Hypoxia induced MT2 and MT3 in HRPTECs

We evaluated the effects of hypoxia on the expressions of MT1, 2 and 3 in HRPTECs. Hypoxia significantly decreased MT1 (0.63-fold vs. normoxia control, p < 0.0001) and increased MT2 (2.12-fold, p < 0.01) and MT3 (78.28-fold, p < 0.01) levels in HRPTECs (Fig. 1A). Given the high hypoxia-induced increment in MT3 expression in HRPTECs, we focused on the role of MT3 in hypoxia-involved DN in the current study.

Fig. 1

Identification of MT3 as hypoxia-induced gene dependent on HIF-1α. A qRT-PCR showed that hypoxia induced MT2 and MT3, not MT1, in HRPTECs. **p < 0.01, ****p < 0.0001 vs. the normoxia control. B HRPTECs were transiently transfected with a negative control, MT3-specific siRNAs (25 nmol/L final level). Twenty-four hours after transfection, the cells were cultured under normoxic (21% O2) and hypoxic (1% O2) conditions overnight. Real-time reverse transcriptase PCR (RT-PCR) (upper panel) and conventional RT-PCR (lower panel) with two primer sets (Supplementary Table 1) revealed that MT3 siRNA efficiently decreased MT3 expression in HRPTECs. C Gene expression microarray analysis of MT3-knockdown HRPTECs under hypoxia. To determine which molecules are regulated by MT3 in human renal proximal tubular cells, we analyzed microarray data via an Affymetrix GeneChip (Human Gene 1.0 ST Array) with siRNA-mediated knockdown of MT3 under hypoxia in HRPTECs

MT3 siRNA decreases the expression of DN-related genes in HRPTECs

We confirmed that hypoxia significantly induced MT3 expression in HRPTECs, as evaluated via qRT-PCR and conventional RT-PCR (Fig. 1B). We compared gene expression between the control and MT3-knockdown HRPTECs under hypoxic conditions via Affymetrix Human Gene 1.0 ST arrays (Fig. 1C). DNA microarray analysis revealed a − 1.5638772-fold change in MT3 expression in MT3 siRNA-treated HRPTECs compared with that in negative control siRNA-treated HRPTECs; thus, the threshold for linear fold change absolute values was set to > 1.5 to cover both sides, as shown in Table 1. Among the downregulated genes, we focused on the top 3 downregulated genes: ceruloplasmin (CP), cytochrome b reductase 1 (CYBRD1), and fibroblast growth factor receptor 2 (FGFR2).

Table 1 The changed over 1.5-fold by MT3 siRNA under hypoxic condition in HRPTECsHIF-1α and Nrf2 regulate MT3 under hypoxia in HRPTECs

To confirm the results of the microarray analysis, we analyzed the regulation of MT3 expression via qRT-PCR. MT3 mRNA expression was significantly upregulated by 21.13-fold in HRPTECs under hypoxia compared with control siRNA-treated HRPTECs under normoxia (Fig. 2A). Hypoxia induces ROS, which activate transcription factors, such as hypoxia-inducible factor 1α (HIF-1α) and Nrf2 [19]. Because HIF-1α can bind to the HRE, we screened a region of the promoter region of the human MT3 gene and searched for transcriptional start sites for HIF via JASPAR (profile score threshold 80%, https://jaspar.elixir.no/). The JASPAR analysis revealed potential binding sites for HIF1A in the promoter of human MT3, indicating that human MT3 could be a direct transcriptional target of HIF (Supplementary Table 5). We also detected fewer putative site(s) for HIF1A in the promoter region of the mouse Mt3 gene than human MT3 (Supplementary Table 6). Because of CTG triplet repeats in the promoter region, which function as negative elements, hypoxia failed to induce expression of the Mt3 gene in cultured mouse proximal tubular cells (Supplementary Fig. 1).

