Autophagy Inhibits the Accumulation of Advanced Glycation End Products by Promoting Lysosomal Biogenesis and Function in the Kidney Proximal Tubules

Diabetic nephropathy is the most common cause of end-stage kidney disease worldwide (1). Although various factors contribute to the pathogenesis of diabetic complications, advanced glycation end products (AGEs), which are formed by the irreversible attachment of reducing sugars onto amino groups of proteins in the hyperglycemic state, are a common factor (2–5). Numerous studies have indicated that AGEs cause structural and functional alternations in various organs, including the kidneys (6). The formation of AGEs can cause generalized cellular dysfunction by inducing cross-linking of extracellular matrices such as collagen (7,8). In addition to the increased production in the hyperglycemic state, decreased clearance by the kidneys is responsible for the accumulation of AGEs in patients with diabetic nephropathy (9). AGEs filtered by glomeruli or delivered from the circulation are endocytosed and then, together with endogenous AGEs produced by the glycation of intracellular proteins, degraded in the lysosomes of the proximal tubular epithelial cells (PTECs) (10,11).

Macroautophagy, hereafter autophagy, is a highly conserved degradation system that functions to regulate intracellular homeostasis (12,13). Cytoplasmic components, such as organelles and proteins, are sequestered into a double-membrane cytosolic vesicle called an autophagosome. Then, the autophagosome fuses with a lysosome, resulting in the degradation of the sequestered materials. It has been demonstrated that autophagy deficiency leads to the accumulation of protein aggregates, resulting in tissue injury in various organs including the central nervous system, liver, and kidneys (14–16). We have recently reported that autophagy plays a protective role not only in acute kidney injury, such as ischemic reperfusion injury and cisplatin nephropathy, but also in chronic disorders such as cyclosporine nephropathy and metabolic acidosis (17–20). Recently, a novel relationship between autophagy and lysosomes has emerged: autophagy sequesters the damaged lysosome to control lysosomal biogenesis in acute hyperuricemic nephropathy, which is termed lysophagy (21,22).

With regard to the role of autophagy in diabetic nephropathy, it has been reported that autophagy in podocytes plays a protective role by maintaining lysosomal homeostasis (23). In addition, it has been demonstrated that AGE overload disrupts the autophagic pathway because of lysosomal membrane permeabilization (LMP) (24). However, it remains unclear whether autophagy in the proximal tubules can exert a protective role against diabetic nephropathy by the maintenance of lysosomes, where a large number of AGEs are endocytosed. Thus, we hypothesized that autophagy contributes to the degradation of AGEs in the diabetic kidney by modulating lysosomal biogenesis. To address this hypothesis, we investigated whether endogenous and exogenous AGE overload could affect autophagic activity and lysosomal biogenesis in vivo and in vitro. In addition, we explored the consequences of autophagy deficiency on kidney function and morphology in a streptozotocin (STZ)–induced murine diabetic nephropathy model, focusing especially on inflammasome activation and kidney fibrosis.

Atg5^F/F:kidney androgen-regulated protein (KAP) mice have been described previously (18). Briefly, we crossed KAP-Cre transgenic mice, which express Cre recombinase in the kidney proximal tubules, with mice bearing an Atg5^flox allele. We have demonstrated efficient knockout of the ATG5 protein in Atg5^F/F:KAP mice (84.2% reduction in kidney cortex homogenates of 8-week-old mice) (17). All experiments were performed using 8-week-old male mice. Mice were allowed access to water and mouse chow ad libitum. All animal experiments were approved by the institutional committees of the Animal Research Committee of Osaka University and the Japanese Animal Protection and Management Law (no. 25). Diabetes was induced by intraperitoneal injection of STZ (50 mg/kg) dissolved in a sodium citrate buffer (pH 4.5) for 4 consecutive days. As a control, vehicle-injected (sodium citrate buffer) mice were followed concurrently. These mice were sacrificed 4 months after STZ or vehicle injection. Spot urine samples were collected before sacrifice.

