Protective effect of the tunneling nanotube-TNFAIP2/M-sec system on podocyte autophagy in diabetic nephropathy

ABSTRACT

Podocyte injury leading to albuminuria is a characteristic feature of diabetic nephropathy (DN). Hyperglycemia and advanced glycation end products (AGEs) are major determinants of DN. However, the underlying mechanisms of podocyte injury remain poorly understood. The cytosolic protein TNFAIP2/M-Sec is required for tunneling nanotubes (TNTs) formation, which are membrane channels that transiently connect cells, allowing organelle transfer. Podocytes express TNFAIP2 and form TNTs, but the potential relevance of the TNFAIP2-TNT system in DN is unknown. We studied TNFAIP2 expression in both human and experimental DN and the renal effect of tnfaip2 deletion in streptozotocin-induced DN. Moreover, we explored the role of the TNFAIP2-TNT system in podocytes exposed to diabetes-related insults. TNFAIP2 was overexpressed by podocytes in both human and experimental DN and exposre of podocytes to high glucose and AGEs induced the TNFAIP2-TNT system. In diabetic mice, tnfaip2 deletion exacerbated albuminuria, renal function loss, podocyte injury, and mesangial expansion. Moreover, blockade of the autophagic flux due to lysosomal dysfunction was observed in diabetes-injured podocytes both in vitro and in vivo and exacerbated by tnfaip2 deletion. TNTs allowed autophagosome and lysosome exchange between podocytes, thereby ameliorating AGE-induced lysosomal dysfunction and apoptosis. This protective effect was abolished by tnfaip2 deletion, TNT inhibition, and donor cell lysosome damage. By contrast, Tnfaip2 overexpression enhanced TNT-mediated transfer and prevented AGE-induced autophagy and lysosome dysfunction and apoptosis. In conclusion, TNFAIP2 plays an important protective role in podocytes in the context of DN by allowing TNT-mediated autophagosome and lysosome exchange and may represent a novel druggable target.

Abbreviations: AGEs: advanced glycation end products; AKT1: AKT serine/threonine kinase 1; AO: acridine orange; ALs: autolysosomes; APs: autophagosomes; BM: bone marrow; BSA: bovine serum albumin; CTSD: cathepsin D; DIC: differential interference contrast; DN: diabetic nephropathy; FSGS: focal segmental glomerulosclerosis; HG: high glucose; KO: knockout; LAMP1: lysosomal-associated membrane protein 1; LMP: lysosomal membrane permeabilization; MAP1LC3/LC3: microtubule-associated protein 1 light chain 3; PI3K: phosphoinositide 3-kinase; STZ: streptozotocin; TNF: tumor necrosis factor; TNFAIP2: tumor necrosis factor, alpha-induced protein 2; TNTs: tunneling nanotubes; WT: wild type.

Introduction

Diabetic nephropathy (DN), a chronic complication of diabetes, is the leading cause of end-stage renal disease (ESRD) in the western world. The complication is characterized by both albuminuria and progressive renal function loss [1,2]. These functional alterations are primarily due to both excessive accumulation of extracellular matrix in the mesangium and podocyte damage.

Podocytes form the glomerular filtration barrier together with the glomerular basement membrane and the fenestrated glomerular endothelium. The junction between adjacent podocyte foot processes (FP), named slit diaphragm (SD), is the major restriction site to protein filtration in the glomeruli. Downregulation of SD proteins, such as NPHS1/Nephrin and NPHS2/Podocin, FP effacement, and podocyte apoptosis are early features of DN and major determinants in the development of albuminuria [3–6].

Both hyperglycemia and advanced glycation end products (AGEs) play an important role in the pathogenesis of the podocyte injury in DN by inducing both oxidative stress and inflammation [7–9]. In addition, dysfunction of intracellular organelles also contributes to podocyte damage [10–15]. However, the mechanisms underlying podocyte impairment in DN are not fully understood. Moreover, most studies focussed on mechanisms of podocyte damage and relatively little is known on counteracting mechanisms of cytoprotection.

Tunneling nanotubes (TNTs) are open-ended straight membrane channels (20–500 nm) without contact to the substrate that interconnect cells over long distances [16–20]. These intercellular bridges have been demonstrated in a variety of cell types in vitro [21–26] and more recently in vivo [27–31]. Cells form TNTs de novo in response to various stresses, including serum deprivation, hypoxia, oxidative stress, and inflammation [24,32–38]. TNTs allow horizontal intercellular transfer of cellular components, including small organelles, such as mitochondria, endosomes, and lysosomes, to target cells [39–44]. Replacement of damaged organelles via TNTs increases resistance of recipient cells to injury.

The protein TNFAIP2/M-Sec (tumor necrosis factor, alpha-induced protein 2) plays a crucial role in TNT formation by interacting with the exocyst complex and triggering polymerization of F-actin [45–47]. Expression of TNFAIP2 is observed predominantly in immune, endothelial, and cancer cells, and it is enhanced by both retinoic acid and inflammatory cytokines, including TNF/TNF-α [48–53]. There is little information on the TNFAIP2-TNT system in the kidney. However, podocytes can express TNFAIP2 and their exposure to either serum deprivation or adriamycin induces a TNFAIP2-TNT-dependent transfer of mitochondria from healthy to injured podocytes [54]. In addition, in vivo deletion of Tnfaip2 causes the spontaneous development of focal segmental glomerulosclerosis (FSGS) in the susceptible BALB/c mouse strain [54].

Whether diabetes-induced cellular stress activates the TNFAIP2-TNT system in podocytes and if this system confers protection to podocytes by organelle exchange has not yet been explored.

Therefore, in the present study, we assessed podocyte TNFAIP2 expression in both human and experimental DN, investigated the effect of tnfaip2 deletion in experimental DN, and explored in vitro in diabetes-injured podocytes the mechanism of TNFAIP2-TNT-mediated cytoprotection.

Results

TNFAIP2 expression in human and experimental DN

TNFAIP2 expression was studied in renal biopsies obtained from patients with DN and non-diabetic control subjects (Table 1). Glomerular immunostaining for TNFAIP2 was fourfold greater in patients with DN compared to controls (Figure 1(A)). The pattern of staining suggested a predominant podocyte distribution, as confirmed by colocalization of TNFAIP2 and the podocyte marker SYNPO (synaptopodin) in double immunofluorescence (Figure 1(B)). Of interest, an increased TNFAIP2 expression was also observed in kidney biopsies from patients with microalbuminuria.

Table 1.

Clinical parameters of patients with diabetic kidney disease (DKD) and non-diabetic control subjects.

	Controls	DKD
N	8	15
Age (years)	65.4 ± 6.3	66.5 ± 2.8
Sex (male/female)	6/2	11/4
Diabetes duration (years)	-	19.5 ± 0.7
Systolic blood pressure (mmHg)	120 ± 2.2	143 ± 4.6
Diastolic blood pressure (mmHg)	65 ± 3.1	83 ± 4.2
Serum creatinine (mg/dl)	1 ± 0.1	2.1 ± 0.4
Proteinuria (g/24 h)	-	2.1 ± 0.5
Hypertension (%)	-	100

Data are expressed as mean ± SEM

Figure 1.

TNFAIP2 expression in both human and experimental diabetic nephropathy. (A) Glomerular TNFAIP2 protein expression was assessed by immunohistochemistry in renal cortex sections from control subjects (n = 8) and patients with either incipient (DN-micro, n = 3) or overt diabetic nephropathy (DN-macro, n = 12) (original magnification 200X, scale bar: 100 µm). Focal segmental glomerulosclerosis (FSGS) sections were used as a positive control. The percent area of positive staining is shown in the graph (***p < 0.001 DN vs. control). (B) Double immunofluorescence for TNFAIP2 and the podocyte marker SYNPO (synaptopodin) was carried out on renal sections from DN patients. The merged image shows colocalization (original magnification 200X, scale bar: 100 µm). (C) Representative immunohistochemistry images of glomerular TNFAIP2 protein expression in renal cortex sections from non-diabetic (ND) and diabetic mice (DM) after 12 weeks of diabetes (magnification 400X, scale bar: 50 µm). (D) The graph shows the percent area of glomerular TNFAIP2-positive staining (n = 5 mice per group; ***p < 0.001 DM vs. ND). (E) Tnfaip2 mRNA levels were measured in the glomeruli isolated from ND and DM mice by real time-PCR and results corrected for the expression of the housekeeping gene Hprt (n = 5 mice per group; ***p < 0.001 ND vs. DM). (F) Staining for NPHS2/Podocin and TNFAIP2 on serial renal cortical sections obtained from DM mice showed a podocyte distribution (magnification 400X, scale bar: 50 µm). (G) Representative immunoblotting of TNFAIP2 in cultured podocytes exposed to high glucose concentrations (HG) for 48 h with and without the PI3K inhibitor, LY294002 (LY). TUBB/Tubulin was used as a loading control. (H) Results of densitometry analysis are shown in the graph (n = 3; ***p < 0.001 HG vs. others). (I and J) Immunoblottings showing TNFAIP2 expression in podocytes exposed to AGEs, mechanical stretch (10% elongation) or vehicle (V). Results of densitometry analyses are shown in the graphs (p = ns).

