Tetrahydrobiopterin deficiency in the pathogenesis of Fabry disease

Abstract

Fabry disease is caused by deficient activity of α-galactosidase A and subsequent accumulation of glycosphingolipids (mainly globotriaosylceramide, Gb₃), leading to multisystem organ dysfunction. Oxidative stress and nitric oxide synthase (NOS) uncoupling are thought to contribute to Fabry cardiovascular diseases. We hypothesized that decreased tetrahydrobiopterin (BH4) plays a role in the pathogenesis of Fabry disease. We found that BH4 was decreased in the heart and kidney but not in the liver and aorta of Fabry mice. BH4 was also decreased in the plasma of female Fabry patients, which was not corrected by enzyme replacement therapy (ERT). Gb₃ levels were inversely correlated with BH4 levels in animal tissues and cultured patient cells. To investigate the role of BH4 deficiency in disease phenotypes, 12-month-old Fabry mice were treated with gene transfer-mediated ERT or substrate reduction therapy (SRT) for 6 months. In the Fabry mice receiving SRT but not ERT, BH4 deficiency was restored, concomitant with ameliorated cardiac and renal hypertrophy. Additionally, glutathione levels were decreased in Fabry mouse tissues in a sex-dependent manner. Renal BH4 levels were closely correlated with glutathione levels and inversely correlated with cardiac and kidney weight. In conclusion, this study showed that BH4 deficiency occurs in Fabry disease and may contribute to the pathogenesis of the disease through oxidative stress associated with a reduced antioxidant capacity of cells and NOS uncoupling. This study also suggested dissimilar efficacy of ERT and SRT in correcting pre-existing pathologies in Fabry disease.

Introduction

Fabry disease is caused by deficient activity of lysosomal enzyme α-galactosidase A (α-gal A) (1) and subsequent accumulation of glycosphingolipids (primarily globotriaosylceramide, Gb₃). Fabry disease exhibits a variety of clinical manifestations, including stroke, hypertrophic cardiomyopathy, and progressive renal impairment (2,3). Enzyme replacement therapy (ERT) is currently the standard of care for Fabry patients. Substrate reduction therapy (SRT) and pharmacological chaperone therapy are other promising approaches (4,5). A Fabry knockout mouse model is available (6). In this model, Gb₃ accumulates in most organs and late-onset cardiac and kidney hypertrophy develops (7); both occur in human patients.

The pathogenesis of Fabry disease is not well understood. However, accumulating evidence demonstrates an association of oxidative stress with the disease. Early clinical studies suggested nitric oxide (NO) dysregulation and increased reactive oxygen species (ROS) in Fabry patients (8,9). Subsequent studies also found evidence of oxidative stress (10,11). Most recently, Chimenti et al. revealed profound oxidative damage of proteins and DNA in the cardiomyocytes of Fabry patients with left ventricular hypertrophy (12). Nitrotyrosine is a marker of protein damage mediated by peroxynitrite; the latter compound can be produced by nitric oxide synthase (NOS) uncoupling (13). Increased nitrotyrosine in patients and mice with Fabry disease (8,12,14–16) suggested the presence of NOS uncoupling in this disease. More direct evidence came from the observation of decreased activity and the dimeric form of endothelial NOS (eNOS) in cultured endothelial cells and mesenteric arteries from Fabry mice (15,17). However, the detailed mechanisms of oxidative stress and NOS uncoupling and their role in Fabry disease are not fully understood.

Tetrahydrobiopterin (BH4) is an essential cofactor for the normal enzymatic function of all three isoforms of NOS and three aromatic amino acid hydroxylases (18). Suboptimal BH4 bioavailability may cause NOS uncoupling, whereby superoxide is produced instead of NO, leading to oxidative stress. Recent studies support a critical pathogenic role of this mechanism in various cardiovascular diseases (18,19), including cardiac hypertrophy (20,21). BH4 itself also functions as an antioxidant (22). Therefore, we hypothesized that the level of BH4 is decreased in Fabry disease, which may contribute to the pathogenesis of the disease. In this study, we tested this hypothesis using Fabry mice and Fabry patient-derived samples.

