The AMPD2 gene, a member of the AMPD gene family encoding AMP deaminase, is widely expressed in nonmuscle tissues including kidney, although its functions have not been fully elucidated. In this study, we studied the function of the AMPD2 gene by establishing AMPD2-deficient model animal. We established AMPD2 knockout mice by using gene transfer and homologous recombination in murine ES cells and studied phenotypes and functions in the kidneys of these animals. AMPD activity was decreased from 22.9 mIU/mg protein to 2.5 mIU/mg protein in the kidneys of AMPD knockout mice. In addition to changes in nucleotide metabolism in the kidneys, proteinuria was found in 3-week-old AMPD2 knockout mice, followed by a further increment up to a peak level at 6 weeks old (up to 0.6 g/dL). The major protein component in the urine of AMPD2 knockout mice was found to be albumin, indicating that AMPD2 may have a key role in glomerular filtration. Indeed, an ultrastructure study of glomerulus specimens from these mice showed effacement of the podocyte foot processes, resembling minimal-change nephropathy in humans. Based on our results, we concluded that AMPD2 deficiency induces imbalanced nucleotide metabolism and proteinuria, probably due to podocyte dysfunction.
AMPD may be involved in the maintenance of adenylate energy charge (Morisaki & Holmes 1993; Tavazzi et al. 2000). In addition to the reports of patients with AMPD1 deficiency, other studies have shown that individuals with an AMPD1 mutation were able to tolerate cardiac failure better, suggesting that a mechanism involved with attenuation of heart failure is related to the functions of AMPD (Loh et al. 1999; Kalsi et al. 2003). Furthermore, it was reported that a genetic variant of AMPD1 may be associated with differences in insulin clearance (Goodarzi et al. 2005). Together, these results indicate that AMPDs have other important functions in addition to those known.
To understand the function of AMPD2, we established AMPD2-deficient mice. The AMPD2-deficient mice exhibited proteinuria, and a morphological study showed that these mice had pathological changes of the renal glomerulus that resembled minimal-change nephropathy in humans.
Confirmation of AMPD2 gene knockout
We confirmed the knockout of the AMPD2 gene in the mice as well as the ES cells by PCR screening and Southern blot analysis (Fig. 1A,B). Western blot analysis using anti-AMPD2 showed no immunoreactive AMPD2 protein in liver, kidney, lung, spleen, testis, and brain tissues (Fig. 1C). The total enzymatic activity of AMPD was evaluated in various tissues (Fig. 1C), with a significant decrease in AMPD activity observed in liver, kidney, and brain tissue samples. Based on these results, this is the first report of AMPD2 deficiency in an established mouse line.
The AMPD2-deficient homozygous mice were born with the expected Mendelian ratio and were fertile and visually indistinguishable from their wild-type litter mates. Furthermore, these mice did not show significant differences in food intake or life span as compared with the wild-type mice.
Nucleotide metabolism and proteinuria in AMPD2-deficient mice
To evaluate whether AMPD2 deficiency causes changes in energy balance as well as nucleotide metabolism, the levels of AMP, ADP, ATP, GTP, and IMP were measured in kidney tissues from AMPD2-deficient and wild-type mice (Table 1). There were no significant changes in AMPD2-deficient mice at the age of 2 weeks. However, a significant increase in AMP and decreases in ATP and GTP were observed in the kidneys of AMPD2-deficient mice at the age of 12 or 24 weeks as compared with those of the wild-type mice. Because mRNA of other enzymes including adenosine deaminase and adenosine kinase was not changed in AMPD2-deficient mice (data not shown), the nucleotide change was thought to be specific to AMPD2 deficiency. These findings showed that energy and nucleotide metabolism balance in the kidneys were disrupted in AMPD2-deficient mice.
Table 1. The nucleotide levels in kidney
nmol/mg tissue, n = 6–15, *P < 0.05.
