significantly higher than those in WT mice, both of which were ameliorated in DDAH-1 Tg Nx mice (Figure 3, B and C). Although reticulocytes were not decreased in WT Nx mice, they were significantly increased in DDAH-1 Tg Nx mice (Table 3). Plasma Epo levels were comparably elevated in WT Nx mice and DDAH-1Tg Nx mice compared with those of their respective sham-operated mice (Figure 3D). Markers of renal function, such as BUN and plasma creatinine levels, were not different between WT Nx mice and DDAH-1 Tg Nx mice. Nx mice had a higher erythropoietin level compared with sham (Figure 3D). These observations suggest that erythrocyte ADMA may cause anemia in Nx mice, whose effects are not dependent on Epo levels.

Figure 3.

Erythrocyte ADMA may cause anemia in Nx mice, whose effects are not dependent on Epo levels. DDAH-1 expression (A), plasma ADMA (B), erythrocyte ADMA (C), and plasma Epo (D) in WT and DDAH-1 Tg mice with or without 5/6 Nx. (A) DDAH-1 protein expression was detected by western blotting using a monoclonal antibody raised against hDDAH-1 in partially purified erythrocyte lysates prepared from WT (n=3) and DDAH-1 Tg mice (n=3). (B and C) Plasma and erythrocyte ADMA were measured by LC/MS. (D) Plasma Epo levels measured by a commercially-available ELISA kit. WT mice, n=9; WT Nx mice, n=14–23; DDAH-1 Tg mice, n=8–9; DDAH-1 Tg Nx mice, n=20–22. Columns express the means and error bars indicate SEM. Statistical significance was determined using a Mann–Whitney U test. *P<0.01.

Table 3.

Characteristics of the model mice with advanced CKD

Variable	WT Mice (n=12)	WT Nx Mice (n=24)	DDAH-1 Tg Mice (n=9)	DDAH-1 Tg Nx Mice (n=24)
Body weight, g	26.0±2.2	24.2±2.0	26.0±1.6	23.1±2.5a,b
Systolic BP, mmHg	110.5±10.8	114.1±10.0	107.0±9.1	109.8±10.2
Diastolic BP, mmHg	66.1±13.1	69.8±13.6	68.6±8.0	68.8±11.3
RBC (× 104/μl)	860.8±45.0	693.2±76.0a	873.7±22.5	730.1±84.6a,b
Hemoglobin, g/dl	12.9±0.6	9.7±1.2a	13.0±0.4	10.4±1.3a,b,c
Hematocrit, %	40.8±2.1	32.0±4.0a	41.2±1.0	33.8±4.8a,b,c
Mean corpuscular volume, fl	47.5±0.5	45.6±1.3a	47.2±0.7	46.8±1.7c
Mean corpuscular hemoglobin, pg/cell	14.95±0.17	13.94±0.44a	14.83±0.16	14.25±0.39c
Reticulocytes, %	3.85±0.45	4.26±0.73	4.10±0.52	4.81±1.15c
BUN, mg/dl	33.6±4.5	87.7±14.3a	26.2±3.7a	85.3±17.2a,b
Creatinine, mg/dl	0.12±0.02	0.36±0.08a	0.10±0.00	0.32±0.07a,b
WBC ( × 102/μl)	28.5±13.7	25.2±11.9	34.7±16.2	28.6±15.4
Platelets (×104/μl)	30.9±17.3	40.6±20.8	18.6±12.6	42.0±16.3b
TSAT, %	26.9±3.7	19.1±6.6b	33.0±5.2b	20.5±12.9a,b
TIBC, mg/dl	252.8±12.6	267.5±21.9a	233.8±23.2a	272.6±28.0a,b
Ferritin, ng/ml	24.1±5.2	27.5±7.4	24.2±5.6	29.9±5.9a,b

Animals were euthanized at the end wk 12 after Nx. Statistical significance was assessed with a two-tailed unpaired paired t test and approved with P<0.05. Differences in TSAT, TIBC, and plasma ADMA levels were tested using a Mann–Whitney U test and approved with P<0.05. Data are expressed as mean±SD. RBC, red blood cell counts; WBC, white blood cell counts; TSAT, transferrin saturation; TIBC, total iron biding capacity.

^aSignificant compared with WT.

^bSignificant compared with DDAH-1.

^cSignificant compared with WT Nx.

