Non-purine selective xanthine oxidase inhibitor ameliorates glomerular endothelial injury in Ins^Akita diabetic mice

Abstract

Endothelial dysfunction represents a predominant early feature of diabetes, rendering patients with diabetes prone to renal complications, e.g., proteinuria. Recent studies have indicated a possible role for xanthine oxidase (XO) in the pathogenesis of vascular dysfunctions associated with diabetes. In the present study, we investigated the contribution of XO activation on the progression of diabetic nephropathy in a mouse model using selective XO inhibitors. Male Ins2^Akita heterozygous mice were used with wild-type mice as controls. Akita mice were treated with topiroxostat (Topi) or vehicle for 4 wk. Serum uric acid levels were significantly reduced in Akita + Topi mice compared with Akita + vehicle mice. The Akita + Topi group had a significant reduction in urinary albumin excretion compared with the Akita + vehicle group. Mesangial expansion, glomerular collagen type IV deposition, and glomerular endothelial injury (assessed by lectin staining and transmission electron microscopy) were considerably reduced in the Akita + topi group compared with the Akita + vehicle group. Furthermore, glomerular permeability was significantly higher in the Akita + vehicle group compared with the wild-type group. These changes were reduced with the administration of Topi. We conclude that XO inhibitors preserve glomerular endothelial functions and rescue compromised glomerular permeability, suggesting that XO activation plays a vital role in the pathogenesis of diabetic nephropathy.

INTRODUCTION

Diabetic nephropathy (DN) is a major cause of end-stage renal disease (38). Many factors have been reported to be involved in the pathogenesis of DN, such as inflammatory cytokines, growth factors, vasopressor peptides (ANG II and endothelin-1), advanced glycation end products, and reactive oxygen species (ROS). ROS are believed to be generated from several sources in the diabetic kidney, including glycolysis, mitochondrial dysfunctions, accentuation of the polyol pathway, excessive production of advanced glycation end products, uncoupling of nitric oxide (NO) synthase (NOS), and activation of NAD(P)H oxidase and xanthine oxidase (XO) systems (9). The consensus among various investigators is that ROS are the common denominators in the contribution to glomerular endothelial dysfunctions, and we have previously elucidated the molecular mechanism pertaining to such cellular dysfunctions in DN (18). Also, recent studies have indicated an important role of the XO system in the pathogenesis of vascular dysfunctions associated with diabetes per se (8). However, it is unclear whether XO activity is consistently involved in the pathogenesis of DN.

The enzyme xanthine oxidoreductase (XOR) plays an important role in the catabolism of purines. XOR oxidizes hypoxanthine to xanthine and xanthine to uric acid. XOR exists in two interconvertible forms: XO and xanthine dehydrogenase (XDH) (3). In the process of hypoxanthine/xanthine metabolism, XDH catalyzes electron transfer to NAD⁺ and generates NADH. XO transfers electrons to O₂ and generates H₂O₂ and ${\text{O}}_2^{-}$. Thus, the XO form is regarded as the source of ROS (3, 12). XO is mainly expressed in the liver and intestine (24). XO is not a source of ROS that induce tissue damage under normal physiological conditions, but XO generates excessive ROS in several pathological conditions, such as chronic heart failure (7), obesity (40), atherosclerosis (17), hypertension (4), and end-stage renal disease (5). ROS generated via the XO system induce inflammation through the activation of NF-κB in the liver (31). In this regard, increased XO expression and oxidative stress have been reported in the kidney (20, 22) in a rat model of streptozotocin-induced diabetes. Following an increased hepatic expression XO, it is then transported via the circulation to target organs in diabetes (8), where it gets trapped in the endothelium and inevitably causes dysfunction (1, 3).

Topiroxostat (Topi), a non-purine selective XO inhibitor, is traditionally used for the treatment of hyperuricemia and gout. Recent investigations have demonstrated that non-purine selective XO inhibitor ameliorates renal dysfunctions in several experimental animal models, including those of subtotal nephrectomy (35), ischemia-reperfusion renal injury (39), unilateral ureteral obstruction (30), and streptozotocin-induced diabetes (19, 20). In humans, Topi has been reported to reduce urinary albumin excretion in patients with chronic kidney disease (stage 3) (11). In view of the above observations, we sought to clarify the relationship between XO activation and the progression of DN and to characterize the renoprotective effects of selective XO inhibitors via elucidating improved glomerular endothelial functions in Ins2^Akita heterozygous mice.