Fig. 2

Regulation of MT3 expression and downstream genes in human renal proximal tubular epithelial cells (HRPTECs). A Hypoxia significantly induced MT3 expression, and MT3 siRNA efficiently decreased MT3 expression in HRPTECs. MT3 siRNA significantly inhibited hypoxia-induced CP B and CYBRD1 C expression. D Hypoxia did not significantly increase FGFR2 expression because of wide variation. MT3 siRNA tended to decrease FGFR2 expression independent of oxygen conditions. E Oxygen conditions did not affect KL expression, but MT3 siRNA significantly and equally decreased KL expression under normoxia and hypoxia. F In contrast to the results for the genes described above, hypoxia decreased FGF23 expression and MT3 siRNA increased FGF23 expression. G Oxygen conditions did not change KLB expression in the same manner as KL expression, but MT3 siRNA increased KLB expression but not KL expression. H Hypoxia inducible factor (HIF)−1α siRNA efficiently inhibited HIF1A compared with negative siRNA under normoxic condition. As hypoxia alone significantly inhibited HIF1A compared with those in normoxic conditions, HIF-1α siRNA tended toward decrease in HIF1A expression under hypoxia. I HIF1A siRNA drastically decreased hypoxia-induced MT3 expression. Except for FGFR2 expression L, HIF1A siRNA did not decrease CP J, CYRBD1 K, or KL M expression. N Hypoxia significantly decreased NFE2L2 expression. Conformingly, Nrf2 siRNA decreased NFE2L2 expression. Nrf2 siRNA also decreased MT3 O, CP P, CYBRD1 Q, FGFR2 R, and KL S expression. Statistical comparisons were performed via one-way analysis of variance ANOVA with post hoc Bonferroni multiple comparison tests. *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001

Accordingly, compared with control siRNA, HIF-1α siRNA markedly decreased MT3 expression under hypoxia in HRPTECs (Fig. 2I). Nrf2 is another major redox-responsive molecule that regulates genes involved in oxidative stress and inflammation. MT3 expression was also significantly downregulated by Nrf2 siRNA under hypoxia (Fig. 2O). In addition, the antioxidant N-acetylcysteine (NAC) failed to restore hypoxia-induced MT3 expression, whereas the mitochondrial respiratory inhibitor NADPH oxidase (NOX) abolished the stimulatory effect of hypoxia on MT3 expression (Supplementary Fig. 2), indicating that hypoxia-induced MT3 expression is dependent on the mitochondrial respiratory system or NOX.

MT3 and Nrf2, but not HIF-1α, siRNAs decrease CP, CYBRD1 and KL expression in HRPTECs

Consistent with the results of the microarray analysis, MT3 siRNA significantly decreased hypoxia-induced CP expression (Fig. 2B). Regarding CYBRD1 expression, hypoxia produced various effects because of its limited ability and the use of primary cultured cells derived from different human donors (Figs. 2C and 3C). Hypoxia failed to induce FGFR2 expression, and MT3 siRNA just tended to decrease FGFR2 expression (Fig. 2D). We also analyzed the mRNA expression of the FGF–Klotho–FGFR complex. MT3 siRNA significantly decreased KL expression under both normoxia and hypoxia (Fig. 2E). In contrast, MT3 siRNA significantly increased FGF23 expression under hypoxia (Fig. 2F) and the expression of β-klotho (KLB), which shares 41.2% amino acid identity with KL, under both normoxia and hypoxia (Fig. 2G), suggesting the potential regulation of FGF‒KL‒FGFR complexes, such as mutual negative feedback.

Fig. 3

Effects of mimicking the diabetic microenvironment on the expression of MT3 and its downstream genes in human renal proximal tubular epithelial cells (HRPTECs). High glucose (D-glucose, 25.5 mM) significantly decreased hypoxia-induced CP expression B but failed to decrease MT3 A, CYBRD1 C, KL D, or FGFR2 E expression. In contrast, palmitic acid (PA; 150 μM) decreased MT3 F, CP G, CYBRD1 H, FGFR2 I, and KL J expression. K Oleic acid (OA; 150 μM) did not affect MT3 expression under normoxia or hypoxia. Statistical comparisons were performed via one-way analysis of variance (ANOVA) with post hoc Bonferroni multiple comparison tests. *p < 0.05, **p < 0.01, ***p < 0.001 and ****p < 0.0001

HIF-1α siRNA significantly decreased MT3 expression, but not CP, CYBRD1 or KL expression (Fig. 2J, K, M), indicating that these three genes are directly regulated by MT3, not HIF-1α, under hypoxia. On the other hand, Nrf2 siRNA significantly decreased CP and CYBRD1 expression under hypoxia (Fig. 2P and Q) and KL expression under both normoxia and hypoxia (Fig. 2S). In contrast, Nrf2 siRNA significantly increased FGFR2 expression under normoxia (Fig. 2R), suggesting the complicated regulation in FGFR/KL complexes.