Molecular Mechanism for the Improvement of Anemia Observed in DDAH-1 Tg Nx Mice

To further clarify whether erythrocyte ADMA accumulation contributed to erythropoiesis, we investigated the gene expression of EpoR; erythroferrone, a newly identified erythroid-specific hormone that regulates iron metabolism in response to Epo²⁹; and transferrin receptor 1 in the spleens of the mice. Mouse spleen and bone marrow are hematopoietic tissues that highly express EpoR,³⁰ transferrin receptor 1,³¹ and erythroferrone.²⁸ As shown in Figure 4, mRNA expression levels of the EpoR and erythroferrone were significantly decreased in WT Nx mice compared with those in WT mice. However, both of them were restored in DDAH-1 Tg Nx mice (Figure 4, A and B). Although splenic gene expression of transferrin receptor 1, which is also the downstream target of Epo signaling,³² remained unaltered in WT Nx mice, it was significantly enhanced in DDAH-1 Tg Nx mice (Figure 4C).

Figure 4.

EpoR and erythroferrone were restored in DDAH-1 Tg Nx mice. EpoR (A), erythroferrone (B), and transferrin receptor 1 (C) splenic gene expression were analyzed in WT and DDAH-1 Tg mice with or without 5/6 Nx mice using quantitative real-time RT-PCR (WT mice, n=4; WT Nx mice, n=8; DDAH-1 Tg mice, n=4; DDAH-1 Tg Nx mice, n=8). Columns express the means and error bars indicate SEM. Statistical significance was determined using a Mann–Whitney U test. *P<0.01, **P<0.05.

Discussion

The salient findings of this study are as follows: (1) erythrocyte, but not plasma, levels of ADMA were one of the independent determinants of anemia in predialysis patients; (2) erythrocyte ADMA values were positively associated with ERI in predialysis patients and with plasma Epo levels per hemoglobin values in patients untreated ESA; (3) overexpression of DDAH-1, an enzyme which mainly degradates ADMA, significantly decreased erythrocyte levels of ADMA and simultaneously blocked the progression of renal anemia in the remnant kidney model (Nx mice); and (4) splenic gene expression of EpoR and erythroferrone were decreased in Nx mice, both of which were restored in DDAH-1 Tg Nx mice. Therefore, our present findings suggest that erythrocyte ADMA accumulation might contribute to the development and progression of impaired response to Epo with advanced CKD via suppression of EpoR expression.

In this study, increased erythrocyte ADMA levels were independently associated with decreased hemoglobin levels in predialysis patients, whereas plasma ADMA levels showed no such association (Table 1). Furthermore, ERI was found to be significantly correlated with erythrocyte ADMA levels in subjects with ESA therapy, but plasma ADMA levels were not (Figure 2A, Supplemental Figure 1A). Therefore, the accumulation of erythrocyte ADMA but not plasma might be mainly involved in impaired response to Epo in predialysis patients.

ADMA is a degradation product of methylated proteins, which is metabolized by DDAH-1.¹⁷ Because an active ADMA-metabolizing system existed in circulating erythrocytes,^24,25 we examined here whether erythrocyte ADMA accumulation played a role in impaired response to Epo in a mouse model of advanced CKD. First of all, we confirmed the relationship between plasma and erythrocyte ADMA. As shown in Supplemental Figure 2, A and B, plasma and erythrocyte ADMA levels were not correlated with each other in patients with CKD (r=−0.26, P=0.06) and Nx mice (r=0.16, P=0.60). Further, we performed additional in vitro experiments. As shown in Supplemental Figure 3, when erythrocytes were incubated with 3.5 mM ADMA for 14 hours, erythrocyte ADMA levels were significantly increased by 30%, whereas ADMA values in the medium were reduced by 10%. These observations indicate that although the difference in intracellular ADMA levels between normal patients and patients with advanced CKD was modest (17%–18%), and that the concentration relative to plasma was lower in erythrocytes, plasma ADMA may be taken up by erythrocytes.