METHODS

Animal model.

The experimental protocol (no. 14-058) was approved by the Ethics Review Committee for Animal Experimentation of Kawasaki Medical School. Male C57BL/6 mice and Ins2^Akita heterozygous (Akita) mice were purchased from CLEA Japan (Osaka, Japan) and housed in a temperature- and humidity-controlled room with a 12:12-h light-dark cycle. Mice were fed standard laboratory chow, and they had free access to tap water. Nine-week-old male C57BL/6 mice (body weight: 20–27 g) and Akita mice (body weight: 19–23 g) were used for this study. C57BL/6 mice were used as the control group, and Akita mice were randomly divided into the following two groups (n = 6 mice/group): the vehicle group and the Topi group. Mice in the the Topi group were administered Topi (3 mg·kg⁻¹·day⁻¹) by gavage for 4 wk (26). At the 13th week, 24-h urine samples of the mice were collected. Blood samples were taken from the heart immediately after euthanization. Their body weight and blood pressure were recorded, and they were euthanized. Their kidneys were collected and weighed. Fasting serum glucose, serum creatinine, and serum uric acid levels were measured using enzymatic assays. HbA1c was measured with the A1c-Now-rapid measurement kit (Bayer, Osaka, Japan). Urinary albumin excretion was determined using a murine microalbuminuria ELISA kit (Albuwell M, Exocell, Philadelphia, PA).

Histopathological examination.

Kidneys were fixed in 4% paraformaldehyde and embedded in paraffin for histological analysis. Sections measuring ∼4 μm thick were deparaffinized and subjected to periodic acid-Schiff (PAS) staining. Collagen type IV- and nitrotyrosine-stained slides were photographed by a microscope (Nikon Coolscope, Nikon, Tokyo, Japan), and podocin (NPHS2)-stained slides were photographed using a confocal laser scanning microscope (FV1000, Olympus, Tokyo, Japan). A color image analyzer (WinLoof, Mitani, Fukui, Japan) was used to evaluate each stain. For analysis of diffuse glomerular lesions in diabetes, the percentages of the PAS-positive area were measured. For evaluation of collagen type IV, NPHS2, and nitrotyrosine staining, percentages of the stain-positive area were measured. Lycopersicon esculentum lectin (LEL)-stained slides were photographed with a fluorescence microscope (BZ-9000, Keyence, Osaka, Japan), and the disarray score of LEL staining was blindly assessed by three researchers. Briefly, staining was scored in terms of the amount of degradation (0, none; 1, mild; 2, moderate; 3, severe; and 4, global degradation) as previously described (18). A total of 20 glomeruli were randomly selected each tissue section from each animal, and the average score was calculated. Tissues were also processed for electron microscopy (JEM-1400, JEOL, Tokyo, Japan) to assess ultrastructural alterations in the endothelium.

Immunohistochemical staining.

Deparaffinized kidney sections were heated in a microwave at 600 W for antigen retrieval and then incubated with antibodies against collagen type IV (Abcam, Cambridge, UK), NPHS2 (Abcam), and LEL (Sigma-Aldrich, St. Louis, MO). Cryostat frozen kidney sections measuring ∼5 μm thick were used for immunohistochemical analysis of nitrotyrosine (Upstate, Lake Placid, NY). Primary antibody was detected using the Histofine Simple Stain MAX-PO kit (Nichirei, Tokyo, Japan) and 3,3′-diaminobenzidine (Sigma-Aldrich) for collagen type IV, NPHS2, and nitrotyrosine staining.

In situ detection of NO and ROS.

Levels of NO and ROS generated from ROS coupling were assessed using confocal laser microscopy after renal perfusion using CellROX Deep Red and diaminofluorescein-FM diacetate. The detailed method have been previously described (36).

In vivo imaging of macromolecule hyperfiltration.

In vivo imaging of the glomerular microcirculation was performed as previously described (37). Briefly, for fluorescent probes, 500-kDa fluorescein-dextran solution (green anionic, excitation/emission maxima: 494/518 nm) was obtained from Invitrogen (Tokyo, Japan). The A1R MP+ multiphoton excitation laser scanning fluorescence microscopy (Confocal Microscope System, Nikon) was used. A 70-kDa rhodamine-dextran solution (reddish) was infused through the jugular venous catheter to assess glomerular permeability along with high-molecular-weight dextran (500 kDa).

