HGF Protected Against Diabetic Nephropathy via Autophagy-Lysosome Pathway in Podocyte by Modulating PI3K/Akt-GSK3β-TFEB Axis

Background: Podocyte loss is a detrimental feature and major cause of proteinuria in diabetic nephropathy (DN). Our previous study revealed that hepatocyte growth factor (HGF) prevented high glucose-induced podocyte injury via enhancing autophagy. In the current study, we aimed to assess the role of HGF on podocyte homeostasis in DN and clarify its mechanisms further. Methods: Streptozotocin (STZ)-induced DN mice were applied as a model to verify the impact of HGF on the prevention of type 1 diabetes. To study the mechanism involved, the autophagy ux and related signaling pathway upon HGF were examined and the relevance of lysosome acidication in cultured podocytes were investigated. Results: Diabetic mice treated with HGF had markedly reduced ratio of kidney weight to body weight, urinary albumin excretion, podocyte loss and matrix expansion compared with that in the non-treated counterpart. Simultaneously, HGF-treated diabetic mice exhibited increased autophagy activity as indicated by the decreased accumulation of sequestosome 1 (SQSTM1/ p62) and increased microtubule-associated proteins 1 light chains 3 (LC3) II/LC3I ratio. These benecial effects of HGF were blocked by HGF/c-Met inhibitor Crizotinib or phosphatidylinositide 3-kinases (PI3K) inhibitor LY294002. Moreover, HGF treatment obviously prevented inactivation of the protein kinase B (Akt)-glycogen synthase kinase 3 beta (GSK3β)-transcription factor EB (TFEB) axis in high glucose-stimulated podocytes, which was associated with improved lysosome function and autophagy. Accordingly, adenovirus vector encoding constitutively active GSK3β (Ad-GSK3β-S9A) offset whereas small interfering RNA against GSK3β (GSK3β siRNA) recapitulated salutary effects of HGF on lysosome number and autophagy in podocytes. Conclusions: These results suggested that HGF protected against diabetic nephropathy through restoring podocyte /GSK3β in HGF LY294002 these results suggested that HGF enhanced podocyte autophagy via PI3K/Akt-GSK3β pathway. activity Since is a well-recognized pathway of well in the pathogenesis of DN, we suggest that PI3K-GSK3β-TFEB axis could be a potential therapeutic target beyond accounting for the salutary effects of HGF in DN treatment.


Background
Diabetic nephropathy (DN) is a serious complication of diabetes and becomes the leading cause of endstage renal disease worldwide. Depletion of podocytes, which are the critical constituent of the glomerular ltration barrier, initiates and drives the progression of proteinuria towards DN [1]. However, terminally differentiated podocytes have limited capacity of regeneration in response to injury. Autophagy is an intracellular self-repair mechanism that involves the degradation of defective proteins or organelles via lysosomes. Podocytes possess a relatively high basal level of autophagy, which is crucial for its stress adaptation [2]. Therefore, autophagy impairment has been proposed as a major contributor to the pathogenesis of podocytopathy and proteinuria.
Hepatocyte growth factor (HGF) is derived from some mesenchymal cells or tissues with multiple biological effects. Over the years, more and more animal experiments have con rmed that HGF could achieve signi cant prevention against and treatment contribution to DN [3][4][5], however, previous studies more than 15 mmol/l (280 mg/dl) were considered as diabetic. Diabetic mice were then allocated randomly into DN, DN+HGF, DN+HGF+Crizotinib and DN+HGF+LY294002 groups. Rh-HGF (150 μg/kg, i.v.) was given every other day and Crizotinib (20 mg/kg, i.g.) and LY294002 (6 mg/kg, i.p.) or equivalent vehicle were given daily for 4 weeks. All mice were maintained for 12 weeks and the blood glucose level and body weight were monitored throughout the course of treatment. At the end of the study, 24h-urine was collected in the metabolic cages and mice were then killed under chloral hydrate anesthesia to collected blood samples and kidney tissues. Total protein, creatinine and albumin in urine were detected by standard diagnostic kits (Nanjing Jiancheng Bioengineering Institute, Nanjing, China) according to the manufacturer's instructions.