In this study, DDAH-1 proteins were expressed in the erythrocytes of WT mice, and were overexpressed in DDAH-1 Tg mice (Figure 3A). Furthermore, the decreased levels of hemoglobin, hematocrit, mean corpuscular volume, and mean corpuscular hemoglobin values in Nx mice were significantly improved in DDAH-1 Tg Nx mice, which were associated with reduction of erythrocytes of ADMA, although plasma Epo levels were comparable between the two groups (Figure 3D, Table 3). These observations suggest that erythrocyte ADMA may cause anemia in Nx mice, an animal model of advanced CKD, whose effects are not dependent on Epo levels. Therefore, reduction of erythrocyte accumulation of ADMA by DDAH-1 overexpression may be a novel therapeutic target of impaired response to Epo in advanced CKD. In this study, we showed for the first time that mRNA levels of EpoR and erythroferrone, a newly identified erythroid-specific hormone, were decreased in Nx mice, which were significantly restored in DDAH-1 Tg mice (Figure 4, A and B). Furthermore, although there was no significant difference of transferrin receptor 1 between WT and WT-Nx mice, transferrin receptor 1 mRNA levels were significantly increased in DDAH-1 Tg Nx mice (Figure 4C), all of which except erythropoietin was significantly improved. These findings suggest the involvement of erythrocyte ADMA in EpoR expression.

We can posit an overall scheme of renal anemia in a remnant kidney model (Figure 5, A and B). Epo signal induces various erythroid responses, such as stimulation of iron utilization and erythroid differentiation.²⁹ Under conditions without advanced CKD, Epo levels are initially increased in response to anemia, which could stimulate EpoR, thus leading to transferrin receptor 1 and erythroferrone over-expression. However, under advanced CKD conditions, erythrocyte ADMA is elevated, which could suppress transferrin receptor 1, iron uptake, erythroid differentiation, and erythroferrone production, thereby being involved in Epo resistant anemia via suppression of EpoR (Figure 5B). In this study, the splenic gene expression of transferrin receptor 1 remained unaltered in WT-Nx mice compared with WT mice (Figure 4C). Transferrin receptor 1 is expressed in lymphocytes³³ and monocytes³⁴ in the spleen as well. This is one possible reason why splenic transferrin receptor 1 mRNA levels were not decreased in our advanced CKD model. In other words, ADMA accumulation or transferrin receptor-1 expression in lymphocytes and monocytes may not be affected in advanced CKD. Although it has been shown that erythroferrone reduces hepatic hepcidin expression and resultantly increases iron utility,²⁹ hepatic mRNA levels of hepcidin were not suppressed in DDAH-1 Tg Nx mice (Supplemental Figure 4). Unfortunately, we had no data to show that inflammation of uremia was present. Because IL-1 and IL-6 have also been shown to stimulate hepcidin expression,³⁵ inflammation might overcome the effects of erythroferrone on hepatic hepcidin expression in our model. Further, we performed additional studies to work out the mechanisms relating to NO. Compared with erythrocyte ADMA, the ratio of erythrocyte ADMA to arginine, a substrate for NOS, more strongly associated with ERI in ESA-treated patients (r=0.71, P=0.01), which was independent of eGFR, thus supporting the role of decreased NO production in impaired response to Epo. Given the fact that NO stimulates EpoR expression under low-oxygen conditions,^15,16 erythrocyte ADMA may reduce EpoR expression via inhibition of NO production, thus contributing to Epo-resistant anemia in advanced CKD. Although we cannot determine which portion of erythrocyte cell formation occurs in the spleen of mice, EpoR expression in CFU-erythroid (CFU-E) may be a target of ADMA because CFU-E was increased upon stimulation by NO.³⁶ Therefore, ADMA may reduce EpoR expression in CFU-E in the spleen.

Figure 5.

Epo resistant anemia via suppression of EpoR by erythrocyte ADMA accumulation. (A) Overview of Epo signaling and erythrocyte maturation without advanced CKD. (B) Overall scheme of renal anemia under advanced CKD.

Symmetric dimethylarginine (SDMA) is also described to inhibit NO production. Erythrocyte levels of SDMA, an inert isomer of ADMA corrected for protein (nanomoles per gram protein), were correlated with Hb (r=−0.33, P=0.02), but the association was lost after adjustment for eGFR in this study. Our study suggested that the pathologic role of erythrocyte ADMA, but not SDMA, corrected for total proteins in impaired response to EpoR in CKD mice and patients with CKD.

Finally, in previous studies, Epo-resistant anemia has been associated with increased risk for cardiovascular events.^37,38 In this study, we found that increased erythrocyte ADMA levels were also associated with increased serum levels of NT-proBNP, a surrogate marker for heart failure, independent of age, sex, and other confounders (Supplemental Tables 1 and 2). Erythrocyte ADMA might link impaired response to Epo to high risk of cardiovascular events including heart failure. Further longitudinal studies will be needed to clarify whether erythrocyte ADMA could be a novel biomarker and therapeutic target for Epo-resistant anemia and related cardio-renal complications in advanced CKD.