XOR activity in the liver, kidney, and plasma.

XOR activity was measured as previously described (25). Liver and kidney homogenates were centrifuged at 105,000 g and 4°C for 60 min. Either cytosol or plasma was added to a mixture containing [¹⁵N₂]-labeled xanthine (0.8 mmol/L), NAD⁺ (1 mmol/L), and potassium oxonate (0.013 mmol/L) in 20 mmol/L Tris buffer (pH 8.5) and then incubated at 37°C for 30 min. Thereafter, [¹³C₂,¹⁵N₂]-labeled uric acid (UA) was added as the internal standard, and the mixtures were subsequently heated for 5 min at 95°C and then recentrifuged at 15,000 g and 4°C for 10 min. The supernatant was filtered through an ultrafiltration membrane (Amicon Ultra-0.5 3K centrifugal filtration device, Millipore, Burlington, MA), and [¹⁵N₂]-labeled UA ([¹⁵N₂]UA) was measured by liquid chromatography-mass spectrometry (LC/MS; LTQ-Orbitrap, ThermoFisher Scientific, Waltham, MA). Activity was expressed as [¹⁵N₂]UA nmol·min⁻¹·mg protein⁻¹.

Statistical analysis.

Data are expressed as means ± SE and were analyzed using a two-tailed unpaired Student’s t test or one-way ANOVA for comparison of multiple means. Statistical significance was set at P < 0.05.

RESULTS

Status of body weight, systolic blood pressure, and other physiological parameters in the various groups of mice.

Body weight, systolic blood pressure, and other physiological parameters in the various groups of mice are shown in Table 1. The kidney weight-to-body weight ratio was significantly increased in all Akita mice compared with the wild-type (WT) + vehicle group. There were no significant differences between Akita + vehicle mice and Akita + Topi mice among the various physiological parameters. Serum glucose and HbA1c levels were significantly elevated in all Akita mice, but there was no significant difference between Akita + Topi and Akita + vehicle mice. Serum creatinine levels also showed no significant differences among all groups. Serum UA levels were significantly reduced in Akita + Topi mice compared with Akita + vehicle mice. Interestingly, urinary albumin levels were significantly increased in Akita + vehicle mice compared with WT + vehicle mice. They were significantly reduced in Akita + Topi mice compared with Akita + vehicle mice (Fig. 1).

Table 1.

Biochemical and physiological parameters of the various strains of mice

	Wild-Type + Vehicle Group	Akita + Vehicle Group	Akita + Topiroxostat Group
Body weight, g	26.6 ± 0.5	23.2 ± 0.5*	24.7 ± 0.6*
Left kidney weight/body weight, %	0.63 ± 0.02	1.04 ± 0.04*	1.12 ± 0.06*
Systolic blood pressure, mmHg	119 ± 1	106 ± 2*	100 ± 4*
Serum glucose, mmol/L	13.2 ± 0.3	40.1 ± 0.9*	39.3 ± 0.9*
HbA1c, %	4.6 ± 0.1	10.3 ± 0.7*	10.1 ± 0.6*
Serum creatinine, μmol/L	9.90 ± 0.33	8.31 ± 0.35	8.84 ± 0.51
Serum uric acid, mg/dL	1.56 ± 0.1	1.9 ± 0.2	0.32 ± 0.1*†

Data are shown as means ± SE; n = 6 mice/group.

^*P < 0.05 vs. the wild-type + vehicle group;

^†P < 0.05 vs. the Akita + vehicle group.

Fig. 1.

Urinary albumin excretion in the various strains of mice. Shown is an evaluation of urinary albumin excretion in wild-type (WT) + vehicle, Akita + vehicle, and Akita + topiroxostat (Topi) mice. Data are presented as means ± SE. *P < 0.05 vs. WT + vehicle mice; †P < 0.05 vs. Akita + vehicle mice.

Status of glomerular mesangial matrix changes in the various groups of mice evaluated by PAS and collagen type IV staining.