We used the following antibodies and reagents: antibodies for carboxymethyllysine (CML; Transgenic, Kobe, Japan); collagen type I and BSA (Abcam, Cambridge, U.K.); α-smooth muscle actin (α-SMA), β-actin, transcription factor EB (TFEB), and epidermal growth factor receptor (EGFR; Sigma-Aldrich, St. Louis, MO); F4/80 (Bio-Rad, Hercules, CA); lysosomal-associated membrane protein (Lamp) 1 (BD Biosciences, Franklin Lakes, NJ); LC3, interleukin (IL)-1β, and phosphorylated (p-)S6K1 (Cell Signaling Technology, Beverly, MA); biotinylated secondary antibodies (Vector Laboratories, Burlingame, CA); horseradish peroxidase–conjugated secondary antibodies (DakoCytomation, Glostrup, Denmark); Alexa 488– and 555–conjugated secondary antibodies (Molecular Probes, Eugene, OR); and bafilomycin, STZ, EGF, l-leucyl-l-leucine methyl ester (LLOMe), and wortmannin (Sigma-Aldrich).

Plasma glucose, blood urea nitrogen, serum creatinine, total cholesterol, and triglycerides were measured using the Glu-Test-Wako, BUN-Test-Wako, Cholesterol E-test Wako, Triglyceride E-test Wako (Wako Pure Chemical Industries, Osaka, Japan), and CRE-EN Kainos test (Kainos, Tokyo, Japan), respectively. Urinary albumin concentration was determined using the MICROFLUORAL Microalbumin test (Progen, Heidelburg, Germany). Plasma and urinary CML were measured using the Circulex anti-CML mouse autoantibody ELISA kit (CircuLex, Nagano, Japan). Urinary neutrophil gelatinase–associated lipocalin (NGAL) was measured using Mouse Lipocalin-2/NGAL Quantikine ELISA kit (R&D Systems, Minneapolis, MN).

Atg5-deficient PTECs (Atg5 [−] PTECs) isolated from 3-week-old Atg5^F/F:KAP mice and Atg5-revertant PTECs (Atg5 [+] PTECs) were described previously (17). Construction of GFP–galectin 3–expressing retrovirus vector and establishment of GFP–galectin 3 stably expressing Atg5 (+) and Atg5 (−) PTECs were described previously (21).

According to previous reports, AGE-BSA and BSA were prepared by incubating endotoxin-free BSA (Sigma-Aldrich) with 0.5 mol/L d-glucose in 0.4 mol/L PBS or with PBS alone at 37°C for 8 weeks, respectively (25,26). Free glucose was removed through Amicon centrifugal filters with a 50-kDa molecular weight cutoff (Millipore, Temecula, CA). The concentrations of AGE-BSA and BSA were estimated by the Bradford method.

Atg5 (−) and Atg5 (+) PTECs were cultured in DMEM (Sigma-Aldrich) containing 5% FBS. To evaluate the effect of high glucose on intracellular AGE formation, confluent Atg5 (−) and Atg5 (+) PTECs were incubated for up to 72 h in DMEM containing either 5.6 or 25 mmol/L glucose and 1% FBS. To examine the effect of extracellular AGEs (and BSA) administration, confluent Atg5 (−) and Atg5 (+) PTECs were incubated for 6 or 24 h in DMEM (containing 5.6 mmol/L glucose and 1% FBS) with vehicle (PBS) or 40 μmol/L BSA or AGE-BSA.

Western blot analysis was conducted as described previously (18). Band intensities were measured using ImageJ software (National Institutes of Health, Bethesda, MD).

Atg5 (−) and Atg5 (+) PTECs were precultured in DMEM without FBS for 5 h and incubated in DMEM containing 1% FBS with vehicle (PBS) or 40 μmol/L BSA or AGE-BSA for 24 h. After adding 100 ng/mL EGF (time point: 0 min), cells were cultured for the indicated times, and then Western blot analysis for EGFR was performed.