1.00 ± 0.11
0.98 ± 0.10
0.55 ± 0.06
0.96 ± 0.10*
0.65 ± 0.08
1.33 ± 0.12*
1.61 ± 0.07
1.50 ± 0.10
1.50 ± 0.06
1.46 ± 0.08
1.67 ± 0.08
2.14 ± 0.17*
4.24 ± 0.60
3.79 ± 0.20
4.01 ± 0.15
3.12 ± 0.22*
3.95 ± 0.30
3.35 ± 0.37*
0.46 ± 0.02
0.34 ± 0.03*
0.41 ± 0.03
0.31 ± 0.05*
0.41 ± 0.03
0.30 ± 0.04*
0.16 ± 0.02
0.15 ± 0.02
0.14 ± 0.01
0.13 ± 0.03
0.10 ± 0.02
0.10 ± 0.02
In contrast to their otherwise unremarkable characteristics, the AMPD2-deficient mice showed proteinuria (Fig. 2A,B). Proteinuria was first distinguished at the age of 3 weeks after birth and increased until it reached the highest level at the age of 6 weeks (up to 0.6 g/dL). Thereafter, urinary protein levels tended to decrease (Fig. 2B). No proteinuria was observed in AMPD2(+/-) mice, which showed no change in nucleotide levels in kidney (data not shown).
Urinalysis showed no significant differences in color, specific gravity, pH, glucose, average urine amount, and creatinine clearance in the AMPD2-deficient mice. To evaluate whether proteinuria had an effect on systemic metabolism, blood biochemistry analyses including total protein, blood urea nitrogen, and creatinine were carried out; however, no significant differences were found for those in the AMPD2-deficient mice (serum total protein, 4.8 ± 0.4 g/dL; serum urea nitrogen, 18.3 ± 2.9 mg/dL; serum creatinine, 0.09 ± 0.02 mg/dL; n = 6).
To elucidate the protein composition of urine from the AMPD2-deficient mice, urine samples were electrophoresed on 7.5% SDS-PAGE gels (Fig. 2A). The results showed that the protein in urine was albumin, which suggested glomerular filtration function was impaired.
Histological changes in AMPD2-deficient kidney
There were no apparent differences, including shape, color, and weight, observed in the AMPD2-deficient mouse kidneys. HE-stained sections showed no obvious change in the glomerulus; though, tubules filled with protein casts were observed (Fig. 3A). There was no inflammatory cell infiltration in the AMPD2-deficient kidneys, and PAS staining did not show any significant differences in the kidneys of the AMPD2-deficient mice (Fig. 3B). Furthermore, those mice older than 1.5 years old did not show fibrosis in the kidneys (Fig. 3C). When histological changes of AMPD2-deficient kidneys were compared between mice at different ages, no obvious change was found in the glomerulus or no protein cast in tubules even in AMPD-deficient kidney at 2 weeks of age (Fig. 3D). However, tubules filled with protein casts were observed without obvious change in the glomerulus in AMPD-deficient kidney at 12 or 24 weeks of age (Fig. 3D).
Next, ultrastructural analysis using transmission electron microscopy was carried out in the AMPD2-deficient mice as well as wild-type mice. In the AMPD2-deficient mice, effacement of the podocyte foot processes was found (Fig. 4A), which suggests that AMPD2 deficiency results in podocyte dysfunction. When ultrastructure findings of AMPD2-deficient kidneys were compared among mice at different ages, there was no obvious change in the glomerulus even in AMPD2-deficient kidney at 2 or 6 weeks of age (Fig. 4B). The AMPD2-deficient kidney at 8 weeks showed effacement in some of the podocyte foot processes; though, AMPD2-deficient kidney at 12 or 32 weeks of age showed apparent effacement of the podocyte foot processes (Fig. 4B). Furthermore, relevant protein expression in the kidney of AMPD2-deficient mice at various ages was investigated, showing no different expression in any of those proteins in AMPD2-deficient mice at any age (Fig. 5).
AMPD2 expression in mouse podocytes
To determine whether AMPD2 plays an important role in podocytes, AMPD2 expression in Mouse podocyte cells (MPCs) was evaluated for protein as well as RNA levels. The results showed that AMPD2 was expressed in MPCs at both mRNA and protein levels (Fig. 6). Although undifferentiated podocytes expressed both AMPD2 and AMPD3, AMPD2 expression was increased in MPCs after differentiation with IFN-γ (−) in collagen type-4 coated dishes at 37°C and the increase in AMPD3 expression was lower than that of AMPD2 expression in differentiated MPCs (Fig. 6).