Results

Decreased BH4 in Fabry mouse tissues and Fabry patient plasma

We measured BH4 levels in various tissues from Fabry mice of both sexes at 5 and 12 months of age. These two time points were chosen because they represent ‘asymptomatic’ and ‘symptomatic’ disease stages, respectively, in regard to cardiac and renal hypertrophy in Fabry mice (7). Compared with wild-type (WT) mice, BH4 levels in the heart and kidneys were significantly decreased in both hemizygous male (Fig. 1A) and homozygous female (Supplementary Material, Fig. S1A) Fabry mice at both time points. BH4 deficiency in the heart and kidneys was further confirmed in 18-month-old Fabry mice (see below) (Fig. 3E). BH4 deficiency (% of WT) progressed in an age-dependent manner in the kidney, but not in the heart (Supplementary Material, Fig. S1B). In contrast to the heart and the kidney, BH4 levels in the Fabry mouse liver were either unchanged or increased compared with those of WT mice (Fig. 1A, Supplementary Material, Fig. S1A). Fabry mice also exhibited decreased BH4 levels in the plasma (Fig. 1A). In these initial analyses, we did not measure dihydrobiopterin (BH2).

Figure 1

Decreased BH4 in Fabry mouse tissues and Fabry patient plasma. (A) BH4 levels in the heart, kidney, liver and plasma of male Fabry and WT mice at 5 and 12 months (n = 8–10). (B) BH4 levels and BH4/BH2 ratios in plasma from healthy controls and Fabry patients (Pt) with and without ERT (n = 5–23). *P < 0.05, **P < 0.01, ***P < 0.001 determined via the t-test.

To determine whether BH4 is also decreased in human patients, we tested Fabry patient plasma. Female Fabry patients who did not receive ERT showed significantly decreased plasma BH4 levels and BH4/BH2 ratios compared with those of normal controls (Fig. 1B). The decreased BH4 level and BH4/BH2 ratio were not corrected in female Fabry patients receiving ERT (Fig. 1B). In contrast, there was no difference in the plasma BH4 level and BH4/BH2 ratio between male Fabry patients (with or without ERT) and normal controls (Fig. 1B). There was an inverse correlation between patients’ BH4 and/or BH2 levels and their estimated glomerular filtration rate (eGFR) (P < 0.05) (Supplementary Materials, Fig. S2A–C, Table S1). No correlation was observed between plasma BH4, BH2 and the BH4/BH2 ratio and age, proteinuria, left ventricular mass index (LVMI), and the Mainz severity score index (MSSI) in these Fabry patients (Supplementary Material, Table S1). Similar to our findings, a previous study showed that in chronic kidney disease patients plasma levels of total biopterins and BH2 are inversely correlated with creatinine clearance (23). This suggests that renal failure is associated with elevated plasma BH4 level, and the latter might mask decreased plasma BH4. To test this possibility, we compared BH4 levels in Fabry patients who had normal eGFR (> 80 ml/min/1.73m²) with normal controls. Plasma BH4 levels in ERT-treated male patients were significantly lower than that in normal controls (Supplementary Material, Fig. S2D).

Gb₃ accumulation is associated with decreased BH4

There was a significant inverse correlation between Gb₃ and BH4 levels in Fabry and WT mouse kidneys (Fig. 2A), suggesting a potential association between these two molecules. To further determine the role of glycosphingolipids in BH4, we examined whether manipulation of Gb₃ levels leads to altered BH4 levels in cultured cells. Fabry patient-derived microvascular endothelial cells (IMFE1) (10) were treated with either recombinant α-gal A or GZ161 (an inhibitor of glucosylceramide synthase) (24) for 2 weeks. Immunofluorescence analysis of Gb₃ revealed two different staining patterns in mock-treated IMFE1 cells: a cytoplasmic punctate (lysosome-like) distribution and a diffuse plasma membrane-like distribution (Fig. 2B and C). As expected, most of the lysosome-localized Gb₃ was depleted in enzyme-treated cells; however, there was no clear reduction of membranous Gb₃ in these cells (Fig. 2D), suggesting that the exogenous enzyme delivered to lysosomes is able to reduce lysosome-accumulated Gb₃ but not membrane-localized Gb₃. In contrast, both lysosomal and membranous Gb₃ staining was cleared in GZ161-treated cells (Fig. 2E). Quantitative analysis via mass spectrometry showed a significant decrease in Gb₃ in GZ161-treated cells, but not in enzyme-treated cells (Fig. 2F). The absence of a decrease in Gb₃ in lysates from enzyme-treated cells suggested that, in this experiment, Gb₃ predominates at the plasma membrane (i.e. that lysosome-accumulated Gb₃ constitutes a relatively minor fraction of the Gb₃ pool). Compared with mock- or enzyme-treated IMFE1 cells, in which the BH4 level was under the detection limit, the BH4 level was increased in GZ161-treated cells (Fig. 2G). These data suggested that abnormal metabolism of Gb₃ contributes to BH4 deficiency and that in addition to the Gb₃ present in lysosomes, the Gb₃ fraction accumulating at the plasma membrane (presumably caveolae) also plays a role in this process.