Quantitative RT-PCR was performed as described previously (27). Primers were as follows: nucleotide binding domain, leucine-rich repeats–containing family, pyrin domain-containing (NLRP) 3, 5′-ATGCTGCTTCGACATCTCCT-3′ and 5′-AACCAATGCGAGATCCTGAC-3′; apoptotic speck protein containing a caspase recruitment domain (ASC), 5′-ACAGAAGTGGACGGAGTGCT-3′ and 5′-CTCCAGGTCCATCACCAAGT-3′; caspase-1, 5′-CACAGCTCTGGAGATGGTGA-3′ and 5′-TCTTTCAAGCTTGGGCACTT-3′; MCP-1, 5′-CCCAATCAGTAGGCTGGAGA-3′ and 5′-TCTGGACCCATTCCTTCTTG-3′; NGAL, 5′-CTACAACCAGTTCGCCATGG-3′ and 5′-ACACTCACCACCCATTCAGT-3′; α-SMA, 5′-AGATCAAGATCATTGCCCCTCC-3′ and 5′-TTGTGTGCTAGAGGCAGAGC-3′; collagen type I, 5′-ACGCCATCAAGGTCTACTGC-3′ and 5′-ACTCGAACGGGAATCCATCG-3′; and 18S ribosomal RNA, 5′-AAACGGCTACCACATCCAAG-3′ and 5′-CCTCCAATGGATCCTCGTTA-3′. The relative amounts of RNA were depicted as ratios to 18S ribosomal RNA.

Immunohistochemical staining for CML, Lamp1, megalin, F4/80, IL-1β, α-SMA, and collagen type I was performed as described previously (18). Antigen retrieval was performed by autoclaving in 0.01 mmol/L citrate buffer (pH 6) for 10 min at 120°C (for CML, Lamp1, megalin, IL-1β, α-SMA, and collagen type I) or by treatment with proteinase K solution (20 μg/mL) in TE buffer (10 mmol/L Tris-HCl [pH 8] and 1 mmol/L EDTA) for 3 min at room temperature (for F4/80).

Immunocytochemistry for TFEB using Atg5 (+) and Atg5 (−) PTECs, which were exposed to PBS, BSA, or AGE-BSA, was performed as described previously (21).

To assess LMP, the number of GFP-galectin 3–positive dots were counted in Atg5 (+) and Atg5 (−) PTECs stably expressing GFP-galectin 3, which were exposed to PBS, BSA, or AGE-BSA (or LLOMe as a positive control). Cells were observed using an FV1000 confocal microscope (Olympus, Tokyo, Japan).

All results are presented as means ± SE. Statistical analysis was conducted using JMP software (SAS institute, Cary, NC). Multiple group comparisons were performed using ANOVA with posttesting according to the Tukey-Kramer test. A P value <0.05 was considered statistically significant.

STZ (50 mg/kg) or a vehicle was intraperitoneally injected into control (Atg5^F/F) and KAP-Cre/floxed Atg5 (Atg5^F/F:KAP) mice, in which autophagy is deficient in the PTECs, at 8 weeks of age for 4 consecutive days. Four months after treatment, we found no significant difference in plasma glucose levels, kidney weight, blood pressure, or kidney function as assessed by blood urea nitrogen and serum creatinine between STZ-treated control and Atg5^F/F:KAP mice (Table 1). Notably, urinary albumin excretion, as evaluated by the albumin-to-creatinine ratio, was not elevated in either of the STZ-treated mice (control and Atg5^F/F:KAP), in agreement with the previous reports (28,29). Although periodic acid Schiff staining revealed no apparent morphological changes in STZ-treated control and Atg5^F/F:KAP mice, immunohistochemical and Western blot analysis for CML demonstrated that more AGEs accumulated in the STZ-treated Atg5^F/F:KAP mice than in the STZ-treated control mice (Fig. 1A–C). The amount of CML in urine significantly increased by STZ treatment in both control and Atg5^F/F:KAP mice (Fig. 1D). Of interest, the plasma CML level significantly increased only in STZ-treated Atg5^F/F:KAP mice (Fig. 1E). In addition, accumulation of CML in the glomerulus was prominent in STZ-treated Atg5^F/F:KAP mice compared with STZ-treated control mice (Supplementary Fig. 1A). Moreover, the intimal thickness of intrarenal artery by CML accumulation was observed in STZ-treated Atg5^F/F:KAP mice, whereas it was rarely observed in STZ-treated control mice (Supplementary Fig. 1B). These results indicate that autophagy in the proximal tubules suppresses the accumulation of AGEs in the whole kidney.