AMPD is widely distributed among mammalian cell types and plays an important role in the stabilization of energy charge (Wang et al. 1997; Korzeniewski 2006). However, the precise functional relevance of AMPD2 in kidney tissues has not been clearly established (Nowak & Kaletha 1992; Mahnke-Zizelman & Sabina 2002). To further study the function of AMPD2, we established AMPD2-deficient mice and detected very low levels of AMPD activity in liver, kidney, and brain samples. AMPD2-deficient mice exhibited proteinuria; though, there were no remarkable phenotypic differences between AMPD2-deficient and wild-type mice, such as body composition, food intake, and life span. Therefore, we conducted additional investigations into the characteristics of these mice.
AMPD2-deficient mice showed an imbalance of energy and nucleotide metabolism in the kidneys. Adenylosuccinate synthetase, adenylosuccinate lyase, and AMPD have been proposed to form a functional unit, termed the purine nucleotide cycle (Van den Berghe et al. 1992). This cycle converts AMP into IMP and then reconverts IMP into AMP via adenylosuccinate, thereby producing NH3 and forming fumarate from aspartate. In skeletal muscle tissues, the purine nucleotide cycle has been shown to function during intense exercise and plays a role in the regeneration of ATP by pulling the adenylate kinase reaction in the direction of ATP formation (Operti et al. 1998; Hellsten et al. 1999; Fukui et al. 2001). In the kidneys, the purine nucleotide cycle has been shown to account for the release of NH3 under a normal acid-base status, however, not under acidotic conditions (Bogusky et al. 1981). In a hypoxia-like condition, the rate of AMPD was increased by 41% in previously studied kidneys (Stepinski et al. 1996). In the present study, AMPD2-deficient mice showed increased AMP and decreased ATP and GTP levels in the kidneys. Therefore, AMPD is also thought to play an important role in the regeneration of ATP in kidneys in the purine nucleotide cycle.
AMPD2-deficient mice showed proteinuria from the age of 3 weeks after birth. Analysis of the protein composition of urine from these mice demonstrated that albumin was the major component, suggesting that the glomerular filtration function is primarily affected. However, no significant changes were observed histologically in glomerulus with HE and PAS staining at any period of age. Also, no apparent change in tubules was observed at 2 weeks of age, whereas tubules in AMPD2-deficient kidney filled with protein casts were seen at 12 or 24 weeks of age. There were no histological differences between the AMPD2-deficient kidneys at 12 and 24 weeks of age. Ultrastructure analysis of the nephrons by electron microscopy showed effacement of the podocyte foot processes in AMPD2-deficient mice. The ultrastructure findings of those were not different between the AMPD-deficient kidneys at 12 weeks and 32 weeks of age; though, milder changes were seen in those at 8 weeks of age. Interestingly, the AMPD2-deficient mice at not only 2 weeks but also 6 weeks of age did not exhibit podocyte foot process effacement; though, they did show proteinuria at 6 weeks of age. Therefore, we observed discrepancy between proteinuria and podocyte foot process effacement as reported before (Van den Berg et al. 2004).
Expression of the AMPD2 gene and protein in MPCs suggested that the gene plays an important role in renal glomerular function. Furthermore, AMPD2 expression was found to be apparently increased during the course of IFN-γ (−)-induced differentiation of MPCs, whereas AMPD3 expression was not. Therefore, AMPD2 is thought to be more relevant in podocyte function than other isoforms of AMP deaminase. Nevertheless, additional investigations are needed to identify how AMPD2 deficiency causes podocyte dysfunction because specific high expression of AMPD2 was not confirmed in podocytes in comparison with other cell types in the kidney. Also, further investigations will be needed to study whether nucleotide concentrations in podocytes were changed as those found in the whole kidney and caused cellular dysfunction.