PAS-positive area expansion was seen in Akita + vehicle mice, and it was more than in WT + vehicle mice (Fig. 2, A and D). Topi treatment reduced mesangial expansion in Akita mice (Fig. 2, A and D). The collagen type IV-positive area was also increased in Akita + vehicle mice compared with WT + vehicle mice. Topi treatment reduced collagen deposition in Akita mice (Fig. 2, B and E). Podocyte damage was evaluated using NPHS2 immunohistochemical staining (Fig. 2C). Quantitative analysis showed no significant differences in the NPHS2-stained area among the various groups (Fig. 2F). Morphological changes in the glomerular capillary endothelium were also assessed by scanning and transmission electron microscopy (Fig. 3). A decrease in glomerular endothelial fenestrae was observed in Akita + vehicle mice, while fenestrations were preserved in Akita + Topi mice (Fig. 3).

Fig. 2.

Morphological evaluation of glomerular alterations. A−C: glomerular periodic acid-Schiff (PAS) staining (A), collagen type IV immunohistochemical staining (B), and podocin (NPHS2) immunohistochemical staining (C). Bars = 20 μm. D−F: quantitative determination of the PAS-positive area (D), collagen type IV-positive area (E), and NPHS2-positive area (F). Data are presented as means ± SE. *P < 0.05 vs. wild-type (WT) + vehicle mice; †P < 0.05 vs. Akita + vehicle. Topi, topiroxostat.

Fig. 3.

Ultrastructural evaluation of the glomerular capillary wall in the various strains of mice. Shown are cross-sectional transmission electron micrographs of the glomerular capillary in wild-type (WT) + vehicle, Akita + vehicle, and Akita + topiroxostat (Topi) mice. Bars = 1 μm. Arrowheads indicate endothelial fenestrae.

Status of hepatic and renal XOR activity in the various groups of mice.

Topi treatment significantly reduced liver XOR activity (Fig. 4A). XOR activity in the kidney was significantly increased in Akita + vehicle mice compared with WT + vehicle mice. However, a significant reduction in XOR activity was seen in Akita + Topi mice (Fig. 4B). The status of ROS and NO in kidney tissue was evaluated by confocal laser microscopy. CellROX Deep Red and diaminofluorescein-FM diacetate were used to detect ROS and NO in the same kidney tissue section. Increased ROS generation and decreased NO bioavailability were observed in the glomerular compartment of Akita + vehicle mice, but both were normalized to large extent in Akita mice that received Topi (Fig. 4C).

Fig. 4.

Status of xanthine oxidoreductase (XOR) activity in the liver and kidney and of nitric oxide (NO) and reactive oxygen species (ROS). A and B: XOR activity in the liver (A) and kidney (B). C: detection of NO and ROS in the same tissue sections. Data are presented as means ± SE. *P < 0.05 vs wild-type (WT) + vehicle mice; †P < 0.05 vs. Akita + vehicle mice. Topi, topiroxostat; G, glomerulus.

Status of nitrotyrosine and LEL staining in the glomerular compartment.

In Akita + vehicle mice, only the glomerular mesangial area showed positivity following nitrotyrosine staining (Fig. 5A). The extent of staining area became notably much larger in Akita + vehicle mice than in WT + vehicle mice, and it was reduced following Topi treatment (Fig. 5C). Glomerular endothelial surface layer disruption was detected using LEL staining in Akita + vehicle mice, and it was significantly smaller compared with the WT + vehicle mice (Fig. 5B). The disruption in lectin stainability was significantly normalized with the administration of Topi (Fig. 5D).

Fig. 5.

Status of glomerular oxidative stress and of glomerular endothelial injury. A and B: immunohistochemical staining for nitrotyrosine (A) and Lycopersicon esculentum lectin (LEL; B). Bars = 20 μm. C: quantitative determination of the nitrotyrosine-positive area. D: disarray score of LEL staining. Data are presented as means ± SE. *P < 0.05 vs. wild-type (WT) + vehicle; †P < 0.05 vs. Akita + vehicle. Topi, topiroxostat.

Assessment of glomerular capillary permeability of macromolecules following intravenous injection of 70-kDa rhodamine-conjugated dextran.