Figure 2

Correlation between Gb₃ and BH4 levels. (A) Correlation between Gb₃ and BH4 levels in 18-month-old mouse kidneys. (B–G) IMFE1 cells were treated with either α-gal A or GZ161 for 2 weeks, and the correlation between Gb₃ and BH4 levels was assessed. (B and C) Immunofluorescence showing two different distribution patterns of Gb₃ in mock-treated IMFE1 cells (arrows: punctate cytoplasmic (presumably lysosomal) distribution; arrowheads: plasma membrane-like distribution, where unlike the cytoplasmic distribution, there is no clear ‘empty nucleus’). Left: Gb₃ staining (red); Right: merged with DAPI nuclear staining. (D) Enzyme treatment cleared cytoplasmic (lysosomal) Gb₃, but not plasma membrane-localized Gb₃ (arrowheads). Left: Gb₃ staining alone; Right: merged with DAPI nuclear staining. (E) There was no detectable Gb₃ staining in GZ161-treated cells. (F) Gb₃ contents in mock-, enzyme- or GZ161-treated IMFE1 cells (n = 3). (G) BH4 levels in mock-, enzyme- or GZ161-treated IMFE1 cells (n = 3–4). ND, not detectable. ***P < 0.001 determined via the t-test.

Correlation between BH4 restoration and amelioration of disease phenotypes by specific therapies

Fabry mice develop cardiac and renal hypertrophy (7). Hence, we asked whether partial BH4 deficiency is involved in the pathogenesis of these disease phenotypes. To this end, we tested whether BH4 deficiency can be restored by ERT or SRT and, if so, whether this restoration is associated with a correction of hypertrophy. We tested adeno-associated virus (AAV)-mediated ERT and GZ161-mediated SRT in male Fabry mice. The treatments were initiated at 12 months of age, at which time cardiac and renal hypertrophy is apparent (7), and their therapeutic effects were evaluated at 18 months.

AAV-aGal expresses human α-gal A under a liver-specific promoter; if administered intravenously, α-gal A is stably expressed in the liver and is persistently secreted into the circulation and taken up by other organs (25). AAV-aGal-injected Fabry mice (Fabry-AAV) exhibited supraphysiological levels of α-gal A activity in their plasma, liver, heart and kidneys (3–157-fold WT activity) (Fig. 3A and B). The cellular distribution of human α-gal A in Fabry-AAV mouse tissues was assessed via immunohistochemistry. In the heart, α-gal A was detected in capillaries and perivascular cells but not in cardiomyocytes, and α-gal A was predominantly distributed to cortical tubular cells in the kidneys (Fig. 3C). Fabry-AAV mice showed significantly decreased cardiac, renal and liver Gb₃ levels compared with those of sham-treated Fabry mice (Fabry-sham) (Fig. 3D). However, BH4 levels in the heart, kidneys and aorta of Fabry-AAV mice were not changed relative to Fabry-sham mice (Fig. 3E). The BH4/BH2 ratios in these organs were not different between WT-sham, Fabry-sham and Fabry-AAV mice (Supplementary Material, Fig. S3A). The heart-to-body weight ratio (heart/BW) and kidney/BW ratio of Fabry-AAV mice were not reduced compared with those of Fabry-sham mice (Fig. 3F and G).

Figure 3

Effects of AAV-mediated ERT on BH4 and disease phenotypes. (A and B) α-Gal A activities in various organs (n = 6–11). (C) Cellular distribution of human α-gal A in the heart and kidney, determined via immunohistochemistry. Heart: α-gal A was detected in capillaries and perivascular cells (arrows), but not cardiomyocytes; Kidney: arrows indicate immunostaining-positive tubular epithelial cells. Scale bar: 50 μm. (D) Gb₃ levels in the liver, heart and kidneys (n = 6–10). (E) BH4 levels in heart, kidney and aorta (n = 6–10). (F and G) Heart and kidney weights normalized by body weight (n = 6–10). *P < 0.05, **P < 0.01, ***P < 0.001 determined via the t-test.

GZ161-treated Fabry mice (Fabry-GZ161) exhibited decreased cardiac and renal Gb₃ levels relative to those of placebo-treated Fabry mice (Fabry-placebo) (Fig. 4A). Compared with Fabry-placebo mice, cardiac BH4 levels were increased (P = 0.013) and renal BH4 trended toward a higher level (P = 0.063) in Fabry-GZ161 mice (Fig. 4B). The BH4/BH2 ratios in the heart and kidney were not different between Fabry-GZ161 and Fabry-placebo mice (Supplementary Material, Fig. S3B). Concomitant with the restoration of BH4, the heart/BW and kidney/BW ratios were reduced in Fabry-GZ161 mice (Fig. 4C). Furthermore, there was a significant inverse correlation between kidney BH4 levels and the weights of these organs in Fabry-placebo and Fabry-GZ161 mice (Fig. 4D). However, there was no correlation between cardiac BH4 levels and heart and kidney weight (Supplementary Material, Fig. S3C). In fact, a similar pattern of correlation (i.e. significant correlation between kidney BH4 (but not cardiac BH4) and heart and/or kidney weight) was also observed in untreated Fabry and WT mice (Supplementary Material, Fig. S4).