Several genetic models for proteinuria and nephrotic syndrome have been reported (Ghiggeri et al. 2003; Wolf & Stahl 2003), and genetic studies of human familial cases and genetically modified animal models have identified genes responsible for glomerular filtration dysfunction, which include CD2AP, ACTN4, NPHS1, and NPHS2 (Wolf & Stahl 2003; Caridi et al. 2004; Aucella et al. 2005). However, defects in those genes were reported to cause not only proteinuria, but also nephrosclerosis eventually (Lahdenkari et al. 2004; Aucella et al. 2005), whereas the present AMPD2-deficient mice did not show any further glomerulus changes during their life span. In addition, these mice did not show any changes in total protein, urea nitrogen, or creatinine levels in serum. In addition, no cellular inflammatory reaction or fibrotic changes were observed in the kidneys, even in 1.5-year-old animals. Therefore, these AMPD2-deficient mice resembled minimal-change nephropathy in humans, in contrast to other genes reported to cause proteinuria and eventually nephrosclerosis. Regarding the relation between adenosine metabolism and nephropathy, it has been reported that adenosine A2A receptor activation attenuates inflammation and injury in diabetic nephropathy (Awad et al. 2006). Also, A1 receptor is an absolute requirement for normal tubuloglomerular feedback (Schnermann & Briggs 2008), and it was reported that A1 antagonists protect against decline in renal function seen with diuretic therapy (Gill et al. 2009). Furthermore, it has been reported that ENTPD1, an ectoenzyme metabolizing nucleotides and regulating the activity of type 2 purinergic receptors (P2X and P2Y), is a critical factor preventing microvascular injury or nephropathy in diabetes (Friedman et al. 2007). However, such findings were explained by extracellular nucleotide/adenosine-stimulated inflammation signals. Because AMPD2-deficient kidney showed rather lower ATP level, proteinuria in these mice seems to be not directly related to the purinergic receptor function found in ENTPD1-deficient mice (Friedman et al. 2007). However, it was reported that treatment of podocytes with ATP depletion induced redistribution of Neph1 and ZO-1 to the intracellular compartments, resulting in dysfunction (Wagner et al. 2008). Because AMPD2-deficient kidney altered adenylate energy charge and then decreased intracellular ATP, such conditions could induce podocyte dysfunction; though, the previous report did not use the murine model with altered adenylate energy charge but that with ATP depletion by ischemia.
In summary, this is the first report of establishment of AMPD2-deficient mice, which were used to study the function of AMPD2 in kidney tissues. We found that AMPD2 deficiency caused imbalances of energy and nucleotide metabolism in the mice kidneys. In addition, proteinuria featuring a phenotype resembling minimal-change nephropathy in humans was evident in these mice and podocyte foot process effacement was observed, suggesting that AMPD2 plays an important role in glomerular filtration function. These results are the first to show a novel important metabolic mechanism related to AMPD2 in renal glomerular function; though, additional investigations into the functions of AMPD2 in podocytes remain to be carried out.
Generation of AMPD2 mutant animals
Mouse AMPD2 cDNA clones were isolated from DNA fragments amplified by reverse transcription polymerase chain (RT-PCR) amplification using RNA isolated from P19 cells derived from a 129/sv mouse. Mouse AMPD2 genomic clones were isolated from PCR-amplified fragments of genomic DNA from P19 cells.
Long- and short-arm genomic fragments (6.8 and 1.4 kb; total 8.8 kb) of the mouse AMPD2 gene were reconstructed into a pSK vector, divided by insertion of a puromycin resistant gene driven by a PGK promoter (Fig. 1A). As a result of homologous recombination, parts of AMPD2, which contained catalytic activity sites, were presumed to be deleted.
The targeting vector was linearized with NotI and electroporated into D3 embryonic stem cells. Homologous recombination was confirmed by Southern hybridization on the both long- and short-arm sides. After undergoing a karyotype test, cells were injected into blastocysts derived from C57BL/6 mice to produce chimeric mice. Then, the chimeric males were bred to C57BL/6J females and germ-line transmission was determined. Mice with the targeted allele were back-crossed to C57BL/6 mice more than 10 times before analysis.
Genotyping and RNA analysis
Total DNA was extracted from ES cells or mouse tails using standard methods. PCR for genotype screening was carried out using the following primers for the wild-type AMPD2 gene; 5′-TTGATAGCGTGGATGATGAG-3′ and 5′-CCCGTGGGAGATGTTCTCGG-3′, and for the knockedout AMPD2 gene; 5′-TTGATAGCGTGGATGATGAG-3′ and 5′-AGACAATAGCAGGCATGCTG-3′. Southern hybridization was also carried out for screening (Fig. 1A).
Total RNA was extracted from tissues or cells using a standard protocol. Total RNA (1 μg) was reverse-transcribed by Superscript III (Invitrogen, Carlsbad, CA, USA). Quantitative real-time PCR was carried out with an ABI Prism 7000 Sequence Detection System using a SYBR Green Reagent Kit (Applied Biosystems, Foster City, CA, USA).