Consistent with this, a significant increase in both immunostaining for TNFAIP2 and Tnfaip2 mRNA levels was observed in the glomeruli from C57BL/6 mice after 12 weeks of streptozotocin (STZ)-induced diabetes (Figure 1(C-E)). Moreover, podocytes were the main cell type overexpressing TNFAIP2 within the diabetic glomeruli, as shown by immunostaining for TNFAIP2 and NPHS2 in serial cortex sections (Figure 1(F)). Of interest, a significant increase in Tnfaip2 mRNA levels was also observed in isolated glomeruli in a very early stage (6 weeks) of experimental diabetes (control animals: 0.57 ± 0.07; diabetic mice: 1.19 ± 0.09; p < 0.01).

Effect of diabetes-related insults on the TNFAIP2-TNT system in cultured podocytes

To explore the underlying mechanism of TNFAIP2 overexpression, cultured podocytes were exposed to diabetes-related insults. Podocyte incubation in a high glucose (HG) (25 mM) milieu induced an over twofold increase in TNFAIP2 expression via a PI3K (phosphoinositide 3-kinase)-dependent mechanism (Figure 1(G, H)), while no changes were observed in response to AGE-bovine serum albumin (BSA) and mechanical stretch, mimicking glomerular capillary hypertension (Figure 1(I, J)). HG-induced PI3K-dependent Tnfaip2 expression occurred via a transcriptional mechanism as PI3K inhibition using LY294002 (LY) also prevented Tnfaip2 mRNA expression (NG: 1.0 ± 0.01; NG + LY: 1.14 ± 0.09; HG: 2.81 ± 0.52; HG + LY: 1.24 ± 0.26).

Given the key role of TNFAIP2 in TNT formation, we next investigated if diabetes-related insults can also promote TNT formation. Exposure to either HG or AGE-BSA caused formation of TNT-like structures that fulfilled the morphological criteria of TNTs, including the lack of contact to the substrate and the presence of an actin backbone (Figure 2(A-D)). Importantly, both HG and AGE-BSA increased the number of podocytes connected by TNTs in a PI3K-AKT1-dependent and TP53/TRP53/P53-independent manner (Figure 2(E, F)). Tnfaip2 silencing completely prevented diabetes-induced TNT formation (Figure 2(G-I)), confirming that TNFAIP2 is required for TNT development.

Figure 2.

The TNFAIP2-TNT system in cultured podocytes exposed to diabetes-related insults. (A) Podocytes pre-exposed to high glucose concentrations (HG) for 48 h were stained with wheat germ agglutinins (WGA) Alexa Fluor 488 to reveal TNTs. A representative image showing a TNT-like channel, interconnecting two podocytes, is shown (magnification 630X, scale bar: 50 µm). (B) Serial Z-stack images, acquired with a step-size of 0.25-µm proved that TNTs did not adhere to the substrate. In panel C, colors represent the Z-depth (depth coding; red: bottom, blue: top). (D) F-actin was labeled with Cell-Light Actin-RFP to show that TNTs have an actin backbone (magnification 630X, scale bar: 50 µm). (E and F) Podocytes were exposed to normal (NG) or high glucose (HG) (E) and vehicle or AGEs (F) with and without inhibitors of AKT1 (AKT1-VIII inhibitor (AKT1i); 1 µM), PI3K (LY-294002 (LY); 50 µM), and TP53 (p-nitro-pifithrin (TP53i); 5 µM), stained with WGA Alexa Fluor 488, and analyzed by fluorescent live cell microscopy. The graphs show the percentage of podocytes connected by one or more TNTs (n = 3; ***p < 0.001 HG and HG+p53i vs. others; **p < 0.01 AGEs and AGEs+p53i vs. others). (G) Murine podocytes were transfected with either Tnfaip2 shRNA (tnfaip2^−/−) or a mock plasmid (Tnfaip2^+/+). Knockdown efficiency was assessed by immunoblotting and TUBB/tubulin used as internal control. (h and i) tnfaip2^−/− and Tnfaip2^+/+ podocytes were exposed to (H) vehicle or AGEs and (I) NG or HG for 48 h, stained with WGA Alexa Fluor 488, and analyzed by fluorescent live cell microscopy to reveal TNTs. The graphs show the percentage of podocytes connected by one or more TNTs (***p < 0.001 Tnfaip2^+/+ HG and Tnfaip2^+/+AGEs vs. others).

Diabetes induction in tnfaip2-knockout (KO) C57BL/6 mice

To assess the functional relevance of TNFAIP2 in vivo, we knocked out Tnfaip2 in C57BL/6 mice, a strain that is susceptible to STZ-induced diabetes, but resistant to FSGS. Homozygous tnfaip2 KO C57BL/6 mice were viable, born at normal Mendelian ratios, and grew normally. There were no differences in albuminuria, renal function, and glomerular structure between tnfaip2 KO and wild type (WT) animals even after long term follow-up (26 weeks of age). Therefore, tnfaip2 deletion per se did not induce a renal phenotype in this strain (Fig. S1).

We then induced diabetes by STZ administration in both WT and tnfaip2 KO C57BL/6 mice. Twelve weeks after diabetes onset, diabetic WT and tnfaip2 KO mice had comparable values of blood glucose, glycated hemoglobin, body weight, kidney weight-to-body weight ratio, and systolic blood pressure. tnfaip2 deletion did not affect albuminuria in non-diabetic mice, but significantly enhanced the rise in ALB (albumin):creatinine ratio (ACR) induced by diabetes. Furthermore, a significant reduction in creatinine clearance was exclusively observed in diabetic tnfaip2 KO mice (Table 2), indicating that Tnfaip2 abrogation worsened functional parameters of DN.

Table 2.

Metabolic and physiological parameters in WT and tnfaip2 KO mice.

	ND-WT	ND-tnfaip2 KO	DM-WT	DM-tnfaip2 KO
Body weight (g)	30.20 ± 0.97	28.92 ± 1.73	21.65 ± 1.31a	20.80 ± 0.86 a
Blood glucose (mg/dl)	136.20 ± 2.80	150.70 ± 8.30	403.30 ± 40.1 a	382.50 ± 23.90 a
Glycated Hb (%)	4.75 ± 0.15	5.12 ± 0.10	10.92 ± 0.55 a	11.41 ± 0.33 a
SBP (mmHg)	108.00 ± 3.20	105.00 ± 3.10	103.00 ± 1.30	106.00 ± 2.50
KW:BW ratio	6.61 ± 0.32	5.45 ± 0.21	7.57 ± 2.11b	7.45 ± 0.42b
ACR (µg/mg)	36.80 ± 7.70	28.80 ± 7.90	68.20 ± 14.00a	144.10 ± 36.50c
CrCl (ml/min)	0.43 ± 1.37	0.40 ± 3.20	0.31 ± 3.74	0.17 ± 0.050d

Data are expressed as mean ± SEM; ND: non-diabetic; DM: diabetic; KW:BW: kidney weight:body weight; SBP: systolic blood pressure; ACR: ALB (albumin)-to-creatinine ratio; CrCl: creatinine clearance; ^ap<0.001; ^bp<0.05 DM groups vs. ND groups; ^cp<0.05 DM-tnfaip2 KO vs. DM-WT; ^dp<0.05 DM-tnfaip2 KO vs. others.

Structural abnormalities in diabetic tnfaip2 KO C57BL/6 mice

Histological assessment by periodic acid Schiff (PAS) staining revealed a mild mesangial expansion in diabetic mice. The degree of glomerular injury was greater in diabetic tnfaip2 KO mice with a more prominent mesangial expansion, as assessed by both light and electron microscopy (Figure 3(A-C)). Furthermore, diabetes-induced overexpression of both FN1 (fibronectin 1) and the prosclerotic cytokine Tgfb1 (transforming growth factor, beta 1) was enhanced in mice lacking Tnfaip2 (Figure 3(D-F)).

Figure 3.