Figure 4

Effects of GZ161-mediated SRT on BH4 and disease phenotypes. (A) Gb₃ levels in the heart and kidneys (n = 5). (B) BH4 levels in the heart and kidneys (n = 8). (C) Heart and kidney weights normalized according to body weight (n = 9–11). (D) Correlations between kidney BH4 levels and heart and kidney weights. *P < 0.05, **P < 0.01 determined via the t-test.

Collectively, these data demonstrated a close correlation between BH4 levels and cardiac and renal hypertrophy, suggesting that BH4 deficiency may be involved in the pathogenesis of these phenotypes.

BH4 and eNOS uncoupling and glutathione deficiency

To study potential mechanisms through which BH4 deficiency might exert its function in Fabry disease, we tested the relationship between BH4 and eNOS uncoupling and oxidative stress.

NOS uncoupling is associated with decreased dimeric form of NOS protein in gel electrophoresis. BH4 structurally stabilizes the NOS protein in the active homodimeric form (19). Previous studies have shown decreased dimeric form and total expression of eNOS as well as altered phosphorylation states of the eNOS protein in Fabry mouse arteries (15). Consistent with this, the dimeric form of eNOS was decreased in Fabry mouse heart (Supplementary Material, Fig. S5A) and kidney homogenates compared with WT controls. However, dimer levels were not increased in Fabry-GZ161 mouse hearts compared to Fabry-placebo controls (Supplementary Material, Fig. S5B). Total eNOS expression levels in the heart and kidneys were similar between Fabry and WT mice (Supplementary Material, Fig. S5C) and were not changed in Fabry-GZ161 mice compared to Fabry-placebo mice (Supplementary Material, Fig. S5D). GZ161-treated IMFE1 cells, in which Gb₃ was decreased and the BH4 level was increased (Fig. 2), did not show significant changes in the dimeric form, total expression, or phosphorylation (Ser1177) of eNOS compared with mock-treated cells (Supplementary Material, Fig. S5E). In conclusion, there was no correlation between BH4 levels and the dimerization or other abnormalities of eNOS protein in these experiments.

BH4 and some other pterins function as antioxidants (22). We measured glutathione levels in Fabry mouse tissues as a general indicator of cellular oxidative stress. Compared with WT controls, the levels of GSH and/or GSSG in the male Fabry mouse heart and kidney were decreased significantly (Fig. 5A). The GSH/GSSG ratio was similar between Fabry and WT mice (Fig. 5A), suggesting that the decrease in GSH is not due to oxidation of GSH to GSSG. Glutathione levels were less affected in homozygote female Fabry mice than in male Fabry mice (Fig. 5B). Interestingly, renal BH4 levels strongly correlated with GSH and GSSG levels (Fig. 5C). These data suggested that BH4 deficiency, together with reduced abundance of glutathione, may lead to a decreased antioxidant capacity of the tissue and, thus, contribute to oxidative stress in Fabry disease.

Figure 5

BH4 and glutathione deficiency. (A) GSH and GSSG levels and GSH/GSSG ratios in the heart and kidneys from 12-month-old male Fabry and WT mice (n = 8). (B) The same parameters in 12-month-old female Fabry and WT mice (n = 8). (C) Correlations between BH4 levels and GSH and GSSG levels in male mouse kidneys. *P < 0.05 determined via the t-test.

Potential mechanism of BH4 deficiency in Fabry mice

We investigated potential mechanisms for the decreased BH4 levels in Fabry disease. We found that GTP cyclohydrolase I (GTPCH1), the enzyme that catalyzes the first and rate-limiting step in the de novo synthetic pathway of BH4, was slightly, but significantly, downregulated in the Fabry mouse heart compared with WT controls (Fig. 6A and B). The GTPCH1 protein level was not changed in Fabry mouse kidneys and was increased in the Fabry mouse liver (Fig. 6B). GZ161 treatment led to an increased GTPCH1 level in Fabry mouse hearts (Fig. 6C). These data suggested that the decreased cardiac BH4 in Fabry mice is caused at least partly by downregulated GTPCH1 expression.