Transmission electron microscopy
Kidney tissues from mice were xed in 2.5% glutaraldehyde overnight and processed for standard scanning at the Electron Microscopy Core Laboratory, School of Basic Medical Sciences, Fudan University. Five glomeruli were randomly selected from each mouse and 3-5 electron micrographs were taken. The glomerular basement membrane (GBM) thickness and the number of foot processes were calculated as described [12].

Histology and immunohistochemistry analysis
The paraformaldehyde-xed and para n-embedded sections of kidney were stained with hematoxylineosin (H&E) and periodic acid-Schiff (PAS) and glomerulosclerotic index was assessed on PAS staining as described previously [13]. For immunohistochemistry staining, the sections were dewaxed and hydrated. The endogenous peroxidase was removed using 3% H 2 O 2 and the antigen was retrieved in citrate buffer. After that, nonspeci c binding sites were blocked with 5% normal goat serum and the sections were incubated with primary antibody at 4 °C overnight (NPHS2/Podocin: 1:500), and secondary antibody in 37 °C for 45 min. The expression was detected by diaminobenzidine and hematoxylin was used as the nuclear counterstain. Images were captured by Nikon camera system (Nikon, Tokyo, Japan) and immunoreactivity was analyzed using Image J software (National Institutes of Health, Bethesda, MD, USA).

Glomerulus isolation and western blotting analysis
The procedure of glomerulus isolation from mice is similar to the isolation from rat described previously [14]. Brie y, the kidney cortex from mice were minced and passed through a system of sieves (BD Falcon, San Diego, CA, USA) with decreasing pore diameters (100, 70, and 40μm) to obtain suspension of glomeruli in PBS. The glomeruli were homogenized in the RIPA lysis buffer (Sangon Biotech, Shanghai, China) containing protease inhibitor cocktail (Selleck, Houston, TX, USA) and vortexed every 10 min for 5 times on ice, which was followed by centrifugation. The protein lysate was measured by BCA protein assay reagent kit (Thermo Piece, Waltham, MA, USA) and used for western blot as described [11]. The primary antibodies were used at the indicated dilution: 1:1000 (LC3B, p-GSK3β, p-Akt, PI3K), 1:2000 (Akt, GSK3β), 1:10000 (p62, GADPH). The bands were quanti ed by densitometry measurement using Gel-pro software (Media Cybernetics, Bethesda, MD, USA).

Cell culture and treatment
Conditionally immortalized podocytes MPC-5 were kindly provided by Professor Peter Mundel (Albert Einstein College of Medicine, New York, NY, USA) and cultured as previously reported in the literature [15].
All experiments were performed with differentiated podocytes, free of mycoplasma infection. HG culture medium was made by supplementing normal glucose RPMI 1640 medium (Thermo Gibco, Waltham, MA, USA) containing 5.5 mmol/l D-glucose with additional D-glucose (Sinopharm, Shanghai, China) for a nal D-glucose concentration at 30 mmol/l. Podocytes cultured in normal or high glucose were stimulated with or without 25 ng/ml HGF in the absence or presence of 2.5 μmol/l SU11274 (S1080, Selleck, Houston, TX, USA) for 24 h. SU11274 was added 1 h before HGF supplementation. After treatment, podocytes grown on 60mm culture dishes (Corning, New York, NY, USA) were harvested and the lysate measurement and western blotting were the same as above.

Immuno uorescence staining of podocytes
Podocytes on the glass sides were xed with 4% paraformaldehyde for 15 min at room temperature. After rinsing twice with PBS, cells were permeabilized with 0.1% Triton (Sinopharm, Shanghai, China) for 5min, incubated with primary antibody (TFEB: 1:50) at 4 °C overnight and secondary antibody at room temperature for 1 h, and counterstained with DAPI. Images were taken using Zeiss Axioplan2 Imaging System (Zeiss, Oberkochen, Germany).