To investigate whether endogenous AGE overload affects autophagic flux in the proximal tubular cells, autophagy-deficient PTECs isolated from Atg5^F/F:KAP mice (Atg5 [−] PTECs) and Atg5-revertant cells (Atg5 [+] PTECs) were exposed to normal (5.6 mmol/L) or high (25 mmol/L) glucose medium for a prolonged period (72 h). It was demonstrated that more CML accumulated in the Atg5 (−) PTECs incubated in the high-glucose medium compared with the other conditions (Fig. 2A). Next, we incubated Atg5 (+) PTECs in normal or high-glucose states for up to 72 h to examine autophagic flux. LC3-II level was increased in the cells incubated in the high-glucose medium for 72 h, which was unchanged after bafilomycin treatment, whereas a substantial conversion of LC3-I to LC3-II by bafilomycin treatment was observed in other conditions (Fig. 2B). We excluded the possibility that the observation with regard to autophagic activity was not attributed to high osmolarity itself by the addition of mannitol to the normal glucose medium. Moreover, wortmannin, an inhibitor of autophagosome formation, suppressed the increase of LC3-II by bafilomycin in the high-glucose state (Supplementary Fig. 2A). These results suggest that the high-glucose state gradually blunted in the later step of autophagy.

Next, to examine the role of autophagy when PTECs are exposed to exogenous AGEs, Atg5 (+) and Atg5 (−) PTECs were exposed to AGE-BSA or BSA for 24 h. We found more AGE-BSA accumulated in the AGE-BSA–treated Atg5 (−) PTECs than in the AGE-BSA–treated Atg5 (+) PTECs, whereas we observed less BSA in the BSA-treated Atg5 (−) PTECs compared with the BSA-treated Atg5 (+) PTECs (Fig. 2C). We also examined autophagic flux in AGE-BSA– or BSA-treated Atg5 (+) PTECs for up to 24 h. LC3-II level was not significantly increased in the cells treated with AGE-BSA for 24 h, which was unchanged after bafilomycin treatment similar to endogenous AGEs, whereas a substantial conversion of LC3-I to LC3-II by bafilomycin treatment was observed in other conditions (Fig. 2D). Moreover, wortmannin suppressed the increase of LC3-II by bafilomycin in AGE-BSA–treated Atg5 (+) PTECs (Supplementary Fig. 2B). These results suggest that exogenous AGEs also gradually disrupt the later step of autophagy.

The observation that high glucose and AGE overload led to the gradual stagnation of autophagic flux prompted us to investigate whether they could modulate lysosomal biogenesis and function. Atg5 (+) and Atg5 (−) PTECs were exposed to high glucose (or normal glucose) or AGE-BSA (or BSA). The protein levels of Lamp1 were comparable between normal and high-glucose conditions in both Atg5 (+) and Atg5 (−) PTECs, indicating that high glucose itself does not affect lysosomal biogenesis (Fig. 3A). In contrast, we observed increased protein levels of Lamp1 in BSA- and AGE-BSA–exposed Atg5 (+) PTECs, but not in BSA- or AGE-BSA–exposed Atg5 (−) PTECs, suggesting that autophagy upregulates lysosomal biogenesis upon protein overload (Fig. 3B). In line with these observations, the protein level of TFEB, a master regulator of lysosomal biogenesis, was higher in Atg5 (+) PTECs than in Atg5 (−) PTECs, and the difference was prominent after AGE overload (Fig. 3B). In addition, nuclear translocation of TFEB was impaired in Atg5 (−) PTECs irrespective of protein overload (Fig. 3C). To investigate the mechanism underlying autophagy-dependent TFEB activation, we explored the mammalian target of rapamycin (mTOR) pathway, which regulates the nuclear translocation of TFEB (30). Protein levels of p-S6K1 (downstream from the mTOR pathway) were comparably increased after 24 h of exposure to BSA or AGE-BSA in both Atg5 (+) and Atg5 (−) PTECs (Supplementary Fig. 3).