Figure 6

Decreased GTPCH1 expression in the Fabry mouse heart. (A) Representative western blot images for GTPCH1 in Fabry and WT mouse heart homogenates. (B) GTPCH1 expression levels in heart, kidney and liver homogenates from 12-month-old male Fabry and WT mice assessed via western blotting and densitometry (n = 5–7). (C) Western blot for GTPCH1 in placebo- or GZ161-treated Fabry mouse heart homogenates. Lower: Intensity of the bands analyzed through densitometry (n = 6). *P < 0.05, **P < 0.01 determined via the t-test.

Discussion

The present study showed that decreased BH4 levels occur in Fabry disease and may contribute to some of its clinical manifestations. Given the known biological functions of BH4 as a cofactor and antioxidant, its contribution to disease pathogenesis is likely through oxidative stress associated with NOS uncoupling and/or a reduced capacity in scavenging ROS.

BH4 levels were decreased in Fabry mice in an organ-dependent manner. Of particular interest is that BH4 was decreased in disease-relevant organs (heart and kidney), but not in the liver that is unaffected in this disease. The absence of a significant change in the Fabry mouse aorta suggests that the decrease in BH4 in the heart and kidney may be due to changes in non-vascular cells (e.g. cardiomyocytes). The significantly decreased BH4 level in female Fabry patient plasma suggests that BH4 deficiency may also exist in human patients. However, the plasma BH4 level was not changed in male Fabry patients. The reason for this sex-dependent change is not clear. Negative correlation between renal function and plasma BH4 in male patients may contribute to the unchanged BH4 levels in these patients (Supplementary Material, Fig. S2A). Nevertheless, the biological role of BH4 in human plasma remains to be clarified. The source of BH4 in plasma and its homeostatic regulation are not clear either. A clinical study on coronary artery disease (CAD) showed that there is no association between plasma and tissue (blood vessel wall) BH4 levels in CAD patients and that plasma and vascular BH4 levels are oppositely correlated with endothelial function, suggesting that they behave as two independent compartments in CAD (26). Further investigations will be necessary to determine whether BH4 level is decreased in Fabry patient tissues.

Decreased BH4 in the Fabry mouse heart appeared to be due to downregulation of GTPCH1. Cellular BH4 levels are determined by the balance between de novo synthesis, loss of BH4 via oxidation (oxidization either to BH2 through interactions with ROS or to tetrahydrobiopterin-4a-carbinolamine during aromatic amino acid hydroxylase reactions), and regeneration of BH4 from these oxidized forms (18). These pathways involve multiple enzymes, such as sepiapterin reductase, dihydrofolate reductase and dihydropteridine reductase, and altered expression/activity of any of these enzymes may lead to aberrant BH4 levels. The normal BH4/BH2 ratio in Fabry mouse tissues implies that the decrease in BH4 is not likely due to mere oxidation of BH4 to BH2. Furthermore, it remains unclear whether the partial BH4 deficiency is Fabry disease-specific. If it is a consequence of the common pathogenic cascades in sphingolipidoses (e.g. altered lipid trafficking, autophagy and inflammation), it is possible that BH4 deficiency may also occur in other lysosomal storage disorders.

The association of BH4 restoration with ameliorated disease phenotypes in interventional studies of Fabry mice strongly suggested the pathogenic role of BH4 deficiency. Although we cannot completely exclude the possibility that the altered BH4 level is secondary to cardiac and renal hypertrophy, this theory seems unlikely because BH4 deficiency already exists at 5 months of age, when these phenotypes are not yet developed. BH4 bioavailability is a key determinant of NOS uncoupling; thus, it is possible that BH4 deficiency plays roles in Fabry disease via NOS uncoupling. However, the alleviation of BH4 deficiency and cardiac hypertrophy observed in GZ161-treated mice was not associated with the correction of eNOS dimer levels in heart tissues. This finding may be explained by cell type-specific (cardiomyocyte vs. endothelium) alterations of BH4 and eNOS and their role in cardiac hypertrophy. BH4 showed no change in the Fabry mouse aorta, suggesting that the eNOS uncoupling in the blood vessels of Fabry mice (15) is not due to BH4 deficiency. In the heart, both cardiomyocytes and endothelial cells express eNOS, and a large portion of the eNOS detected via immunoblotting comes from the endothelium due to its high expression of eNOS. Thus, it is unclear whether myocardial eNOS is uncoupled in Fabry mice and whether this is corrected by the restoration of BH4. Moens et al. showed that, in pressure overload-induced cardiac hypertrophy that is associated with eNOS uncoupling, endothelium-specific increases in BH4 synthesis do not prevent hypertrophic remodelling and cardiac dysfunction (21), suggesting a lower impact of endothelial BH4 and eNOS on cardiac hypertrophy.