Effect of tnfaip2 deletion on diabetes-induced glomerular structural abnormalities, markers of fibrosis, and inflammation. Renal cortex samples from diabetic (DM) and non-diabetic (ND) both WT (ND-WT, DM-WT) and tnfaip2 KO (ND-KO, DM-KO) mice were studied 12 weeks after diabetes onset. (A) Representative periodic acid Schiff (PAS) staining images (magnification 400X, scale bar: 50 µm) and (B) quantification of glomerulosclerosis are shown (n = 6 mice per group; ***p < 0.001 DM-WT vs. ND-WT; DM-KO vs. DM-WT). (C) Electron microscopy images showing a more prominent mesangial expansion in the glomeruli from DM-KO mice compared to DM-WT mice (magnification 3200X). (D) Immunofluorescence images of glomerular FN1 (fibronectin 1) are shown (magnification 400X, scale bar: 50 µm). (E) Quantification of the percent area of glomerular FN1-positive staining (n = 6 mice per group; ***p < 0.001 DM-WT vs. ND-WT; DM-KO vs. DM-WT). (F) Tgfb1 mRNA levels were measured by real-time PCR in total renal cortex and corrected for the expression of the housekeeping gene Hprt (n = 5 mice per group; *p < 0.05 DM-WT vs. ND-WT; DM-KO vs. DM-WT). (G) Glomerular macrophage accrual was evaluated by counting the number of LGALS3/MAC2-positive cells per glomerulus (n = 6 mice per group; ***p < 0.001 DM-WT vs. ND-WT; DM-KO vs. DM-WT). mRNA levels of Ly6c2 (H), Ccl2 (I), Ccr2 (J), and Tnf (K) were measured in the renal cortex by real time-PCR and the results corrected for the expression of the housekeeping gene Hprt (n = 5 mice per group; *p < 0.05 DM-WT vs. ND-WT; **p < 0.01 DM-KO vs. DM-WT). (L) Schematic illustration of the protocol used for generating diabetic (DM) tnfaip2 KO chimeric mice. Recipient tnfaip2 KO mice were lethally irradiated, and then reconstituted with bone marrow (BM) from either WT (DM-KO-c^WT) or KO (DM-KO-c^KO) mice. (M) PCR genotyping of peripheral blood cells from chimeric animals. M: marker; NC: no template control. (N) ALB (albumin) excretion rate (AER) was measured after 12 weeks of diabetes in transplanted animals. (*p < 0.05 DM-WT-c^WT vs. ND-WT-c^WT; DM-KO-c^WT and DM-KO-c^KO vs. DM-WT-c^WT. (O) Representative images of PAS staining of renal cortex sections from transplanted animals (magnification 400X, scale bar: 50 µm).

Glomerular inflammation in diabetic tnfaip2 KO mice and chimeric animals

Diabetes induced glomerular infiltration of LGALS3/MAC2⁺ (lectin, galactose binding, soluble 3) macrophages and an increased expression of Ly6c2 (lymphocyte antigen 6 complex, locus C2) a marker of inflammatory monocytes. Moreover, overexpression of Ccl2/MCP-1 (chemokine (C-C motif) ligand 2) and its cognate receptor Ccr2 (chemokine (C-C motif) receptor 2) was also observed, indicating activation of the CCL2-CCR2 system that play a key role in monocyte accrual in the diabetic glomeruli. These proinflammatory effects were magnified in diabetic tnfaip2 KO mice (Figure 3(G-J)). Surprisingly, expression of Tnf, which is known to induce TNFAIP2, was not affected by either diabetes or tnfaip2 deletion (Figure 3(K)).

Although TNFAIP2 is expressed by inflammatory cells, our study in chimeric animals showed that transplantation of bone marrow (BM) cells of WT mice in diabetic tnfaip2 KO mice did not ameliorate albuminuria and mesangial expansion (Table 3 and Figure 3(L-O)). This indicates that lack of Tnfaip2 in BM cells did not play a primary role in explaining worsening of DN in tnfaip2 KO mice. Genotyping confirmed reconstitution of Tnfaip2 in peripheral blood cells of chimeric tnfaip2 KO mice transplanted with BM from WT animals (Figure 3(M)).

Table 3.

Metabolic and physiological parameters of chimeric animals.

	ND-WT-cWT	ND-KO-ctnfaip2 KO	DM-WT-cWT	DM-KO-ctnfaip2 KO	DM-KO-cWT
BW (g)	26.02 ± 0.33	26.42 ± 0.51	22.19 ± 0.83a	21.77 ± 0.75a	22.85 ± 1.52a
BG (mg/dl)	124.2 ± 3.64	123.4 ± 3.07	381.0 ± 1.15b	397.0 ± 11.70b	384.0 ± 34.90b
Glycated Hb (%)	4.82 ± 0.19	5.06 ± 0.12	10.48 ± 0.25b	10.64 ± 0.26b	10.54 ± 0.31b
SBP (mmHg)	114.20 ± 5.27	111.20 ± 4.59	108.80 ± 1.15	112.60 ± 1.60	111.4 ± 1.77
KW:BW ratio	5.85 ± 0.01	5.99 ± 0.01	7.29 ± 0.56c	7.40 ± 0.10c	7.30 ± 0.64c
AER (µg/18 h)	19.60 (18.80–19.10)	22.12 (20.20–22.30)	51.46d (34.90–66.00)	87.63e (77.60–107.90)	84.87e (64.30–117.20)

Data are expressed as mean ± SEM or geometric mean (25°-75° percentile); ND; non-diabetic; DM: diabetic; SBP: Systolic Blood Pressure; BW: body weight; BG: blood glucose; KW:BW: kidney weight:body weight; AER: ALB (albumin) Excretion Rate; ^ap<0.05 ^bp<0.001; ^cp<0.01 DM groups vs. ND groups; ^dp<0.05 DM-WT-c^WT vs. ND groups; ^ep<0.05 DM-KO-c^{tnfaip2 KO} and DM-KO-c^WT vs. DM-WT-c^WT.

Effect of diabetes and/or tnfaip2 deletion on podocytes

There were no podocyte abnormalities in non-diabetic tnfaip2 KO mice. However, downregulation of both NPHS1/Nephrin and NPHS2/Podocin as well as FP effacement were observed in diabetic mice and exacerbated by Tnfaip2 abrogation (Figure 4(A-D)). Moreover, both apoptosis and podocyte loss, as assessed by counting the number of CDKN1C⁺ (cyclin-dependent kinase inhibitor 1C (P57)) cells, were only observed in diabetic tnfaip2 KO mice (Figure 4(E, F)).

Figure 4.

Effect of tnfaip2 deletion on diabetes-induced podocyte abnormalities. Renal cortex samples from diabetic (DM) and non-diabetic (ND) WT (ND-WT, DM-WT) and tnfaip2 KO (ND-KO, DM-KO) mice were studied 12 weeks after diabetes onset. (A) Representative immunofluorescence images of NPHS2/Podocin and NPHS1/Nephrin (magnification 400X, scale bar: 50 µm). (B) The graph shows the percent area of NPHS1/Nephrin and NPHS2/Podocin positive staining (n = 6 mice per group; ***p < 0.001 DM-WT vs. ND-WT; *p < 0.05 DM-KO vs. DM-WT). (C) Nphs1 and Nphs2 mRNA levels were measured by real-time PCR in total renal cortex and the results corrected for the expression of Wt1 (n = 5 mice per group; ***p < 0.001 DM-WT vs. ND-WT; *p < 0.05 DM-KO vs. DM-WT). (D) Electron microscopy images showing glomeruli from ND-WT, DM-WT, and DM-KO mice. The extent of foot processes effacement was greater in DM-KO than in DM-WT animals (magnification 3200X). (E) Apoptosis was assessed by TUNEL assay (green) and nuclei counterstained with DAPI (magnification 400X, scale bar: 50 µm). The percentage of apoptotic cells per glomerular area is shown in the graph (n = 6 mice per group; ***p < 0.001 DM-KO vs. others). (F) Representative immunostaining for CDKN1C/P57 (podocyte marker) (magnification 400X, scale bar: 50 µm). The number of CDKN1C/P57-positive cells per glomerular area is reported in the graph (n = 6 per group; ***p < 0.001 DM-KO vs. others). (G) Primary podocytes obtained from both WT and tnfaip2 KO mice were exposed to either vehicle or AGEs for 48 h and apoptosis assessed by TUNEL assay (apoptotic cells in pink). Nuclei were counterstained with DAPI (magnification 100X, scale bar: 200 µm). The percentage of apoptotic cells is shown in the graph (***p < 0.001 WT-AGEs vs. vehicle; KO-AGEs vs. WT-AGEs). (H) Primary podocytes obtained from both WT and tnfaip2 KO mice were exposed to either vehicle or AGEs for 48 h in the presence/absence of latrunculin-B (LatB, 100 nM) and apoptosis assessed by ANXA5/Annexin V staining (green). Nuclei counterstained with DAPI (magnification 100X, scale bar: 200 µm). The percentage of apoptotic cells is shown in the graph (***p < 0.001 AGEs vs. vehicle; AGEs-LatB. vs. AGEs).

Consistent with these in vivo findings, exposure of cultured podocytes to AGE-BSA induced apoptosis and this was aggravated by both Tnfaip2 deficiency and pre-incubation with latrunculin-B that prevents actin polymerization and thus TNT formation (Figure 4(G, H)). Therefore, lack of the TNFAIP2-TNT system appeared to sensitize podocytes to the deleterious effects of diabetes both in vitro and in vivo.

Effect of diabetes and/or tnfaip2 deletion on macroautophagy/autophagy

Autophagy is a key cellular mechanism for the degradation of dysfunctional organelles by sequestration within double-membrane vesicles autophagosomes (APs) and subsequent lysosomal fusion and degradation. We found that diabetes induced glomerular overexpression of the AP marker MAP1LC3/LC3 (microtubule-associated protein 1 light chain 3) (Figure 5(A)). Moreover, glomerular immunostaining for the autophagy receptor SQSTM1/p62 (sequestosome 1) was increased in diabetic mice, particularly in podocytes (Figure 5(D, C)). Similarly, exposure of cultured podocytes to either AGE-BSA or HG enhanced both LC3-II expression and the number of SQSTM1/p62⁺-LC3⁺ pucta (Figure 5(D-I)). These diabetes-induced alterations were magnified by tnfaip2 deletion in both cultured podocytes and diabetic mice.