Lysosome staining and ow cytometric analysis
The acidi cation of lysosomes was determined with Lysosome Staining Kit (E607506, Sangon Biotech, Shanghai, China) according to the manufacturer's protocol. Cultured podocytes were incubated with LysoBrite Red for 2 h. The cells were then washed with PBS to remove any excess lysosomal marker, and the live cell images were taken using the EVOS uorescence microscope (Life technologies, Grand Island, NY, USA). For uorescence quanti cation, the podocytes were digested by trypsin (Thermo Gibco, Waltham, MA, USA) and resuspended with PBS. The uorescent intensity value of LysoBrite Red was quanti ed by ow cytometry using BD LSRFortessa (BD Biosciences, San Diego, CA, USA). The MFI (mean uorescent intensity) of the uorescence was calculated and presented as a histogram photograph by FlowJo software (Tree Star Inc., Ashland, OR, USA).

Statistical analysis
The experiments were all independently repeated at least three times. Results were expressed as means ± SEM. Differences between two and more groups were performed by means of the unpaired t test and ANOVA, respectively. Statistical analysis was conducted using the GraphPad Prism software (GraphPad, La Jolla, CA, USA). P<0.05 was considered to indicate statistically signi cant difference.

HGF alleviated podocyte injury and improved kidney function in STZ-induced DN mice
To validate the effect of HGF on podocytes in DN, we rst injected rh-HGF into Streptozotocin (STZ)induced diabetic mice. As shown in Supplementary gure, all the STZ groups had hyperglycemia and lower body weight compared with the control counterparts. Firstly, we showed decreased expression of NPHS2/Podocin, a podocyte-speci c transmembrane protein, in renal biopsies from DN patients by immunohistochemical detection, suggesting the occurrence of podocyte injury (Fig. 1a). Secondly, we found lower expression of this marker in STZ-induced DN mice than in the control group. Moreover, HGF treatment signi cantly prevented the loss of podocytes, which could be abrogated by Crizotinib (Criz), a speci c inhibitor of HGF receptor. Consistently, matrix expansion and glycogen accumulation assessed by periodic acid-Schiff (PAS) staining were shown in the mesangial region of nontreated and HGF+Criztreated DN mice but not in HGF-treated diabetic mice and nondiabetic control mice (Fig. 1b). Transmission electron microscopy (TEM) analysis further revealed thicker glomerular basement membrane (GBM) and much more abnormal changes of podocytes such as foot process effacement and vacuolization in DN or HGF+Criz treated mice (Fig. 1c). As an index of renal function, urinary albumin to creatinine ratio (UACR) was measured and expectedly higher levels of UACR was observed in DN or HGF+Criz treated mice relative to normal and HGF-treated mice (Fig. 1d). The ratio of kidney to body weight, another parameter indicating renal hypertrophy, was also increased in DN mice but markedly decreased in HGF-treated DN mice (Fig. 1e).
HGF restored podocyte autophagy in STZ-induced DN mice Autophagy de ciency is a well-recognized contributor to podocyte injury and proteinuria in DN [16]. We examined if autophagy regulation would be one of the mechanisms by which HGF reduces renal damage.
Consistent with other studies, double staining of immuno uorescence with autophagy related proteins and Synaptopodin (Synapt), one of the podocyte cytoskeleton proteins, located fewer dots of LC3 ( Fig.  2a) and more accumulated SQSTM1/p62 (Fig. 2b) in podocytes from DN patients compared with the control subjects, indicating impaired autophagy ux. Similar results were also obtained in STZ-induced DN mice tissue (Fig. 2c,d). Protein of renal cortex was collected, and western blotting analysis displayed decreased ratio of LC3II/LC3I and increased p62 expression (Fig. 2e). Of note, we found that HGF signi cantly recovered the level of LC3 and relatively suppressed p62 expression, which could be almost completely abolished by HGF inhibitor Crizotinib (Fig. 2c-e). These results indicated that restoring podocyte autophagy was the relevant mechanism underlying the salutary effects of HGF in DN described above.
HGF-restored podocyte autophagy in STZ-induced DN mice involved Akt-GSK3β signaling pathway As a vital component of glycogen metabolism, GSK3β has been con rmed as a downstream target of phosphoinositide 3-kinase (PI3K) /Akt pathway [17]. Some studies proposed that GSK3β activity is inversely correlated with autophagy[18-20], and our previous research also veri ed HGF promoted autophagic ux by inhibition of high glucose-reduced Ser9 phosphorylation of GSK3β, an indicator of GSK3β activation [11]. In this study, we logically hypothesized and examined whether HGF played the role through the Akt-GSK3β signaling pathway in vivo. Immunoblotting (Fig. 3a) or double immuno uorescence staining (Fig. 3b) showed decreased ratio of p-Akt(ser473) /Akt and p-GSK3β (ser9) /GSK3β in DN mice group compared with control group. HGF treatment nearly normalized the phosphorylation of Akt and GSK3β to the level in control group, which was e ciently blocked by HGF inhibitor Crizotinib or PI3K/Akt inhibitor LY294002 (LY). Taken together, these results suggested that HGF enhanced podocyte autophagy via PI3K/Akt-GSK3β pathway.