We next examined the effect of autophagy on lysosomal function using an EGFR degradation assay, in which the capacity of EGF-induced degradation of EGFR reflects lysosomal function. After 24 h of exposure to BSA or AGE-BSA, the degradation of EGFR protein was comparably promoted in Atg5 (+) PTECs, whereas the degradation was significantly delayed in Atg5 (−) PTECs (Fig. 3D). Collectively, autophagy upregulates lysosomal biogenesis and function in response to protein or AGE overload.

Previous reports that autophagy is induced to degrade the damaged lysosome after the uptake of mineral crystals such as silica and monosodium urate or after photodamage prompted us to investigate whether BSA or AGE overload could cause lysosomal damage (21,31). We explored galectin 3, which is localized to the endosomal compartment from cytoplasm in response to endosomal membrane rupture (32). The lysosomotropic compound LLOMe, which evokes lysosomal damage, caused the formation of GFP-positive dots in Atg5 (+) PTECs stably expressing GFP–galectin 3, whereas few GFP-positive dots were observed in both Atg5 (+) and Atg5 (−) PTECs during BSA or AGE-BSA exposure (Fig. 3E). These data indicate that LMP was not induced during BSA or AGE overload.

To recapitulate the in vitro observation that autophagy contributes to the upregulation of lysosomal biogenesis in response to AGE overload in vivo, we next examined the effect of autophagy on lysosomal biogenesis in the proximal tubules of diabetic mice. The protein level of Lamp1 in the kidney of STZ-treated control mice was increased compared with that in the nondiabetic state (Fig. 4A). Intriguingly, this upregulation was not evident in the kidney of STZ-treated Atg5^F/F:KAP mice (Fig. 4A). Consistently, immunohistochemical analysis indicated that the increase in Lamp1-positive lysosomes in the proximal tubules of STZ-treated control mice was not observed in those of STZ-treated Atg5^F/F:KAP mice (Fig. 4B). These results suggest that autophagy contributes to the upregulation of lysosomal biogenesis in the diabetic state in vivo.

Because it has been reported that AGE accumulation could evoke chronic inflammation, we examined the effect of autophagy on macrophage infiltration and inflammasome activation induced by the diabetic state (33,34). Immunohistochemistry for F4/80 showed more macrophage infiltration in the kidneys of STZ-treated Atg5^F/F:KAP mice compared with those in nondiabetic mice (control and Atg5^F/F:KAP) and STZ-treated control mice (Fig. 5A). Moreover, mRNA levels of NLRP3, ASC, and caspase-1 increased in the kidneys of STZ-treated Atg5^F/F:KAP mice compared with STZ-treated control mice (Fig. 5B). In addition, the protein level of IL-1β, as assessed by Western blot analysis and immunohistochemistry, increased in the kidneys of STZ-treated control mice compared with vehicle-treated control mice, which was more pronounced in the Atg5^F/F:KAP mice (Fig. 5C and D). These results indicate that exacerbation of macrophage infiltration and NLRP3-mediated inflammation induced by the diabetic state becomes apparent with autophagy deficiency.

To determine whether lysosomal overload of AGE-BSA in the proximal tubules directly induces macrophage infiltration and inflammation as observed in vivo, we evaluated MCP-1 expression in vitro. BSA or AGE-BSA exposure did not affect mRNA levels of MCP-1 in Atg5 (+) PTECs, whereas significantly increased expression of MCP-1 was detected in Atg5 (−) PTECs (Fig. 5E), suggesting that lysosomal overload induced by BSA or AGE-BSA exposure can directly induce inflammation in autophagy-deficient PTECs.