There are several limitations in this study. First, our cross-sectional study was conducted with a small population. Second, it is difficult to infer ADMA levels in erythroid precursor cells from ADMA levels in mature erythrocytes, and the role of ADMA in erythroid precursor cells in Epo-resistant anemia remains unclear. Third, erythrocyte ADMA levels were adjusted for erythrocyte protein concentration in this study. Further analysis is needed to be certain whether this adjustment is correct. Fourth, this study could not provide further description of what portion of erythroid cell formation occurs in the spleen in mice. Fifth, darbepoetin and continuous Epo activator were subcutaneously administered to 22 and four patients, respectively. Although the conversion formulas for subcutaneous routes remain unknown, we used the conversion formulas for IV injection in this study.

Concise Methods

Patients

The Ethical Committee of Kurume University, School of Medicine approved this study and all patients provided their informed consent. A total of 54 predialysis patients with ESRD were admitted at Kurume University Hospital from April of 2013 to June of 2014. We excluded patients with AKI or liver injury and subjects in the perioperative periods. The cause of renal disease was as follows: diabetic nephropathy (n=24), GN (n=12), nephrosclerosis (n=6), autosomal dominant polycystic kidney disease (n=2), lupus nephritis (n=1), and unknown (n=9). Patients received β-blocker (n=16), statin (n=17), vitamin D (n=21), phosphorus binder (n=14), activated carbon (n=12), and diuretic agent (n=27). Blood samples were taken from 31 apparently healthy volunteers who served as controls (24 men, 7 women; mean (±SD) age, 39.7±7.9 years; mean body mass index, 22.4±3.2 kg/m²; mean eGFR, 91.5±15.8 ml/min per 1.73 m²). Any volunteers with CKD, cancer, cardiovascular disease, liver dysfunction, or morbid obesity were excluded.

Measurement of Clinical Variables

Blood was drawn in the morning after an overnight fast. Japanese eGFR was calculated with the equation: GFR (ml/min per 1.73 m²) = 194 × serum creatinine^−1.094 × age^−0.287 × 0.739 (if female). Complete blood count and serum biochemical data were measured using the standard laboratory methods in the central laboratory of our hospital. Serum Epo concentrations were evaluated using a commercially-available Epo ELISA (Roche Diagnostics, Mannheim, Germany). ERI was calculated by the following formula: weekly ESA dose/body weight per blood hemoglobin concentration.^39–41 Weekly ESA doses of darbepoetin α and methoxy polyethylene glycol-epoetin β were calculated by the following formula: dose per month before the date of blood sampling/4 wk × 250 × 1.4 (if methoxy polyethylene glycol-epoetin β).^42–45

Measurement of Erythrocyte Fragility

The osmotic fragility of erythrocytes was determined using a series of NaCl solutions with concentrations ranging from 0.1% to 0.8% according to the method of Parpart AK et al.⁴⁶ In brief, 20 ml heparinized blood was added into 1 ml of each NaCl solution and stored for 30 minutes at room temperature followed by gentle mixing. After centrifugation at 3000 × g for 5 minutes, absorbance at 540 nm in the supernatant was measured and plotted against a series of NaCl solutions for the measurement of osmotic fragility of erythrocytes.⁴⁶

Measurement of ADMA

ADMA was determined by a previously described method using high-performance liquid chromatography/tandem mass spectrometry (LC/MS) with partial modification.^47,48 Erythrocyte ADMA levels were corrected for intracellular protein concentration. In brief, heparinized blood was centrifuged at 2300 × g for 10 minutes at 4°C and the plasma fraction was collected. The erythrocyte fraction after removal of the buffy coat was washed in three volumes of PBS. After the PBS and residual buffy coat were removed followed by centrifugation, 200 μl erythrocyte fraction was lysed with an equal volume of 20 mM phosphate buffer (pH 7.4; buffer A) containing 5 mM β-mercaptoethanol and centrifuged at 20,000 × g for 10 minutes at 4°C. The lysate was used to measure both ADMA and protein concentrations. The protein concentration was measured using the lysate diluted (1:300) with distilled water and the BCA protein assay reagent (Thermo Scientific, Waltham, MA). A 10 μl aliquot of erythrocyte lysate or plasma was mixed with 70 μl 0.1% FA-acetonitrile and 20 μl internal standards and stirred for 30 seconds; then, it was homogenized with a sonicator for 5 minutes. The precipitant was removed by centrifugation at 16,400 × g for 10 minutes. The supernatant was filtered through a 0.2-μm filter and used to determine guanidine compounds containing ADMA by LC/MS analysis. Intererythrocyte concentration was expressed as nanomoles of compound per gram protein in lysate.