HGF improved lysosome function in HG-stimulated mouse podocytes
Lysosomes-mediated intracellular degradation is the major compartment in autophagy activation. Dysfunctional lysosomes participate in podocytes damage, leading to glomerular disease and proteinuria [16]. Thus, we further examined whether correcting lysosome dysfunction could account for HGF-enhanced autophagy ux in HG-stimulated cultured podocytes. Lysosome staining reagent was used to label and track the acidic intracellular compartments (lysosomes) in live cells, and the uorescence intensity increases with the decrease of pH. As shown in Figure 4a and 4b, live cell imaging and quantitative analysis by ow cytometry displayed declined uorescence intensity in HG-stimulated mouse podocytes in comparison to the normal glucose (NG) group, indicating weakened lysosomal acidi cation. However, HGF co-treatment partially recovered lysosomal acidi cation and HGF inhibitor SU11274 antagonized this effect. Transcription factor EB (TFEB), as the master regulators of lysosome biogenesis, is dephosphorylated followed by translocation to the nucleus to activate the transcription of genes responsible for coordinated lysosomal expression and regulation [21]. High glucose sequestered more TFEB granules in cytoplasm while pre-HGF treatment caused more TFEB granules entry into the nucleus as evidenced by distribution of uorescence intensity. HGF inhibitor SU11274 completely blocked the effect of HGF on nuclear translocation of TFEB (Fig. 4c). Together, these results implied that HGF played an important role in maintaining lysosome function in HG-stimulated podocytes.
Overexpression of GSK3β abrogated the effect of HGF on lysosome function in HG-stimulated mouse podocytes Several studies have revealed negative effect of GSK3β on autophagy may be relevant with lysosomes [19,22,23]. Based on the ndings described above, we speculated that GSK3β activity regulation may be attributed to the improved lysosome function by HGF. Podocytes were transfected with adenovirus vector encoding the constitutively active GSK3β (Ad-S9A), dominant-negative GSK3β (Ad-K85A), or wild-type GSK3β (Ad-WT) plasmid and then subjected to HG+HGF treatment. The virus carrying GFP vector was used as the control transfection treatment, and successful adenoviral transfection of GSK3β was corroborated by the immunoblotting for hemagglutinin (HA, the tag of GSK3β adenovirus vector) or increased GSK3β protein expression (Fig. 5a). Lysosome red uorescence intensity greatly reduced in Ad-GSK3β-S9A but not in Ad-GSK3β-WT and Ad-GSK3β-K85A group (Fig. 5b,c). In addition, the nuclear translocation of TFEB was also depressed in GSK3β transfection groups, with the strongest cytosolic TFEB expression being observed in Ad-GSK3β-S9A group (Fig. 5d).
Knockdown of GSK3β partially recapitulated the effect of HGF on lysosome function in HG-stimulated mouse podocytes To further verify the role of GSK3β, podocytes were transfected by small interfering RNA against GSK3β (GSK3β siRNA) before exposure to HG or HG+HGF (Fig. 6a). As expected, knocking down GSK3β increased lysosome uorescence intensity (Fig. 6b,c), as well as promoted TFEB translocation (Fig. 6d). Moreover, HGF treatment in the Ctrl siRNA group certainly enhanced the lysosome acidi cation and function, however, there was no signi cant difference between HG+GSK3β siRNA group and HG+HGF+GSK3β siRNA group regarding lysosome uorescence intensity and TFEB translocation ( Fig.   6b-d), indicating the silencing of GSK3β replaced the inhibitory effect of HGF on GSK3β and thus highlighting the importance of targeting GSK3β for DN treatment. All these data in Figure 5 and 6 illustrated that GSK3β activity modi ed by Ser9 phosphorylation status was essential for improved lysosome function and autophagy by HGF in HG-stimulated podocytes.