We next investigated the effect of autophagy on kidney injury and fibrosis in the diabetic state. mRNA expression and urinary concentration of the tubular injury marker, NGAL, was increased in STZ-treated Atg5^F/F:KAP mice compared with STZ-treated control mice (Fig. 6A). Because it has been reported that AGEs induce fibrotic change in the diabetic state, we evaluated the expression of α-SMA and collagen type I, hallmarks of fibrosis (25). Immunohistochemistry for α-SMA and collagen type I demonstrated that the kidneys of STZ-treated Atg5^F/F:KAP mice exhibited increased areas of positive staining compared with STZ-treated control mice (Fig. 6B and C). mRNA expression of α-SMA and collagen type I were more significantly increased in the kidneys of STZ-treated Atg5^F/F:KAP mice than those in STZ-treated control mice (Fig. 6D and E). These results indicate that autophagy protects the kidneys from injury and fibrosis in the diabetic state.

In this study, we demonstrated that 1) lysosomal biogenesis and function in the kidney proximal tubules are upregulated in the diabetic state, at least partially, to cope with AGE overload; 2) both endogenous and exogenous AGEs gradually disrupt autophagic flux; and 3) autophagy in PTECs plays an essential role in the degradation of AGEs via lysosomal degradation.

Our study cannot completely exclude the possibility that high glucose itself stimulates lysosomal biogenesis in vivo, because we cannot segregate the effects of hyperglycemia and AGE overload on AGE accumulation in autophagy deficiency; however, our result that high glucose does not stimulate lysosomal biogenesis in vitro supports the possibility that lysosomal biogenesis and function are upregulated in response to AGE overload.

In this study, more AGEs accumulated in the proximal tubules of proximal tubule-specific, autophagy-deficient mice than in those of control mice in the diabetic state. One possible explanation is that the amount of intracellular proteins that could be a target of glycation increase because of the lack of canonical (bulk) autophagy in the diabetic state, which is supported by the in vitro study using a high-glucose medium. Alternatively, but not exclusively, accumulation of AGEs in the autophagy-deficient kidney is attributed to insufficient degradation in the lysosomes of the proximal tubules. The fact that plasma concentration and glomerular deposition of CML increased in proximal tubule-specific, autophagy-deficient diabetic mice indicates that autophagy in the PTECs copes with the increasing number of AGEs that exist in urine and plasma by promoting upregulation of the endocytosis-lysosomal system. How are lysosomal biogenesis and function upregulated by autophagy during AGE overload? We previously reported that ruptured lysosomes are directly sequestered by autophagy in the PTECs when exposed to monosodium urate crystals, termed lysophagy (21,35). In this study, LMP, as assessed by GFP–galectin 3–positive dots and electron microscopic analysis of the PTECs, was not observed, even during AGE overload (Fig. 3E and data not shown). Recently, it has been reported that AGE-BSA retards autophagic activity because of LMP in HK-2 cells (24). The precise mechanism to explain this discrepancy remains unknown, but it may be related to the experimental conditions such as the type of cells used or the concentration of AGE-BSA; however, our result that the nuclear translocation of TFEB was impaired in autophagy-deficient PTECs even without AGE overload could shed light on an essential role of autophagy in lysosomal biogenesis other than lysophagy. Because the levels of p-S6K1 protein were comparable between Atg5 (+) and Atg5 (−) PTECs, autophagy-dependent TFEB activation seems to be mTOR independent, at least in our experimental conditions.

Similar to AGE-BSA, BSA activated the endocytosis-lysosomal system in autophagy-competent PTECs, as is evidenced by the Lamp1 protein level and EGFR assay, although BSA did not retard autophagic activity. In autophagy-deficient PTECs, unlike AGE-BSA, BSA did not accumulate despite the lack of upregulation of lysosomal function. This is presumably because BSA may be more easily degraded than AGE-BSA, which is irreversibly glycated. Another possibility is that an endocytosis pathway is disturbed in autophagy-deficient PTECs because of the nonautophagic effect of Atg5, as is suggested by our observation (Fig. 2C). In either case, BSA-induced upregulation of lysosomal biogenesis and function via autophagy could be critical for proteinuric glomerular diseases including advanced diabetic nephropathy.