Animal Preparation

Eight-week-old C57BL/6J background male mice and human DDAH-1 Tg male mice were purchased from Charles River (Charles River Laboratories, Sulzfeld, Germany). Offspring were screened for transgene expression by PCR as described previously.⁴⁹ In the previous reports,^49,50 reductions of plasma ADMA levels in sham and Nx DDAH-1 Tg mice were about 15% and 50%, respectively. So, the magnitudes were similar with those in the present experiments. After ketamine anesthesia (45 mg/kg intraperitoneally; Sankyo, Tokyo, Japan), 8-week-old mice were subjected to Nx (right nephrectomy with surgical resection of the lower and upper thirds of left kidney) as described previously.⁵¹ Then, the mice were euthanized, blood samples were obtained from each mouse by retro-cardiac puncture, and the kidneys, livers, and spleens were removed at 12 weeks after Nx. These samples were frozen in liquid N₂ and stored at −80°C before using for biochemical, western blot, or RNA analysis. All experimental procedures were conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by the ethical committee of our institution.

Western Blot Analysis

The erythrocyte lysate for western blotting was partially purified using Diethylaminoethyl (DEAE)-Sepharose Fast Flow (GE Healthcare Bio-Sciences AB, Uppsala, Sweden) to remove the bulk of the hemoglobin.²⁵ In brief, 100 μl lysate was added to 1 ml buffer A containing 200 μl bed volume DEAE sepharose equilibrated with the same buffer and rotated overnight at 4°C. After the DEAE sepharose was washed four times with 1 ml buffer A, the proteins were eluted three times with 200 μl buffer A containing 0.3 M NaCl. Then proteins were subjected to immunoblotting using primary antibodies against DDAH-1⁵² (1:3000 dilution) and β actin CLONE AC-15 (1:1000 dilution; Sigma, St. Louis, MO) and a peroxidase-conjugated goat anti-mouse secondary antibody (1:2000 dilution). The immune complexes were visualized with an enhanced chemiluminescence detection system using ImmunoStar LD (Wako, Osaka, Japan).

Real-Time RT-PCR

Total RNA (1 μg) extracted from each mouse spleen and liver was used to synthesize cDNA with the ReverTra Ace qPCR RT Master Mix with gDNA Remover (Toyobo, Osaka, Japan). Quantitative real-time RT-PCR was performed using a QuantiTect SYBRGreen PCR kit (Qiagen; Venlo, The Netherlands) according to the supplier’s recommendations. Primers and probe used for the analysis of mouse erythropoietin receptor (Epor), transferrin receptor (Tfrc-1), erythroferrone (Fam132b), and hepcidin (Hamp) genes were Mm_Epor_SG, Mm_Tfrc_1_SG, Mm_Fam132b_1_SG, and Mm_Hamp_1, respectively (Qiagen, Venlo, The Netherlands). Hypoxanthine guanine phosphoribosyl transferase (Hprt) (Mm_Hprt_1_SG) was used as an endogenous control (Qiagen, Venlo, The Netherlands). PCR cycling conditions were as follows: an initial denaturation step of 95°C for 15 minutes followed by 45 cycles of denaturation (15 seconds at 94°C), annealing (30 seconds at 60°C), and extension (30 seconds at 72°C). The relative amount of target gene mRNA was normalized to Hprt by the δ-δ CT method.⁵³

Statistical Analyses

Distributed variables were expressed as means±SD, and non-normally distributed variables as median (range). Ferritin and NT-proBNP were log-transformed to exhibit normal distribution before multiple regression analysis. Statistical analyses were compared by GraphPad Prism 5.0 for Windows (GraphPad Software Inc., La Jolla, CA). Multiple regression analysis was performed using IBM SPSS statistics ver. 20 (IBM, Chicago, IL). A level of P<0.05 was accepted as statistically significant.

Disclosures

None.