Figure 5.

Effect of tnfaip2 deletion on diabetes-induced autophagy abnormalities. Markers of autophagy were studied in the glomeruli of diabetic and non-diabetic WT (ND-WT, DM-WT) and tnfaip2 KO (ND-KO, DM-KO) mice and in primary podocytes from ND-WT and ND-KO animals exposed to normal glucose (NG), high glucose (HG), vehicle, or AGEs for 48 h. (A) Immunoblotting showing MAP1LC3/LC3-II expression in isolated glomeruli (TUBB/Tubulin loading control). The graph shows the results of densitometry analysis (n = 5 per group; *p < 0.05 DM-WT vs. ND-WT; ***p < 0.001 DM-KO vs. DM-WT). (B) Representative images of glomerular SQSTM1/p62 immunostaining (magnification 400X, scale bar: 50 µm). (C) The graph shows the number of SQSTM1/p62-positive cells per glomerular area (n = 6 per group; ***p < 0.001 DM-WT vs. ND-WT; DM-KO vs. DM-WT). (D and E) Immunoblotting showing MAP1LC3/LC3-II expression (TUBB/Tubulin loading control) in cultured podocytes exposed to (D) vehicle or AGEs and (E) NG or HG. Results of densitometry analysis are shown in the graph (n = 3; *p < 0.05 WT-AGE vs. vehicle; ***p < 0.001 KO-AGE vs. WT-AGE; **p < 0.01 WT-HG vs. WT-NG and KO-HG). (F and G) Co-staining (yellow dots) for SQSTM1/p62 (green) and MAP1LC3/LC3-II (red) in cultured podocytes exposed to (F) vehicle or AGEs and (G) NG or HG. Nuclei were counterstained with DAPI (magnification X630, bar: 50 µm). (H and I) Quantification of the percent SQSTM1/p62⁺ and MAP1LC3/LC3⁺ area is shown in the graphs (n = 3; ***p < 0.001 WT-AGE/WT-HG vs. vehicle/NG; KO-AGE/KO-HG vs. WT-AGE/WT-HG). mRNA levels of (J) Becn1, (K) Atg5, (L) Atg7 were measured by real time-PCR in the renal cortex of experimental animals and the results corrected for the expression of the housekeeping gene Hprt (n = 5 per group; p = ns). (M and N) Becn1 mRNA levels were assessed by real time-PCR in cultured podocytes exposed to (M) vehicle or AGEs, (N) NG or HG. Results were corrected for the expression of the housekeeping gene Rn18s (n = 3; p = ns).

The concurrent rise in both LC3-II and SQSTM1/p62 together with the unchanged expression of the autophagy inducers Becn1 (beclin 1, autophagy related), Atg5 (autophagy related 5), and Atg7 (autophagy related 7) (Figure 5(J-N)), suggested that tnfaip2 deletion altered autophagy by inhibiting AP digestion.

Consistent with this, monitoring of the autophagic flux by tandem fluorescent-tagged LC3 (RFP-GFP-LC3) in AGE-BSA-treated podocytes showed that lack of Tnfaip2 increased the number of APs (GFP⁺ mRFP⁺ yellow puncta), but reduced the number of red puncta (GFP⁻mRFP⁺) that form when yellow APs fuse with lysosomes and the green fluorescence of GFP is quenched by the low pH (Figure 6(A, B)). Importantly, a similar effect was also observed in AGE-BSA-treated WT podocytes pre-exposed to latrunculin-B indicating that both lack of Tnfaip2 and inhibition of TNT formation can magnify abnormalities in AP clearance in the context of diabetes. The calculated GFP: RFP ratio was increased by exposure to AGEs and even further by tnfaip2 deletion and latrunctulin B, indicating reduced autophagic flux (Figure 6(C)).

Figure 6.

Effect of tnfaip2 deletion and/or diabetes on both autophagic flux and autolysosome (AL) formation. (A) Autophagy flux was monitored by tandem fluorescent-tagged LC3 in primary WT and tnfaip2 KO podocytes exposed to either vehicle or AGEs for 48 h in the presence or absence of latrunculin B (LatB). Nuclei were counterstained with DAPI. Yellow puncta autophagosomes (APs) and free red puncta (acidic AL) are shown in the RFP+GFP-LC3 panels (magnification X630, bar: 50 µm). (B) Quantification of the number of yellow and free red puncta per cell is shown in the graph (n = 3; ***p < 0.001 WT-AGE vs. vehicle; **p < 0.01 KO-AGE and LatB-AGE vs. WT-AGE). (C) The relative GFP:RFP ratio was assessed by ImageJ and results are shown in the graph (**p < 0.01 WT-AGE vs. vehicle; *p < 0.05 KO-AGE and LatB-AGE vs. WT-AGE). (D) Fusion of APs with lysosomes was assessed by double immunofluorescence for MAPLC3/LC3 (red) and LAMP1 (green). Merged images showed co-localization (magnification X630, bar: 50 µm). (E) The graph shows the percentage of the yellow positive area per cell (***p < 0.001 WT-AGE vs. WT-vehicle; KO-AGE vs. WT-AGE).

Effect of diabetes and/or tnfaip2 deletion on lysosomes

To establish if AP accumulation was due to either failure to fuse with lysosomes or inhibition of lysosomal acidification, we assessed fusion by double immunostaining for LC3 and the lysosome marker LAMP1 (lysosomal-associated membrane protein 1). There was a significant increase in the number of LC3⁺ LAMP1⁺ autolysosomes (ALs) in AGE-BSA treated podocytes compared to control cells and this diabetes-induced effect was magnified by tnfaip2 deletion (Figure 6(D, E)). Therefore, AP-lysosome fusion was preserved and altered lysosome acidification was likely the predominant mechanism of AP accumulation.

Consistent with this, AGE-BSA-treated WT podocytes and to a greater extent AGE-treated tnfaip2 KO cells showed a reduction in acridine orange (AO)-stained acidic organelles and an enhanced release of the lysosomal enzyme CTSD (cathepsin D) into the cytosol, which are both markers of lysosomal dysfunction and lysosomal membrane permeabilization (LMP) (Figure 7(A, B)). Moreover, activity of CTSD was significantly reduced in AGE-BSA-treated podocytes and even more in AGE-BSA-treated tnfaip2 KO cells (Figure 7(E)). Re-expression of Tnfaip2 abolished the deleterious effect of tnfaip2 deletion on AGE-BSA-induced SQSTM1/p62 accumulation, LMP, CTSD activity (Figure 7(A-E)), and podocyte apoptosis (Figure 7(F, G)), confirming specificity of these findings.

Figure 7.

Effect of tnfaip2 deletion and re-expression on diabetes-induced changes on lysosomes, autophagosomes (APs), and apoptosis. Lysosome both integrity and function was assessed in primary murine WT and tnfaip2 KO podocytes exposed to either vehicle or AGEs for 48 h. The effect of tnfaip2 re-expression on lysosomal integrity and function, AP accumulation, and apoptosis was tested by transfecting KO podocytes with an adenovirus expressing either Tnfaip2 (KO^Adv+) or a control mock vector (KO^Adv-). (A) Representative immunofluorescence images of podocytes stained with Acridine Orange (AO) (magnification X100, bar: 200 µm). Quantification of green to red AO fluorescence is reported in the graph below (n = 3; ***p < 0.001 WT-AGE and KO^Adv+-AGE vs. WT-vehicle; KO-AGE and KO^Adv–AGE vs. WT-AGE). (B) Colocalization of CTSD (red) with LAMP1⁺ lysosomes (green). Nuclei were stained with DAPI (magnification X630, bar: 50 µm). (C) Tnfaip2 mRNA levels were measured by real-time PCR in WT, KO^Adv+ and KO^Adv- podocytes and the results corrected for the expression of Rn18s (n = 3; ***p < 0.001 KO^Adv- vs. others). (D) Effect of Tnfaip2 re-expression on AP accumulation as shown by costaining (yellow dots) for SQSTM1/p62 (green) and MAP1LC3/LC3 (red). Nuclei were counterstained with DAPI (magnification X630, bar: 50 µm). (E) Quantification of CTSD activity expressed as relative fluorescence units (RFU) per number of cells (**p < 0.01 WT-AGE and KO^Adv+-AGE vs. WT-vehicle; KO-AGE and KO^Adv–AGE vs. WT-AGE). (F) Effect of Tnfaip2 re-expression on podocyte apoptosis assessed by TUNEL assay (apoptotic cells in pink) (magnification 100X, scale bar: 200 µm). (G) The percentage of apoptotic cells is shown in the graph (**p < 0.01 WT-AGE and KO^Adv+-AGE vs. WT-vehicle; KO-AGE and KO^Adv–AGE vs. WT-AGE).