Another prominent finding of this study is that tubular injury, inflammation, and interstitial fibrosis are exacerbated in the diabetic proximal tubule–specific, autophagy-deficient kidney. Considering that the blood glucose levels were comparable, it is possible that AGE accumulation because of autophagy deficiency leads to such changes. In fact, many reports have shown that AGEs induce extracellular matrix expansion and epithelial mesenchymal transition and inflammation (26,36). In our study, the PTECs produced more MCP-1 after AGE-BSA exposure in the autophagy-deficient PTECs than in autophagy-competent PTECs. In addition, mRNA levels of NLRP3, ASC, and caspase-1 and the protein level of IL-1β increased, especially in the diabetic proximal tubule–specific, autophagy-deficient kidney. These results indicate that autophagy suppresses kidney inflammation and fibrosis by upregulation of lysosomal degradation of AGEs.

Considering that AGEs blunt autophagic flux in cultured PTECs, it is easy to infer that AGEs should accumulate even in autophagy-competent cells treated with AGEs, but actually, that has not been the case. Previous reports have suggested that AGEs are directly endocytosed to lysosomes for degradation (10,11). Although AGEs place an inordinate level of stress on the lysosomal systems (which will blunt the autophagic flux), lysosomal function is preserved compared with the autophagy-deficient kidney, which will allow the degradation of AGEs. Although we cannot easily assess autophagic flux in vivo, immunostaining for SQSTM1/p62, the substrate of autophagy, demonstrated that the numbers of SQSTM1/p62-positive dots were comparable between STZ- and vehicle-treated control mice, whereas they significantly increased in STZ-treated Atg5^F/F:KAP mice compared with vehicle-treated Atg5^F/F:KAP mice, indicating that autophagic flux is not completely inhibited in diabetic control mice (Supplementary Fig. 4).

In addition to hyperglycemia, STZ administration induces various physiological changes such as lipid metabolism, which may affect autophagic activity. However, most physiological changes including total cholesterol and triglycerides were comparable between STZ-treated control and Atg5^F/F:KAP mice. Therefore, it is possible that autophagy deficiency per se leads to lysosomal dysfunction and accumulation of AGEs.

Although several agents that reduce or inhibit AGE production have been experimentally or clinically evaluated, most research efforts and trials have failed to clarify clinical efficacy in diabetic nephropathy (3). Our findings suggest that upregulation of autophagic activity and thereby enhancement of upregulation of lysosomal function for AGE degradation could be a promising therapeutic strategy against diabetic nephropathy. The proposed mechanisms by which autophagy inhibits accumulation of AGEs in diabetic nephropathy are depicted schematically (Fig. 7).

In conclusion, autophagy in the proximal tubules of the kidney suppresses the accumulation of AGEs by stimulating the upregulation of lysosomal biogenesis and function and alleviates inflammation and fibrosis in the diabetic state.

Acknowledgments. The authors thank Noboru Mizushima (University of Tokyo, Tokyo, Japan) for donation of the Atg5^F/F mice, Toshimi Michigami (Osaka Medical Center and Research Institute for Maternal and Child Health, Osaka, Japan) for megalin antibody, and Naoko Horimoto and Megumi Kameda (Osaka University, Osaka, Japan) for technical and secretarial assistance.

Funding. This work was supported by the Japan Foundation for Applied Enzymology (to A.T.) and the Manpei Suzuki Diabetes Foundation and Uehara Memorial Foundation (to T.K.).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. A.T., Y.T., and T.K. designed the study. A.T. conducted all experiments and drafted the manuscript. Y.T. edited and revised the manuscript. I.Mae. gave some helpful advice about in vitro experiments. T.K., T.N., T.Ya., J.M., S.M., J.-y.K., and I.Mat. contributed to discussion and reviewed the manuscript. T.M. and F.N. provided mice and commented on the manuscript. T.Yo. and Y.I. supervised this study. Y.T. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.