AMPD activity and Western blot analysis
AMPD activity was determined using HPLC as reported before (Morisaki et al. 1992). Western blot analysis was also carried out using the anti-AMPD2 antibody and then the anti-rabbit secondary antibody.
Urine was collected from the mice on plastic wrap. Urinary protein was analyzed by electrophoresis of 3 μL of urine on a 7.5% denaturing SDS-PAGE gel, followed by staining with Coomassie brilliant blue.
For routine histological analysis, kidney samples were obtained immediately after euthanasia, then fixed in 4% paraformaldehyde, and embedded in paraffin and sectioned into 5-μm sections. Staining with hematoxylin and eosin (HE) or periodic acid-Schiff (PAS) was carried out using a standard protocol. For electron microscopy, small wedge samples were cut from the kidneys and fixed in 2.5% glutaraldehyde, then dehydrated in graded ethanol and embedded in epoxy resin. Thin sections were stained with uranyl acetate.
Measurement of nucleotides
Fresh-frozen tissues were homogenized in 0.4 mol/L perchloric acid. After centrifugation, the clear supernatant was neutralized, and ten microliters of neutralized supernatant was applied to HPLC (Lachrom Elite, Hitachi, Japan) with a Capcell Pak C18 (Shiseido, Tokyo, Japan) column (Norman et al. 1994).
Immunohistochemical analysis was carried out using the primary antibodies as listed in the legends for Fig. 5. Fresh-frozen kidney tissue sections (5 μm in thickness) were used by a standard protocol. The tissue sections were preblocked for 10 min in PBS containing 5% BSA. The following primary antibodies were used: rabbit polyclonal anti-nephrin antibody (ProSci Inc., Poway, CA, USA); rabbit polyclonal anti-CD2AP antibody(SC-9137) (Santa Cruz Inc., Santa Cruz, CA, USA); rabbit polyclonal anti-α-Actinin-4(ACTN-4) antibody (Zymed Laboratories, South San Francisco, CA, USA); rabbit polyclonal anti-VEGF antibody (Thermo Scientific, Astmoor Runcorn, UK); rabbit polyclonal anti-VEGF-receptor antibody (abcam, Tokyo, Japan); rat monoclonal anti-perlecan antibody (Seikagaku, Tokyo, Japan); and goat polyclonal anti-agrin antibody (Santa Cruz Inc.). For nephrin, CD2AP, ACTN-4, VEGF, and VEGF receptor, the sections were incubated with primary antibody at 4°C overnight and then the sections were incubated with biotinylated the goat anti-rabbit (1:100) (Vector Laboratories, Burlingame, CA, USA) for 1 h at room temperature. After washing twice in PBS, avidin-biotin-peroxidase solution was added for 30 min at room temperature. For perlecan and agrin, after the sections were incubated with primary antibody at 4°C overnight, the sections were incubated with anti-rat or goat antibody labeling with the Dako EnVision reagent/horseradish peroxidase conjugated polymer (Dako, Tokyo, Japan) for 30 min. The color reaction was carried out by adding 0.02% diaminobenzene and 0.005% H2O2 in PBS for 5 min. Counterstaining was carried out with hematoxylin. The negative control experiment was carried out without the primary antibody.
Mouse podocyte cells were grown in RPMI 1640 with 10 U/mL of IFN-γ and 10% fetal bovine serum (FBS) at 33°C. To obtain differentiated podocytes, MPCs were cultured for 10 days in IFN-γ (-) media in collagen type 4 coated dishes (Mundel et al. 1997), with NHL7 cells and embryonic fibroblasts used as control cells.
Results obtained from the knockout mice were compared with those of their wild-type litter mates using a Student’s t test, with P <0.05 considered to be statistically significant. A test of multiple comparisons was also performed using one-way analysis of variance.
We express our thanks to Dr. Peter Mundel at Mount Sinai School of Medicine and Dr. Naoto Kobayashi at Ehime University for providing the MPCs, as well as their useful suggestions. We also thank the members of the Department of Bioscience and Genetics for their technical support. This work was supported by a Grant-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology of Japan, and by grants from the Japan Science and Technology Corporation, the Ministry of Health, Labour and Welfare of Japan, and by the Program for Promotion of Fundamental Studies in Health Sciences of the National Institute of Biomedical Innovation (NIBIO).