Discussion
As the essential mechanism of maintaining podocytes homeostasis, autophagy is suppressed by excess nutrition or energy in diabetes mellitus state with the results of making podocytes more prone to dysfunction and apoptosis [16]. Consistently, we observed podocyte injury, glomerulosclerosis and proteinuria in STZ-induced type 1 DN mouse model, with decreased LC3 (LC3II/LC3I) and increased SQSTM1/p62 expression indicative of impaired autophagy ux. Although there was no difference in blood glucose, HGF treatment had signi cant improved in all aforementioned histological and functional parameters of kidney in DN model, accompanied by restored autophagy, proving that the protective effect of HGF against DN may be achieved by normalizing autophagy level in podocytes without affecting hyperglycemia. This in vivo data is concordant with our previous in vitro study reporting the favorable effects of HGF on high glucose-treated podocyte [11]. Furthermore, in line with our data, Peng KY et al.
demonstrated that HGF had a role in the amelioration of diabetic vascular complications via autophagic clearance of advanced glycation end (AGE) products [24]. Since numerous clinical trials revealed that patients subjected to intensive glycemic control gained little improvement on preservation of renal function compared with patients receiving less aggressive glucose control, the treatment strategy independent of glucose lowering is promising like sodium-glucose cotransporter (SGLT) inhibitors.
As an important regulator of glycogen metabolism, renal expression of total and activated GSK3β was elevated in diabetic patients and mice [25]. Mechanistically, enhanced GSK3β activity was associated with podocyte apoptosis in DN [8]. In our study, expression of p-Akt (ser473) and p-GSK3β (Ser9) in podocytes in DN mice were reduced, which indicated lowered Akt activity and elevated GSK3β activity, respectively. However, these changes in Akt-GSK3β phosphorylation status were remarkably diminished in podocytes of HGF-treated DN mice. Given that GSK3β has been recognized as a downstream signaling molecule of the HGF/c-Met-Akt pathway, we reasoned that HGF could bene t podocytes through the PI3K/Akt-GSK3β signaling pathway in diabetic milieu. As a multifunctional kinase, GSK3β also plays a role in the regulation of autophagy, and the therapeutic effect of GSK3β inhibitors is linked to activated/enhanced autophagy in various models [18,[26][27][28]. In this study, HGF treatment lessened GSK3β activity in parallel with restored autophagy ux in podocyte of diabetic mice, which could be blocked by co-treatment of HGF or PI3K/Akt inhibitors. Therefore, consistent with our previous study in vitro, we con rmed that HGF could promote podocyte autophagy by suppression of GSK3β activity in vivo.
Then, we raised the question of how GSK3β affects autophagy. Azoulay-Alfaguter I et al. found GSK3 reduced lysosomal acidi cation (where GSK3β has a stronger effect than GSK3α) in breast cancer cells [19]. Avrahami L et al. and Parr C et al. con rmed the inhibition of GSK3 promoted lysosomal biogenesis and autophagic degradation, which reduced the accumulation of β-amyloid in Alzheimer's disease mouse model [22,29]. Therefore, we speculated lysosome function might be in uenced by the change of GSK3β during HGF treatment in DN. As we expected, HGF partially recovered the defective lysosomal acidi cation in immortalized mouse podocytes with HG stimulation. However, overexpression of GSK3β weakened the restoration of lysosome acidi cation by HGF. Moreover, the extent of weakening was correlated to the activity pattern of overexpressed GSK3β with constitutively active GSK3β being the strongest and dominant dead GSK3β the