Another possibility is that neuronal NOS (nNOS), rather than eNOS, is the dominant NOS isoform affected by decreased BH4 in cardiac cells. A recent study showed that a cardiomyocyte-specific increase in BH4 levels (by myocardial overexpression of GTPCH1) leads to increased activity of nNOS (but not eNOS) and nNOS-dependent improvement of myocardial relaxation (27), suggesting an important role of BH4 in the normal function of nNOS in cardiomyocytes, and that the requirement for BH4 among different NOS isoforms may differ between cell types. More comprehensive studies are needed to determine myocardial NOS uncoupling and its relationship with BH4 deficiency in Fabry mice.

A striking observation in the present study was that despite being able to achieve high enzyme activity and decreased Gb₃ contents in tissue homogenates, AAV-mediated ERT did not correct either BH4 deficiency or cardiac/renal hypertrophy in Fabry mice. These results are most likely due to insufficient delivery of the enzyme to disease-relevant cell types. For example, in the heart, cardiomyocytes are the primary target of ERT; however, the distribution of the enzyme is limited to vascular and perivascular cells. In the Fabry mouse heart, inclusions (glycosphingolipid deposits) were found in vascular cells and interstitial fibroblasts but not in cardiomyocytes (28), suggesting that the majority of Gb₃ in the heart came from vascular cells and fibroblasts. This is the reason why ERT can significantly decrease Gb₃ levels in heart homogenates, even as it fails to reach the cardiomyocytes. Nevertheless, the cardiac hypertrophy observed in Fabry mice (7,29) suggests that storage of Gb₃ and/or related glycosphingolipids occurs in cardiomyocytes. The degree of storage may be low (as there were no detectable inclusions) but could be sufficient to cause cellular damage. Lyso-Gb₃ (30) is a good candidate for such a storage molecule because lysosphingolipids can be toxic at very low concentrations. Furthermore, due to its relatively hydrophilic nature, lyso-Gb₃ may not form inclusion bodies (i.e. the inclusions observed in tissue sections may not reflect the storage of lyso-Gb₃).

In conclusion, our data suggest that lysosomal storage and dysfunction in Fabry mouse cardiomyocytes are not corrected by ERT. In contrast to enzymes, GZ161 is distributed to most cell types, including neural cells (24), which may be the reason for the different outcomes of these two therapies in our study. We also show the potential role of SRT in ameliorating pre-existing cardiac and renal phenotypes in Fabry disease, even in a null GLA mutant.

Another important finding of this study was the decrease in GSH and/or GSSG levels in male Fabry mouse tissues. The close correlation between glutathione and BH4 in the kidneys suggests that they either are downstream of the same pathway or have a cause-effect relationship. The low cellular GSH level may play an important role in oxidative stress in Fabry disease. In a BH4-deficient state, uncoupled NOS-generated superoxide reacts with NO to form the much more powerful oxidant peroxynitrite. Under normal circumstances, the deleterious effects of peroxynitrite are diminished by GSH, which is the major endogenous scavenger of peroxynitrite (13). Thus, coexistence of decreased GSH with BH4 deficiency in Fabry mice may exacerbate peroxynitrite-dependent cellular damage. The significant increase in nitrotyrosine in Fabry disease (8,12,14–16) supports this hypothesis. In addition to its role as an antioxidant, GSH has diverse functions in various cellular processes such as apoptosis and cytokine production (31). Therefore, it is possible that decreased GSH is also involved in Fabry disease through mechanism(s) other than oxidative stress.

In contrast to male Fabry mice, glutathione levels were not significantly altered in female Fabry mice. A similar finding has been reported in Alzheimer’s disease; GSH is decreased in red blood cells from male but not female Alzheimer’s patients relative to healthy controls (32). Sex differences in GSH level and related enzyme activities in the rat liver have also been reported (33). These findings suggest a role of sex hormones in glutathione metabolism. Previously, we showed that enhanced androgen receptor (AR) signalling contributes to disease phenotypes in Fabry mice (7). Perhaps the role of AR occurs partly via oxidative stress. Indeed, transgenic overexpression of kidney androgen-regulated protein (KAP) leads to increased oxidative stress (34).

In summary, our findings provide new insights into the pathophysiology of Fabry disease in the context of developing new therapeutic approaches. Animal studies have suggested beneficial effects of BH4 supplementation on vascular and heart function (18,19). Future studies should be conducted to test whether supplementation of BH4 or other BH4-targeted approaches (e.g. upregulation of GTPCH1) can prevent and correct the manifestations of Fabry disease.