Using this previously described method (37), a very small volume of filtered 70-kDa rhodamine-labeled dextran was detected in Bowman’s space in WT + vehicle mice (Fig. 6, top, and Supplemental Video S1, available online at https://doi.org/10.6084/m9.figshare.12319595.v1). A massive leakage of dextran (faint reddish color) was observed in Akita + vehicle mice (Fig. 6, middle, and Supplemental Video S2, available online at https://doi.org/10.6084/m9.figshare.12319595.v1). This increased permeability of 70-kDa dextran, however, was notably reduced in the glomeruli of Akita + Topi mice (Supplemental Video S3, available online at https://doi.org/10.6084/m9.figshare.12319595.v1).

Fig. 6.

Evaluation of glomerular permeability in the various strains of mice. A: representative serial images showing filtration of macromolecules in glomeruli per mouse groups. Green indicates 500-kDa fluorescein-labeled dextran solution; red indicates 70-kDs rhodamine-dextran solution. Arrowheads indicate leakage of 70-kDa rhodamine-dextran solution into Bowman's space (*), which is quite remarkable in the Akita + vehicle group (middle). Time per frame = 870 ms. Topi, topiroxostat; G, glomerulus.

DISCUSSION

This study demonstrates that XOR activity is increased in the diabetic kidney compared with the normal kidney and that the non-purine selective XO inhibitor Topi reduces albuminuria and kidney tissue damage in diabetic states via attenuation of glomerular oxidative stress-induced endothelial damage. These effects were observed to be dependent of the serum UA levels. Thus, it seems that XOR activation plays a considerable role in the progression of DN and that XO inhibition may be a good therapeutic avenue in dampening the progression of DN by reducing ROS damage to glomerular endothelial cells.

Recently, many clinical studies have reported on the renoprotective effects and efficacy of Topi. Wada et al. (41) showed that Topi effectively reduced serum UA levels with good tolerability and safety in patients with early diabatic nephropathy. Hosoya et al. (10) demonstrated the safety and efficacy of Topi in Japanese patients with hyperuricemia. Mizukoshi et al. (23) suggested that Topi reduced albuminuria in patients with diabetic nephropathy. These clinical studies demonstrated that Topi is safe and effective compared with existing medications and might suppress the decline in renal function in patients with both hyperuricemia and diabetic nephropathy.

UA has been reported to induce glomerular hypertension, arteriolosclerosis, glomerulosclerosis, and interstitial fibrosis (13). The conceivable mechanism of such injuries may be due the deposition of UA crystals in the distal tubules and collecting ducts. Alternatively, other proposed mechanisms by which UA could contribute to renal injury may be related to the activation of the renin-angiotensin system (6), tubular epithelial cell transition (32), and oxidative stress and endothelial dysfunctions (33, 34). Nakamura et al. (26) have shown that Topi decreased serum UA and urinary albumin excretion and inhibited plasma XOR activity in db/db mice. In the present study, Topi reduced serum UA and urinary albumin excretion in Akita mice. These data were consistent with the previous results. For that reason, non-purine selective XO inhibition reduced urinary albumin excretion and renal pathological changes, characterized by endothelial injury, and it is likely that suppression of XO activity somehow retards the prognosis of DN.

XO is an important source of ROS. In patients with end-stage renal disease, serum XO activity has been reported to be higher than in healthy patients and patients with chronic renal failure. It was also relatively high in patients with chronic renal failure than in healthy patients (5). To date, there is no evidence for a causal relationship between XO activity and the progression of chronic kidney disease, and the association between XO activity and diabetic nephropathy remains unclear. In some of the studies, XO inhibitors was shown to be renoprotective, such as in a sub-total nephrectomy model (35), ischemia-reperfusion renal injury model (39), a unilateral ureteral obstruction model (30), and a diabetic nephropathy model (16, 19, 20). However, it remains unclear whether these effects depend on either UA reduction or XO activity inhibition. Our study suggests the possible involvement of XO activity in DN since renoprotective effects of the non-purine selective XO inhibitor Topi exerted considerable ameliorative effect in our model system.

XO is mainly expressed in the liver and intestine (24), and it is released from these organs into the circulation in a diabetic state (8). XO also binds with sulfated glycosaminoglycans that are expressed on endothelial surfaces, and the glomerular endothelial glycocalyx is quite enriched with such proteoglycans in normal WT mice (1). In this study, no difference in XOR activity between WT and Akita mice in liver tissue was noted, but XOR activity was higher in kidney tissue of Akita mice. Furthermore, there were no effects on podocytes in WT or Akita mice, and injury was confined to the glomerular endothelia, as assessed by lectin staining and electron microscopy. A consistent finding reported in the literature suggests a reduced endothelial glycocalyx in Akita mice (14). These results may suggest that free XO released from the liver might bind and be taken up by the glomerular endothelium in the diabetic state, after which renal XO generates excessive ROS and thus contributes to cellular injury.