TNT-TNFAIP2-mediated lysosome and AP transfer

To functionally link lysosome alterations to the TNFAIP2-TNT system, we next assessed whether TNTs can mediate transfer of lysosomes and APs between podocytes. Cell-Tracker Blue-labeled podocytes (recipients) were treated with AGE-BSA (100 µg/ml) and then co-cultured with normal podocytes (donors), containing green fluorescent protein (GFP)-labeled lysosomes. Fluorescent lysosomes were found along TNTs and in the cytosol of recipient blue cells, indicating lysosome transfer (Figure 8(A)) from healthy to AGE-BSA-treated cells. Flow cytometry analyses showed that TNT-mediated lysosome transfer towards AGE-BSA-treated podocytes was significantly increased. However, transfer was almost abolished when recipient podocytes were cocultured with donor podocytes either lacking Tnfaip2 or pre-exposed to latrunculin-B, indicating a TNFAIP2-TNT-dependent phenomenon. Consistently, re-expression of Tnfaip2 rescued the ability of KO podocytes to form TNTs (WT^Adv-: 19.54 ± 0.39; KO^Adv-: 8.32 ± 0.56; KO^Adv+: 17.43 ± 0.89, % of podocytes connected by TNTs; p < 0.001 KO^Adv- vs. others) and to exchange lysosomes and Tnfaip2 overexpression even magnified lysosome transfer (Figure 8(B, C)). Of relevance, we also demonstrated transfer of lysosomes between podocytes exposed to different concentrations of AGEs (50 vs. 200 µg/ml) (Figure 8(D, E)). AP transfer in the opposite direction was observed from AGE-BSA-treated to normal podocytes (Figure 8(F)). As shown in Fig. S2, transfer was not limited to autophagy-related organelles and shuttling of mitochondria from normal podocytes to AGE-BSA-treated cells was also found.

Figure 8.

TNFAIP2-TNT-mediated lysosome and autophagosome (AP) exchange in AGE-treated podocytes. (A) Podocytes stained with CellTracker Blue were exposed to either vehicle or AGEs for 48 h, and then cocultured with donor podocytes carrying LAMP1-GFP-labeled lysosomes. The images show green lysosomes within the TNT and in the cytosol (*) of a recipient podocyte (magnification X630, bar: 50 µm, DIC images showing the TNT). (B and C) Lysosome transfer (GFP and Blue dually labeled cells) was quantified by flow cytometry in recipient podocytes exposed to either vehicle or AGEs and co-cultured with donor WT and tnfaip2 KO podocytes. WT donor podocytes were treated with latrunculin-B (WT-LatB 100 nM, 1 h) to prevent TNT formation. KO and WT donor podocytes were transfected with an adenovirus expressing either tnfaip2 (KO^Adv+, WT^Adv+) or a mock vector (n = 3; **p < 0.01 WT-AGE and KO^Adv+-AGE vs. WT-Vehicle; KO-AGE, WT^LatB-AGE vs. WT-AGE; p < 0.05 WT^Adv+-AGE vs. WT-AGE). (D) Recipient podocytes were stained with CellTracker Blue and exposed to high AGE concentration (200 µg/ml, 48 h; H-AGE). Donor podocytes with LAMP1-GFP-labeled lysosomes were exposed to lower AGE concentrations (50 µg/ml, 48 h; L-AGE). In coculture, green lysosomes were seen in the cytosol (*) of recipient podocytes (magnification X630, bar: 50 µm, DIC images showing the TNT). (E) Donor (GFP-labeled lysosomes) and recipient (Recip, CellTracker Blue) podocytes were pre-exposed to either H-AGE (200 µg/ml) or L-AGE (50 µg/ml) concentrations and then co-cultured. Lysosome transfer (GFP and Blue dually labeled cells) from donor towards recipient (Recip) podocytes was quantified by flow cytometry (***p < 0.001 Donor-L-AGE to Recip-H-AGE vs. others). (F) Donor podocytes exposed to AGEs for 48 h were transfected with tandem fluorescent-tagged LC3 to label APs (yellow punta), then co-cultured with healthy recipient podocytes. The image shows yellow puncta within the TNT and in the cytosol (*) of a recipient podocyte (magnification X630, bar: 50 µm, DIC images showing the TNT).

Cytoprotective effect of the TNFAIP2-TNT system on AGE-treated podocytes

We then investigated the potential cytoprotective effect of TNFAIP2-TNT-mediated organelle exchange in WT diabetes-injured podocytes. In the co-culture system described above, we found that coincubation of AGE-BSA-injured podocytes with healthy donor podocytes not only enhanced lysosome transfer in a TNFAIP2-TNT dependent manner, but also reduced SQSTM1/p62 accumulation, LMP, and podocyte apoptosis in recipient cells. These protective effects of donor podocytes were magnified by Tnfaip2 overexpression. On the contrary, benefit was no longer observed when leupeptin-treated podocytes, carrying altered lysosomes, were used as donor cells, confirming the importance of lysosome transfer in rescuing damaged cells (Figure 9).

Figure 9.

Effect of TNFAIP2-TNT-mediated lysosome exchange on autophagosome (AP) accumulation, lysosome dysfunction, and apoptosis in a co-culture system. Podocytes were exposed to either vehicle or AGEs for 48 h. The effect of the TNFAIP2-TNT system on autophagy, lysosome dysfunction, and apoptosis was assessed by co-culturing recipient AGE-treated podocytes with donor (D) cells transfected with an adenovirus expressing Tnfaip2 (WT^Adv+), mock vector (WT^Adv-) or pre-exposed to leupeptin (Leup, 200 µg/ml WT^Leup). (A) Immunostaining for SQSTM1/p62 (pink). Blue recipient podocyte (CellTracker Blue) and green donor podocytes (CellTracker Green) (magnification 630X, scale bar: 50 µm) and quantification of SQSTM1/p62⁺ positive area per cell (n = 3; ***p < 0.001 AGE and AGE+Donor-WT^Leup vs. vehicle; **p < 0.01 AGE-Donor-WT^Adv- vs. AGE andAGE-Donor-WT^Adv+). (B) Acridine Orange (AO red and green) staining. Blue donor podocytes (CellTracker Blue) (magnification 630X, scale bar: 50 µm) and quantification of green to red AO fluorescence (n = 3; ***p < 0.001 AGE and AGE+Donor-WT^Leup vs. vehicle; **p < 0.01 AGE-Donor-WT^Adv- vs. AGE and AGE-Donor-WT^Adv+). (C) The graph shows the percentage of apoptotic podocytes as assessed by ANXA5/Annexin V staining (n = 3; ***p < 0.001 AGE and AGE+Donor-WT^Leup vs. vehicle; **p < 0.01 AGE-Donor-WT^Adv- vs. AGE and AGE-Donor-WT^Adv+).

Finally, overexpression of Tnfaip2 in monoculture of AGE-exposed podocytes also improved SQSTM1/p62 accumulation, LMP, CTSD activity and significantly reduced podocyte apoptosis (Figure 10), indicating that benefit was observed even in the absence of exogenous healthy cells functioning as supplier of healthy organelles.

Figure 10.

Effect of Tnfaip2 over-expression in monoculture of AGE-treated podocytes. WT podocytes and podocytes transfected with an adenovirus expressing Tnfaip2 (WT^Adv+) or a mock vector (WT^Adv-) were exposed to either vehicle or AGEs for 48 h. (A) Double immunostaining (yellow dots-autophagosomes) for MAP1LC3/LC3⁺ and SQSTM1/p62⁺ (magnification X630, bar: 50 µm, nuclei: DAPI) and quantification of the percent SQSTM1/p62⁺ and MAP1LC3/LC3⁺ positive area (n = 3; ***p < 0.001 AGE-WT^Adv- vs. others). (B) Acridine Orange (AO) staining (magnification 100X, scale bar: 200 µm) and quantification of green and red AO fluorescence (n = 3; ***p < 0.001 AGE-WT^Adv- vs. others). (C) CTSD activity expressed as relative fluorescence units (RFU) per number of cells (n = 3; ***p < 0.001 AGE-WT^Adv- vs. others). (D) Images of apoptotic podocytes (pink, TUNEL assay) (magnification 100X, scale bar: 200 µm, nuclei: DAPI) and quantification of the percentage of TUNEL positive cells (n = 3; ***p < 0.001 AGE-WT^Adv- vs. others). (E) Tnfaip2 mRNA levels were measured by real-time PCR in WT^Adv+ and WT^Adv- podocytes and the results corrected for the expression of Rn18s (n = 3; ***p < 0.001 WT^Adv+ vs. WT^Adv+).

Discussion

Our study provides evidence that TNFAIP2 is overexpressed by podocytes in both human and experimental DN. Moreover, tnfaip2 deletion exacerbated albuminuria, renal function loss, and structural abnormalities of experimental DN with evidence of enhanced both podocyte injury and impaired autophagy.

In humans, TNFAIP2 overexpression by podocytes was observed not only in patients with established DN, but also in individuals with microalbuminuria, though the number of subjects was small. This suggests that TNFAIP2 induction may be an early event in the natural history of DN. According with this, there was glomerular Tnfaip2 overexpression in an early phase of experimental diabetes.

Podocyte overexpression of TNFAIP2 also occurs in FSGS [54], raising the hypothesis that TNFAIP2 induction may be a common response to podocyte injury. TNFAIP2 upregulation is likely a compensatory mechanism aimed to limit glomerular injury. Indeed, in our study tnfaip2 deletion worsened podocyte injury. Moreover, BALB/c mice lacking Tnfaip2 spontaneously develop podocyte damage, leading to FSGS with aging [54].