Materials and Methods

Animals and treatments

Animal procedures were reviewed and approved by the Institutional Animal Care and Use Committee of Baylor Research Institute. Both the Fabry (6) and wild-type (WT) mice had a B6/129 mixed strain background, with a ∼ 75% C57BL/6 background (7). At the indicated time points, tissues were collected, snap frozen on dry ice and stored at −80 °C until use. For the thoracic aortas, surrounding connective tissues and fats were removed under a stereomicroscope. To obtain mouse plasma for the BH4 assay, blood was collected in tubes containing ethylenediaminetetraacetic acid (EDTA) and dithiothreitol (DTT) to prevent the oxidation of BH4.

For adeno-associated virus (AAV)-mediated enzyme therapy, AAV-aGal was produced and purified as described previously (25) and was injected into 12-month-old male Fabry mice via the jugular vein at a dose of 4x10¹¹ viral particles per mouse. In the un-injected Fabry and WT controls, a sham operation was performed. Plasma was collected from the tail vein at the indicated time points to monitor α-gal A activity in the plasma. At 18 months of age, tissues were harvested for analysis. For substrate reduction therapy using GZ161 (an inhibitor of glucosylceramide synthase) (24), male Fabry mice were fed a diet containing 0.011% GZ161 (equivalent to 20 mg/kg/day). In the placebo groups, the same diet without GZ161 was used. Similar to enzyme therapy, GZ161 was administered beginning at 12 months of age, and its effects were evaluated at 18 months of age.

Human plasma for BH4 analysis

Informed consent was obtained from all subjects. Blood samples from Fabry patients and normal controls were collected in EDTA tubes containing DTT (1 mg/ml). The plasma samples were immediately stored at −80 °C until analysis.

Cell culture and treatments

IMFE1, a microvascular endothelial cell line originating from a skin biopsy of a Fabry patient (10), was cultured in EGM-2MV medium (Lonza). The cells were treated with either agalsidase alfa (at a final concentration of 5 µg/ml) or GZ161 (0.5 µM) for 2 weeks. The culture media were changed every 2–3 days with media containing freshly added compounds; the cells were passaged once during the treatment. Mock-treated control cells were cultured in parallel, and PBS was used in place of the enzyme or GZ161. At the end of the treatments, the cells were confluent, and they were harvested as described below. For the BH4 and Gb₃ assays, cells were trypsinized and rinsed, and cell pellets were stored at −80 °C until use. For eNOS western blotting, cells were rinsed in ice-cold PBS and harvested in lysis buffer (10 mM Tris, 150 mM NaCl, 5 mM EDTA, 1% Triton X-100, pH 7.4) supplemented with a protease inhibitor cocktail, PMSF, sodium orthovanadate and sodium fluoride. Next, the lysates were sonicated and spun at 14,000 rpm for 15 min at 4 °C, and the supernatants were used for protein assays and immunoblotting. For immunocytochemistry, cells grown on coverslips were fixed in 2% paraformaldehyde for 20 min at room temperature for Gb₃ staining.

Analysis of BH4 and BH2

Standards for (6R)-5,6,7,8-tetrahydrobiopterin (BH4) and L-7,8-dihydrobiopterin (BH2) and labelled ¹⁵N-BH4 and ¹⁵N-BH2 stable isotopes were obtained from Schircks Laboratories (Switzerland). Formic acid, perchloric acid (PCA), DTT, heptafluorobutyric acid, ammonium acetate (Sigma), methanol (Fisher), and water of liquid chromatography mass spectrometry (LC-MS/MS) grade were used. Stock calibration standards and internal standards were prepared individually for the BH4, BH2, ¹⁵N-BH4 and ¹⁵N-BH2 solutions at a concentration of 1 mmol/l in Milli-Q water containing 0.2% DTT and stored at −80 °C.

We used two different methods to measure BH4 levels. In initial analyses (data in Fig. 1A, and Supplementary Material, Fig. S1A), BH4 was measured using high performance liquid chromatography (HPLC) with electrochemical detection as previously described (35). We found that this method lacked the required sensitivity to measure BH2 in samples. Therefore, we applied an LC-MS/MS method (36) that provided increased sensitivity to detect both BH4 and BH2 in plasma, tissues and cultured cells. The mass spectrometry data were acquired and processed on a Sciex 5500QTRAP using Analyst 1.5.2 software (ABSciex).

Plasma. The analysis of BH4 and BH2 in human plasma was performed via LC-MS/MS as previously described (36). Working calibrators were prepared in water containing 0.2% DTT in the range of 6.25–200 nmol/l. An internal standard solution containing 15N-BH4 and 15N-BH2 was prepared in water containing 0.2% DTT at a final concentration of 200 nmol/l. Samples were prepared in a Microcon YM-10, 10 kDa NMWL microcentrifugal filter unit through the addition of 180 µL of internal standard solution to 20 µL of standard, quality control solution or plasma. The samples were mixed and centrifuged for 20 min at 14,000 rpm at 4ºC, and the filtrates were used for analysis.