In terms of the effectivity of other XO inhibitors, no differences were observed in the reduction of albuminuria between Topi and febuxostat in our study (data not shown). In contrast, Topi inhibited XOR activity more strongly than febuxostat in the kidney. Other studies have indicated that Topi binds XOR not only by covalent binding with essential molybdenum in the enzyme active site but also by interacting with multiple acid residues of XOR, via multiple hydrogen bonds, hydrophobic interactions, and aromatic interactions (29). In contrast, febuxostat acts on XOR by structure-based inhibition only, i.e., via some of the interactions eluded above (28). The differences in these binding modes may suggest the possibility that these two drugs have different XOR inhibitory effects on the kidney. However, renal outcomes were not affected. Additionally, it has been reported, although not examined in this study, that the UA-lowering drug allopurinol is metabolized to oxypurinol and then binds solely by covalent bonding with molybdenum at the active site of XOR. Further experiments are needed to elucidate the comparative role relative to allopurinol.

We measured XOR activity in the liver and kidney in this study. XOR exists in two interconvertible forms: XO and XDH (3). XO generates ROS during the process of hypoxanthine/xanthine metabolism, whereas XDH does not generate ROS (3, 12). The LC/MS method used in our study cannot separately assay XO/XDH. However, it has been reported that XOR is swiftly converted to XO after release into the plasma from various organs (2, 15, 21), so we regard that XOR activity approximates XO activity, which, of course, is a minor limitation of this study. Nevertheless, inhibition of XO by Topi had a beneficial effect. We used 9-wk-old Akita mice, which were administered Topi for 4 wk. Urinary albumin excretion at the preadministration timeframe was already increased in these mutant mice compared with WT mice (data not shown), so certainly one can consider that non-purine selective inhibitors of XO prevented the progression of DN in Akita mice. Further studies are needed to examine whether the protective effect in the pathogenesis of DN is feasible when Topi is administered at an earlier time point or advanced stage of DN.

The endothelial cell layer is covered with glycocalyx and works to control the permeability of the glomerulus. The thickness of the glycocalyx layer is reduced in patients with diabetes with microalbuminuria (27). We have previously shown that hyperfiltration of macromolecular dextran can be detected in the glomeruli of streptozotocin-induced diabetic rats (37). We have also demonstrated the presence of glomerular glycocalyx in Zucker fatty rats and proved the involvement of oxidative stress-induced heparinase in the reduction of the glycocalyx layer, which suggests that the reduction in the glycocalyx layer has a role in the increased urinary albumin excretion (18). In the present study, although we were unable to examine the detailed molecular mechanism, it is possible that a similar mechanism might be responsible for the increased urinary albumin excretion.

In summary, we demonstrated higher XOR activity in the diabetic kidney compared with the normal kidney. Renal XO produces ROS in endothelial cells, which then leads to perturbation in the endothelial homeostasis and albuminuria. The non-purine selective XO inhibitor Topi reduces albuminuria via attenuation of endothelial damage induced by glomerular oxidative stress derived by XO activation. These effects were observed to be dependent on serum UA levels. Overall, XOR activation can be regarded as to play an important role in the progression of DN, and XO inhibition therapy may have a protective effect in the amelioration of DN due to a reduction of glomerular endothelial ROS.

GRANTS

This work was supported by Grants-in-Aid for Scientific Research (B) from the Japan Society for the Promotion of Science (18H02828) and Research Project Grants from Kawasaki Medical School (27-104 and 28-036) as well as National Institute of Diabetes and Digestive and Kidney Diseases Grant DK60635 (to Y.S.K.).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

M.S. conceived and designed research; S.I. performed experiments; S.I., T.N. and T.M. analyzed data; h.k. prepared figures; h.k. drafted manuscript; h.k., M.S., and Y.S.K. edited and revised manuscript; T.S., Y.S.K. and N.K. approved final version of manuscript.