At variance with BALB/c mice, C57BL/6 tnfaip2 KO mice did not have a renal phenotype in the absence of diabetes. The greater susceptibility to develop FSGS of BALB/c compared to C57BL/6 mice may provide explanation for this strain-dependent effect [55–57]. Lack of Tnfaip2, however, worsened experimental DN in C57BL/6 mice, implying that diabetes-induced glomerular injury is required to disclose the renal protective effect of TNFAIP2 in this strain.

Inflammatory mechanisms are important in the pathogenesis of DN [58] and deletion of Tnfaip2 exacerbated diabetes-induced glomerular inflammation. Moreover, TNFAIP2 is expressed by inflammatory cells and induced by inflammatory cytokines [45,48,50,52]. However, transplantation of BM cells from WT to tnfaip2 KO mice did not modify the phenotype of recipient diabetic animals. Therefore, lack of Tnfaip2 in podocytes rather than in inflammatory cells is likely involved in the worsening of experimental DN. Consistently, in diabetic mice TNFAIP2 upregulation occurred predominantly in podocytes and silencing of Tnfaip2 in isolated cultured podocytes was sufficient to enhance diabetes-induced apoptosis.

In diabetic mice, tnfaip2 deletion augmented podocyte accumulation of APs by blocking the late stages of the autophagy pathway. Podocytes have a high basal autophagy rate [59,60] and autophagy dysregulation is believed to play an important role in pathogenesis of DN. Consistent with our results in diabetic WT mice, podocyte LC3-II and/or SQSTM1/p62 accumulation have been reported in both human and experimental DN [61–67]. Moreover, disruption of the autophagy-lysosome pathway by AGE-BSA was described in both podocytes and tubular epithelial cells [61,68]. However, changes in autophagy induction are also relevant in experimental diabetes [11,69–71] and both early induction and long-term inhibition of autophagy by hyperglycemia have been shown in mice with STZ-induced diabetes [72]. Moreover, global and podocyte-specific deletion of key regulators of autophagy (Atg5, MTOR [mechanistic target of rapamycin kinase] complex 1 [MTORC1], Sirpa [signal-regulatory protein alpha]) has been shown to affect susceptibility to DN [65,72–74].

Diabetes-induced alterations in autophagy may also differ in various renal compartments. Our study showed diabetes-induced LC3-II accumulation in both isolated glomeruli and podocytes, while a recent paper demonstrated LC3-II downregulation in both proximal tubular cells and the renal cortex that consists predominantly of tubules [75]. Opposite effects of diabetes in the glomeruli and the tubules may be due to site-specific differences in autophagy regulation. In line with this, Mir214 that mediated diabetes-induced LC3-II downregulation in the tubuli is not expressed by the glomeruli [75]. In addition, tubular cells at variance with podocytes are an active site of glucose reabsorption and this is likely to be relevant in the context of diabetes-induced changes in autophagy. In keeping with this, inhibition of the proximal tubule SLC5A2/SGLT2 (solute carrier family 5 (sodium/glucose cotransporter), member 2) has been shown to ameliorate autophagy deficiency in tubular cells exposed to HG [76].

Altered AP clearance was due to lysosomal dysfunction as proven by worsening of LMP and reduced CTSD activity in diabetes-injured tnfaip2 KO podocytes. The important role of lysosomes in podocyte injury is increasingly recognised. Mice with podocyte-specific deletion of lysosomal enzymes as Atp6ap2 (ATPase, H+ transporting, lysosomal accessory protein 2; prorenin receptor) and CTSD show impaired autophagy, podocyte damage, podocyte apoptosis, FP effacement, proteinuria, and renal function loss [77,78]. Moreover, LMP can initiate lysosome-dependent cell death through release of lysosomal content, providing a potential explanation for the enhanced apoptosis of diabetic tnfaip2 KO podocytes.

Our in vitro experiments aimed to clarify the mechanism whereby Tnfaip2 deficiency enhances lysosomal dysfunction show that tnfaip2 deletion hampered TNT-mediated transfer among podocytes. Consistent with our previous results [54], we found that podocytes can form TNTs in a TNFAIP2-dependent manner. Several cellular stresses, including oxidative stress, serum deprivation, and hypoxia, have been shown to trigger TNT formation in other cell types [24,33,35,37,38]. Herein, we found that podocyte incubation in a HG milieu induced TNFAIP2 overexpression and that both HG and AGE-BSA were potent TNT inducers. Exposure of mesothelioma cell lines to HG also enhances TNT formation [32]. Moreover, deletion of the AGE receptor Ager/Rage (advanced glycosylation end product-specific receptor) has been shown to reduce TNTs in mesothelial cells, astrocytes, and Chinese hamster ovary cells [79,80].

Mechanistically, it has been previously shown that TNFAIP2 cooperates with the RALA/RAL (v-ral simian leukemia viral oncogene A (ras related)) and the exocyst complex to trigger F-actin polymerization and TNT development [45]. The observation that PI3K-AKT1 signaling mediated TNFAIP2 overexpression and TNT formation suggests that the pro-survival effect of the PI3K-AKT1 pathway may be at least in part mediated by induction of the TNFAIP2-TNT system. In line with our results, the AKT1-PI3K pathway has been previously shown to induce Tnfaip2 mRNA expression in astrocytes [33].

Although TNFAIP2 is required for TNT formation in podocytes and Tnfaip2 overexpression is known to enhance TNT formation [81], AGEs did not increase TNFAIP2 expression, but still promoted TNT formation. This suggests that induction of TNTs can occur via other mechanisms in addition to TNFAIP2 overexpression, though the presence of TNFAIP2 is still required for this to take place. Factors, inducing actin remodeling, which orchestrates TNT formation, are likely candidates.

As hyperglycemia, AGEs, and consequent oxidative stress/inflammation are stressful conditions to cells, it is not surprising that they induce TNT formation. Indeed, the ability of cells to interconnect and to share a common pool of resources via TNTs represents a potent rescue mechanism that provides significant survival advantages by allowing individual cells as part of a supercellular system to maintain redox and metabolic homeostasis [82].

TNTs interconnecting podocytes transferred lysosomes in a TNFAIP2 dependent manner. Lysosome transport from healthy to AGE-injured podocytes was enhanced, suggesting that it was aimed to rescue podocytes with dysfunctional lysosomes by delivering healthy organelles. Transfer of APs in the opposite direction was also observed and may represent a mechanism to dilute the overload of dysfunctional APs among cells, thus preventing extreme organelle dysfunction [82]. Notably, lysosome transfer was also seen when donor and recipient podocytes were pre-exposed to different AGE-BSA concentrations to mimic the in vivo condition in which podocytes are all exposed to the diabetes insult, but may have different degrees of lysosomal dysfunction.

Lysosome transfer via TNTs has been previously demonstrated in other cell types [43,83–86]. Moreover, TNT-mediated lysosomal transfer has been shown to correct cystinosis, a genetic lysosomal disease, through bidirectional lysosome exchange between hematopoietic stem cells (HSC)-derived macrophages and cystinosin-deficient cells [87]. However, lysosome transfer can also be deleterious and it has been implicated in the spreading of neurotoxic aggregates in neurodegenerative diseases [88]. Evidence of AP transfer is more limited; however, AP shuttling via TNTs has been reported in both squamous cell carcinoma [44] and the leukemic niche [89].

Deficiency of Tnfaip2 may enhance diabetes-induced accumulation of damaged APs and lysosomes, possibly explaining worsening of DN in tnfaip2 KO mice. In keeping with this hypothesis, TNFAIP2-TNT-mediated transfer from healthy to AGE-BSA-exposed podocytes ameliorated autophagy, lysosome dysfunction, and apoptosis in recipient cells.

We cannot exclude the possibility that transfer of other organelles was also implicated as TNTs have been shown to transport mitochondria, mRNA, lipid droplets, virus, vesicles in other cell types and mitochondria in podocytes. However, donor podocytes carrying altered lysosomes failed to improve autophagy, lysosome dysfunction, and podocyte apoptosis, suggesting that TNFAIP2 protected podocytes predominantly through TNT-mediated lysosome transfer.

Importantly, overexpression of Tnfaip2 in donor podocytes enhanced lysosome transfer towards podocytes exposed to AGE-BSA and almost abolished the deleterious effect of AGE-BSA on autophagy, lysosome function, and apoptosis in recipient cells. A similar cytoprotective effect of Tnfaip2 overexpression was also observed in monoculture of podocytes exposed to AGE-BSA, a model that more closely resembles the in vivo condition. Therefore, lack of the TNFAIP2-TNT system worsened diabetes-induced podocyte injury, while its induction counteracted the deleterious effects of diabetes.

In conclusion, we demonstrated that the TNFAIP2-TNT system has an important protective role in DN. As several cell types express this system, our findings may also be relevant to other diabetes complications. Evidence that diabetes-injured podocytes can be rescued by Tnfaip2 overexpression suggests that the TNFAIP2-TNT system may represent a novel druggable target.