Tissues and cultured cells. The analysis of BH4 and BH2 in mouse tissues was conducted using LC-MS/MS as previously described (37), with slight modifications in sample preparation. Tissues were homogenized and deproteinized in 0.1 M PCA containing 0.2% DTT using an overhead stirrer with a pestle. Following centrifugation at 14,000 rpm for 10 min at 4ºC, the clear supernatant was diluted with 4 volumes of water containing 0.2% DTT. Working calibrators were prepared in 0.1 M PCA containing 0.2% DTT over the range of 6.25–400 nmol/l. An internal standard solution was prepared as described above, except that 0.4 M PCA with 0.2% DTT was used instead of water. BH4 and BH2 levels were normalized according to the wet weight of the tissues. For cultured cells, samples were homogenized via sonication in 0.1 M PCA containing 0.2% DTT and then centrifuged. The supernatants were used for BH4 analysis, and the pellets were reconstituted in 0.1 N NaOH for protein assays; the data were normalized according to total protein levels.

Analysis of glutathione

Standards for glutathione and GSSG were purchased from Sigma. The levels of GSH and GSSG in tissues were determined by HPLC using a boron-doped diamond cell for electrochemical detection as previously described (38), with some minor modifications. Tissues were homogenized and deproteinized in 4 volumes of 0.1 M PCA, and the lysates were centrifuged at 14,000 rpm for 10 min at 4ºC. The supernatant was then diluted with a mobile phase and injected into the HPLC-EC system.

Enzyme assay

α-Gal A activity was measured using the fluorometric method as described previously (39).

Gb₃ assay

Gb₃ was measured using mass spectrometry as described previously (39). Gb₃ levels were calculated as the sum of the concentrations of 8 isoforms and were normalized according to total protein levels.

Gb₃ immunofluorescence staining

Immunostaining was performed using mouse monoclonal antibody for Gb₃ (Seikagaku, Tokyo, Japan) as described previously (7).

Immunohistochemistry

The localization of human α-gal A in mouse tissues was detected using immunohistochemistry as described previously (39).

Western blotting

Mouse tissues were lysed using a Potter-Elvehjem homogenizer, with subsequent sonication in the lysis buffer used for cultured cells described above. The proteins were fractionated in NuPAGE Bis-Tris gels (Invitrogen) and then transferred to PVDF membranes. The primary antibodies were mouse monoclonal antibodies for eNOS (clone M221, Abcam) and GTPCH1 (Santa Cruz Biotech) and polyclonal antibodies for phosphorylated eNOS (Ser1177) (Cell Signaling) and GAPDH (Santa Cruz Biotech). Protein levels were quantified via densitometry using ImageJ software.

Low-temperature SDS-PAGE (LT-PAGE) was performed to detect eNOS dimers as described previously (40) with some modifications. Lysates (25 µg total protein) were mixed with sample buffer in the presence of 2.5% 2-mercaptoethanol on ice and then subjected to electrophoresis. The gels (NuPAGE 4–12% Bis-Tris or Novex 6% Tris-glycine gels, Invitrogen) and buffers were equilibrated at 4 °C, and electrophoresis was performed in a 4 °C refrigerator. For denatured (eNOS monomer) controls, the samples were heated at 70 °C for 10 min. The blotting step was performed as a regular western blot. Under these conditions, eNOS in unheated IMFE1 lysates is dominantly present as a dimer, along with some monomer bands, whereas heated IMFE1 lysate almost solely produces a single monomer band with a molecular weight of ∼140 kDa (Supplementary Material, Fig. S5A and E). eNOS that was overexpressed in cultured mouse cells using an adenoviral vector encoding human eNOS (41) produced the same result. Similarly, eNOS dimers were detected in mouse tissue homogenates; however, the eNOS monomer bands from mouse tissues were obscured by ‘non-specific’ smears in LT-PAGE (Supplementary Material, Fig. S5A).

Statistical Analysis

Data are presented as the mean ± SEM. Significance was determined using Student’s t-test.

Supplementary Material

Supplementary Material is available at HMG online.

Acknowledgements

We thank Julie Chao (Medical University of South Carolina) for providing the adenovirus Ad.CMV-eNOS; Yang-Ming Yang (New York Medical College) for suggestions on LT-PAGE; and Dilechia Hawthorne and Albert Barnes (Baylor Research Institute, Dallas) for technical assistance.

Conflict of Interest statement. None declared.

Funding

Baylor Research Institute, and in part by a grant from the Cardiovascular Research Review Committee (CVRRC) of the Baylor Scott & White Health System (North Texas Division).

Author notes

†

These authors contributed equally to this study.

‡

Co-senior authors.