weakest. When GSK3β is interfered in the opposite direction, that is GSK3β silence by siRNA, the effect of HGF on lysosome acidi cation was recapitulated as demonstrated by the nding that there was no difference in lysosome acidi cation between HG+siGSK3β and HG+HGF group. Furthermore, lysosomal acidi cation level in HG+HGF+siGSK3β group was comparable with that in HG+siGSK3β group, suggesting that GSK3β is a pivotal downstream target of HGF to in uence lysosome function. We also found GSK3β expression in HG+HGF group was numerically reduced as compared with HG group but it did not reach statistical signi cance.
TFEB is one of the major transcription factors related to autophagy. Studies have reported reduced TFEB phosphorylation by GSK3 inhibitors caused more TFEB nuclear translocation and promoted lysosomal function [20,29,30]. Therefore, we also explored the localization of TFEB in this study. Immuno uorescence staining demonstrated HG made more TFEB sequestered in cytoplasm, which was mostly abolished by HGF. This HGF-promoted TFEB nuclear translocation under HG conditions was blocked by HGF inhibitor SU11274 or partly abrogated by overexpression GSK3β, especially constitutively active GSK3β-S9A. Upon GSK3β silenced by siRNA, TFEB nuclear translocation was enhanced, albeit to a lesser extent in HG group in comparison to HG+HGF group, pointing at mechanisms other than GSK3βdependent signaling pathway whereby HGF could in uence TFEB location. Considering that TFEB entry into nucleus is instrumental to lysosome function, we proposed HGF improved lysosome function and then accelerated autophagic ux in podocytes exposed to HG stimulation via modulation of GSK3β-TFEB axis. In accordance with our ndings, Zhao X et al. con rmed TFEB inhibition mediated the AGEstimulated autophagy impairment in cultured podocytes and db/db mice [31]. As a multifunctional transcription factor, TFEB can be regulated by multiple serine/threonine kinases, driving both expression of autophagy and lysosomal genes to coordinate the autophagy-lysosome degradation system[21].
Overexpression of TFEB promoted the elimination of storage factors in some accumulation or metabolic diseases[32-34]. It warrants further study in future that TFEB may be a promising therapeutic target both in DN and for favorable effects of HGF.

Conclusions
In summary, we show in this study that HGF protects against DN through restoring podocyte autophagy. The mechanism is partly due to improved lysosome function that correlates with enhanced TFEB nuclear translocation and GSK3β activity inhibition. Since PI3K/Akt is a well-recognized signaling pathway of HGF as well as in the pathogenesis of DN, we suggest that PI3K-GSK3β-TFEB axis could be a potential therapeutic target beyond accounting for the salutary effects of HGF in DN treatment.

Consent for publication
Not applicable.

Availability of data and materials
All data generated or analyzed during this study are included in this article.

Competing interests
The authors declare that there is no potential con ict of interest.

Funding
This study was supported by grants from the National Natural Science Foundation of China (NSFC 81370796).
Authors' contributions BH and HL were responsible for the study design, acquisition of data, interpretation of results and preparation and revision of the manuscript. YKL, XL, CYZ, ZHZ and QC contributed to the acquisition and analysis of data. NZ and HL contributed to the study design, interpretation of data and manuscript preparation. All authors were involved in drafting and editing the article, and all authors gave their approval for the nal manuscript to be published. HL is the guarantors of this work. All authors read and approved the nal manuscript.