Materials and methods

Patients

Archival Serra’s fluid-fixed, paraffin-embedded kidney sections from patients with type 2 diabetes with either incipient (n = 3, 30–300 mg/24 h) or overt nephropathy (n = 12, persistent proteinuria >0.5 g/24 h) were studies. Renal tissue from subjects (n = 8), who underwent surgery for localized grade I hypernephroma and did not have proteinuria or glomerular abnormalities, as detected by both light and immunofluorescence microscopy, was used as control. Proteinuria was measured by Pyrogallol-red on three separate urine collections and serum creatinine by the kinetic Jaffè method on a Beckman Synchron CX3 analyser (Beckman Coulter, Italy). Hypertension was defined for blood pressure values >140/90 mmHg measured in three separate occasions [90]. All clinical procedures were in accordance with the Helsinki Declaration.

Mice

Male C57BL/6 mice were purchased from Charles River Laboratories (Milan, Italy). The tfnaip2 KO C57BL/6 mice were generated at the Riken Center in Japan, as we have previously reported [54]. Animals were maintained on a normal diet and both housing and care of laboratory animals were in accordance with Italian law. Procedures were approved by the Ethical Committee of the Turin University, Italy.

Diabetes induction

Diabetes was induced in 8-week-old both WT and tnfaip2 KO C57BL/6 mice by STZ intraperitoneal injection (Merck Life Science, S0130; 55 mg/Kg for 5 consecutive days). Control mice were injected with sodium citrate buffer. Diabetes onset was confirmed by blood glucose levels >250 mg/dl four weeks after the first dose of STZ. Before sacrifice, blood samples were taken via saphenous vein puncture on alert, 4 h–fasted animals, and glucose levels measured using a glucometer (Accu-chek Aviva, Italy). Systolic blood pressure was measured by tail-cuff plethysmography. Urine was collected over 18 h, with each mouse individually housed in a metabolic cage and provided with food and water ad libitum. Urinary ALB concentration was measured by a mouse albumin ELISA kit (Bethyl Laboratories, E99-134). Creatinine clearance was calculated from urine and serum creatinine concentrations, as determined by high-performance liquid chromatography (HPLC) [91]. Glycated hemoglobin was measured by quantitative immunoturbidimetric latex determination (Sentinel Diagnostic, 17,418). Twelve weeks after diabetes onset, mice were euthanized, and kidneys removed and weighted; half a kidney was formalin-fixed and paraffin-embedded for light microscopy analyses and the other half was included in optimal cutting temperature compound (VWR, 00411243) and snap-frozen in N₂. The other kidney was stored at −80°C for RNA and protein analyses. In a subset of mice, glomeruli were isolated using a modified Dynabead method, as previously described, at both 6 and 12 weeks after diabetes induction [54].

Chimeric mice

Eight-week-old male WT and tnfaip2 KO mice were used as BM donors. BM cells were flushed from tibial and femoral cavities under sterile conditions, filtered through 70 μm nylon meshes (Corning, 431,751), and then transplanted without further purification or in vitro expansion. Recipient male mice, aged 8 weeks, underwent whole body-irradiation (8 Gy) and 24 h later injected with 2.0 × 10⁶ BM cells via the tail vein. The tnfaip2 KO mice received a BM transplant from either WT (KO-c^WT; n = 5) or tnfaip2 KO (KO-c^KO; n = 5) animals and WT recipients from WT animals (WT-c^WT; n = 5). DNA was isolated from blood cells (Thermo Fisher Scientific, K1820-02) and chimerism confirmed by polymerase chain reaction (PCR).

Podocytes

Primary podocytes were obtained from both six-week-old C57BL/6 WT and tnfaip2 KO mice, as previously described. Briefly, decapsulated kidneys were digested in 1 mg/ml collagenase A (Merck Life Science, C9407) for 30 min at 37°C. Glomeruli were isolated using Dynabeads (Thermo Fisher Scientific, 14,013) and cultured for 4–5 days on type IV collagen (Merck Life Science, C5533)-coated dishes, in DMEM medium (Gibco, 31,885–023) supplemented with 10% fetal bovine serum (Gibco, 10,270–106), 100 U/ml penicillin/100 and μg/ml streptomycin (Gibco, 15,070–063), 2 mM L-glutamine (Gibco, 25,030–024), 100 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES; Merck Life Science, H3375). Subculture of primary podocytes was performed by detaching glomerular cells with trypsin-ethylenediaminetetraacetic acid (EDTA; Merck Life Science, T4174) followed by sieving through 30-μm pore cell strainers (PluriSelect, 43–50,030-01). For TNT visualization, organelle transfer detection, and immunofluorescence staining, podocytes were seeded onto 35-mm μ-dishes (Ibidi, 80,136; 60,000 cells per dish). Cells were exposed either to HG (25 mM) or NG (7 mM made iso-osmolar with mannitol [Merck Life Science, M1902]), AGE-BSA/BSA (Merck Life Science, 121,800-M; 50, 100, and 200 µg/ml) for 48 h. In 24 h-mechanical stretch experiments, cells were seeded on type I collagen-coated silicon elastomer-base culture plates (Flexcell International, BF-3001C) and connected to a vacuum pump to result in a 10% elongation; control cells were grown in non-deformable, but otherwise identical plates.

In some experiments, cells were pre-incubated with 100 nM latrunculin B (Merck Life Science, L5288), 1 µM AKT1 inhibitor VIII (Merck Life Science, 124,018), 50 µM LY294002 (Merck Life Science, L9908), 5 µM p-nitro-Pifithrin-α (Santa Cruz Biotechnology, sc222176), or 200 μg/ml leupeptin (Merck Life Science, L2884).

To knockdown Tfnaip2, podocytes (3 x 10⁵ cells) were transfected using Lipofectamine 3000 Transfection Reagent (Thermo Fisher Scientific, L3000001) with a plasmid construct encoding Tnfaip2-specific sh-RNA (shRNA: 5’-GGCAATCTACCAAGCAGTGAAGTCAGGAG-3’, scramble negative control: 5’ GCACTACCAGAGCTAACTCAGATAGTACT 3’) cloned in a pRS vector (ExactHuSH, OriGene Technologies Inc, TL513846). To overexpress Tnfaip2, cells were transfected with a mouse Tnfaip2 adenovirus (Vectors Biolabs, AAV-274678). Western blotting was performed to assess the knockdown/overexpression efficiency.

TNTs and organelle transfer

To reveal TNTs, podocytes were labeled with Alexa Fluor®-488 conjugated-wheat germ agglutinin (WGA; Thermo Fisher Scientific, W11261; 5 μg/ml) and imaged by Apotome II Zeiss Microscope controlled by Axiovision software with 0.25-µm Z-stacks. After image acquisition, the number of cells connected by WGA-labeled straight structures that did not adhere to the substrate, as assessed by Z-stack, and were thinner than 1 μm was counted in blind. Data were presented as the percentage of total counted cells (at least 200 cells).

In transfer experiments, recipient podocytes were labeled with the cytoplasmic dye Cell Tracker Blue CMFDA (Thermo Fisher Scientific, C2110; 5 μM). Donor podocytes were transfected with a modified insect virus (Baculovirus) expressing a fusion construct of a green fluorescent protein with the lysosomal protein LAMP1 (Cell Light™ Lysosomes-GFP BacMam 2.0; Thermo Fisher Scientific, C10596). Cells were co-cultured at a ratio of 1:1 for 24 h. To visually assess lysosome transfer, both fluorescent and differential interference contrast (DIC) microscopy were used. To quantify lysosomal transfer efficiency, cells were fixed in 2% w:v paraformaldehyde (PFA; Merck Life Science, 158,127), passed through a cell strainer, then analyzed by flow cytometry (Citoflex, Beckman Coulter, USA). Single positive cells (donor and recipient cells alone) were used to set up the instrument. Data analysis was performed using the Cytexpert software (Version 2.3). To control for TNT-independent transfer, recipient cells were plated in the bottom compartment of a transwell system (Corning, 3401) and donor cells on a 0.4-µm pore-sized insert.

Autophagy, lysosome dysfunction, and apoptosis

Autophagic flux was assessed in podocytes transfected with Premo™ Autophagy Tandem Sensor RFP-GFP-LC3B (Thermo Fisher Scientific, P36239). Results were expressed as the number of yellow and free red puncta per cell as well as relative GFP:RFP ratio (at least 50 cells). CTSD enzymatic activity was measured using a fluorescence-based assay kit (BioVision, K143-100) and results expressed as relative fluorescence units (RFU) per number of cells. For acidic organelle labeling, cells were incubated with 1 µg/ml of AO (Thermo Fisher Scientific, A1301) for 10 min, washed with phosphate-buffered saline (PBS; Merck Life Science, 79,382), and then fixed in 100% methanol. Results were expressed as green to red AO fluorescence (at least 100 cells). Apoptosis was detected by either transferase-mediated dUTP nick end-labeling (TUNEL) assay (Millipore, S7165) or Alexa Fluor® 488-conjugated ANXA5/annexin V (Thermo Fisher Scientific, A13201). Results were expressed as the percentage of apoptotic (TUNEL or ANXA5-positive cells) per glomerular or cellular area.

Microscopy

Light microscopy

Paraffin-embedded renal sections were stained using PAS. The mesangial index (mesangial area:total area of glomerulus) was quantified with Axiovision software (Zeiss, Germany; Version 4.7) in 15 glomeruli per kidney per animal.

Electron microscopy

Renal cortex pieces (1 mm³) were fixed in 2% glutaraldehyde (Merck Life Science, G7651), 4% PFA in PBS for 4 h at room temperature. Samples were then postfixed in 1% osmium tetroxide (Merck Life Science, 201,030) for 2 h, dehydrated, and embedded in Epon 812 (Fisher Scientific, 50,980,391). Ultrathin sections (70 nm) were cut using a Leica EM UC6 ultramicrotome, counterstained with uranyl acetate and lead citrate, and analyzed with an Energy Filter Transmission Electron Microscope (EFTEM, ZEISS LIBRA® 120, Germany) equipped with an yttrium aluminum garnet scintillator slow-scan charge-coupled device camera (Sharp eye, TRS, Moorenweis, Germany). Analysis was performed blindly in 10 glomeruli for each animal.

Fluorescence and DIC microscopy

A Zeiss APOTOME 2 system (Zeiss, Germany) equipped with an incubator for live-cell imaging microscopy and Nomarski optics for DIC microscopy was used in TNT studies. DIC microscopy was used to avoid TNT phototoxic damage. Sequences of optical planes (Z-stack) were acquired for reconstruction in the x-z plane.

mRNA analysis

Total RNA was extracted using the TRIZOL reagent (Thermo Fisher Scientific, 15,596,026). One or two µg of total RNA was reverse transcribed into cDNA using the high-capacity cDNA reverse transcription kit (Thermo Fisher Scientific, 4,368,814). mRNA expression was assessed by real-time PCR using the TaqMan Gene Expression Master Mix (Thermo Fisher Scientific, 4,369,016), and the pre-developed TaqMan primer/probes: Tnfaip2 (Mm00447578), Nphs1 (Mm00497828), Nphs2 (Mm00499929), Tgfb1 (Mm00441724), Ly6c2 (Mm00841873), Ccl2 (Mm00441242), Ccr2 (Mm00438270), Tnf (Mm00443258), Becn1 (Mm01265461), Atg5 (Mm01187303), Atg7 (Mm00512209) (Thermo Fisher Scientific). Amplification was performed using the Abi Prism 7300 (Thermo Fisher Scientific, Italy). Relative quantification was carried out using the 2^−ΔΔCt method using either Hprt (hypoxanthine guanine phosphoribosyl transferase; Mm00446968) or Wt1 (Mm00460570) or Rn18s (Thermo Fisher Scientific, 4,333,760 T) as endogenous controls.

Protein Analysis

Renal cortex sections. immunohistochemistry (IHC) and double immunofluorescence (D-IF)

After antigen retrieval and blocking in 3% BSA (Merck Life Science, A6003), 4-μm kidney paraffin sections were incubated overnight at 4°C with the following primary antibodies: rabbit anti-TNFAIP2 (Abcam, ab196659; 1:100), rabbit anti-NPHS2/podocin (Merck Life Science, P0372; 1:200), rat-anti LGALS3/MAC2 (Cederlane, CL8942AP; 1:100), rabbit anti-CDKN1C/P57 (Santa Cruz Biotechnology, sc-8298; 1:100), rabbit anti-SQSTM1/p62 (Abcam, ab91526; 1:100). After washing, sections were incubated with goat anti-rat (Jackson ImmunoResearch Laboratories, 112–065-003; 1:500) or swine-anti rabbit (DAKO, E0353; 1:200) biotinylated secondary antibodies for 1 h.

For IHC, sections were then incubated with horseradish peroxidase (HRP)-conjugated streptavidin (DAKO, P0397; 1:300) for 1 h. Diaminobenzidine (Merck Life Science, D5905) was used as a chromogen substrate for HRP. Sections were visualized with an Olympus-BX4I microscope and digitized with a high-resolution camera (Zeiss, Germany). On average 30 randomly selected glomeruli were assessed per mouse. Results are expressed either as the number of positive cells per glomerular area or as percent area of positive staining per glomerulus. Two investigators performed evaluations in a blinded fashion.

For D-IF on human biopsies, sections were then incubated with Alexa Fluor 488-conjugated streptavidin (Thermo Fisher Scientific, S11223; 1:100). After further blocking in BSA, sections were incubated with an anti-mouse SYNPO (Progen, 61,094; 1:50) antibody for 18 h at 4°C, followed by RPE-conjugated goat anti-mouse (DAKO, R0480; 1:50). Digitized images were color-combined and assembled into photomontages by using Adobe Photoshop (Universal Imaging Corporation).

Immunofluorescence (IF). renal cortex

Single-antigen IF on mouse tissue was performed on snap-frozen kidney sections (3 µm). Slides were fixed in acetone for 5 min and blocked in 3% BSA, then incubated overnight at 4°C with guinea pig anti-NPHS1/Nephrin (Progen, GP-N2; 1:100), rabbit anti-NPHS2/Podocin, or rabbit anti-FN1/Fibronectin (Merck Life Science, F3648; 1:200) primary antibodies. Following washing, fluorescein isothiocyanate (FITC)-conjugated goat anti-guinea pig (Santa Cruz Biotechnology, sc-2441; 1:50) or swine anti-rabbit (DAKO, F0205; 1:50) secondary antibodies were incubated for 1 h. Sections were examined using a Zeiss APOTOME 2 system equipped with a high-resolution camera (Zeiss, Germany) and quantitated using the Axiovision 4.7 image analysis software (Zeiss, Germany). Results were expressed as the percentage of positively stained tissue within the glomerular tuft. On average 20 randomly selected hilar glomerular tuft cross-sections were assessed per mouse.

Cultured cell

Podocytes were fixed in 4% w:v PFA for 15 min. To avoid disruption of existing TNTs during fixation, 16% w:v PFA was added along the sides of the plates, keeping the overlying culture medium intact, to a final w:v concentration of 4%. Cells were then permeabilized using 0.5% v:v Triton X-100 (Merck Life Science, 9002–93-1) for 5 min at 4°C. Cultures were then blocked with 3% BSA and incubated overnight at 4°C with a mouse anti-SQSTM1/p62 (Santa Cruz Biotechnology, sc-48,402; 1:100) or rat anti-LAMP1 (Santa Cruz Biotechnology, sc-19,992; 1:100) primary antibody. After washing, FITC-conjugated goat anti-mouse (Thermo Fisher Scientific, A16079; 1:50) or goat anti-rat (Merck Life Science, F6258; 1:50) secondary antibodies were added for 1 h. FITC-SQSTM1/p62 labeled cells were, then, incubated overnight at 4°C with a rabbit anti-MAP1LC3/LC3 (Cell Signaling Technology, 43,566; 1:100) followed by incubation with tetramethylrhodamine (TRITC)-conjugated pig anti-rabbit (DAKO, R0156; 1:50). FITC-LAMP1-labeled cells were incubated for 1 h at room temperature with either a rabbit anti-MAP1LC3/LC3 or a goat anti-CTSD (R&D Systems, AF1029; 1:100) antibody, followed by incubation with TRITC-conjugated pig anti-rabbit or Alexa Fluor® 555-conjugated donkey anti-goat (Thermo Fisher Scientific, A21432; 1:1000) antibodies, respectively. Cell nuclei were counterstained with a 4′,6-diamidino-2-phenylindole (DAPI) solution. Digitalised images were color-combined and assembled into photomontages by using Adobe Photoshop (Universal Imaging Corporation).

Immunoblotting

Equal amounts of protein samples were separated on sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis and electro-transferred to nitrocellulose membranes. Following blocking, membranes were incubated with anti-TNFAIP2 or anti-MAP1LC3/LC3 primary antibodies overnight at 4°C. After washing, secondary HRP-linked antibodies (GE Healthcare, RPN4301; 1:10,000) were added for 1 h. Detection was performed using Super signal PICO (Thermo Fisher scientific, 34,580) and visualized on a Gel-Doc system (Bio-Rad, Italy). Band intensities were quantified by densitometry. TUBB/beta-tubulin (Santa Cruz Biotechnology, sc-9104; 1:1000) was used as a loading control.

Statistical analysis

Data are presented as mean ± SEM, geometric mean (25°-75° percentile), or fold change over control. Non-normally distributed variables were log transformed prior to analyses. Data were analyzed using the t test or ANOVA as appropriate. Least significant difference test was used for post hoc comparisons. P values <0.05 were considered statistically significant.

Funding Statement

Barutta F. was the recipient of a Juvenile Diabetes Research Foundation Postdoctoral Fellowship (3-PDF-2014-109-A-N). The work was supported by the European Foundation for the Study of Diabetes (EFSD), the Ferrero Foundation (Alba), the University of Turin (ex. 60% grant) and the Italian Ministry for Education, University and Research (Ministero dell’Istruzione, dell’Università e della Ricerca - MIUR) under the program “Dipartimenti di Eccellenza 2018–2022” project D15D18000410001.

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplementary Material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/15548627.2022.2080382