PA, not oleic acid or high glucose, inhibits hypoxia-induced MT3 mRNA in HRPTECs

To study the effects of glucotoxicity and lipotoxicity in diabetes, we analyzed MT3, CP, CYBRD1, FGFR2, and KL expression under conditions of high glucose (HG; 25.5 mM) and cytotoxic saturated free fatty acid (FFA) palmitic acid (PA) (150 μM) supplementation (Supplementary methods). HG conditions did not affect the expression of MT3, CYBRD1, FGFR2 or KL (Fig. 3A, C–E) but did affect the expression of CP under hypoxia (Fig. 3B). In contrast, PA supplementation decreased the expression of MT3, CP, CYBRD1, and FGFR2 under hypoxia (Fig. 3F–I) and tended to decrease KL expression under hypoxia (p = 0.0883) (Fig. 3J). In contrast, supplementation with oleic acid, an unsaturated fatty acid, did not change MT3 expression (Fig. 3K).

STZ-induced diabetic MT3-BACTg mice present increased plasma and urinary zinc levels

We then investigated the effects of MT3 on DN using STZ-induced diabetic MT3-BACTg mice (Fig. 4). The STZ-induced diabetic MT3-BACTg mice presented increased fasting blood glucose levels (p < 0.001), HbA1c levels (p < 0.001), water intake (p < 0.05), and food intake (p < 0.05) but decreased body weight (p < 0.001). Diabetes did not change SBP. Compared with control mice, diabetic mice presented polyuria (p < 0.0001), increased urinary albumin excretion (p < 0.01), and urinary neutrophil gelatinase-associated lipocalin (NGAL) excretion (p < 0.001) concomitant with hyperglycemia and polydipsia. The plasma cystatin C level was decreased (p < 0.01) in the STZ group compared with the control group, indicating hyperglycemia-induced hyperfiltration. Additionally, urinary zinc excretion (p < 0.01) and the plasma zinc level (p < 0.001) were increased in the STZ group compared with the control group.

Fig. 4

Streptozotocin-induced diabetic humanized bacterial artificial chromosome transgenic mice (MT3-BACTg mice) presented increased plasma zinc levels and hyperzincuria accompanied by diabetic nephropathy. A Fasting blood glucose (FBG). B Glycated hemoglobin (HbA1c). C Water intake for 24 h. D Food intake for 24 h. E Body weight. F Systolic blood pressure (SBP). G Urinary volume for 24 h. H Urinary albumin per day. I Urinary NGAL per day. J Plasma cystatin C. K Urinary zinc per day. L Plasma zinc level. Black circles: control MT3-BACTg (n = 10). Black squares: streptozotocin-induced diabetic MT3-BACTg mice (n = 7). Statistical comparisons were performed via Student’s t tests or the Mann‒Whitney U test. Urinary albumin and NGAL were log(e)transformed for statistical analysis. *p < 0.05, **p < 0.01, ***p < 0.001 and ****p < 0.0001

MT3 expression is increased in the renal tubules of diabetic mice, accompanied by CP, CYRBD1, FGFR2, and KL protein expression

Periodic acid-Schiff (PAS) staining revealed that STZ-induced diabetic MT3-BACTg mice exhibited mesangial expansion, nodular glomerular sclerosis, and proximal tubular injury, as shown by STZ-induced diabetic renal histological changes (Fig. 5A d, g, h, j). Intriguingly, MT3-BACTg mice presented dilated peritubular capillaries surrounding the glomeruli (Fig. 5A e, f). Diabetes increased the concomitant expression of the MT3, CP, CYRBD1, FGFR2, and KL proteins in renal tubules (Fig. 5B). Unexpectedly, no significant difference was observed in HIF-1α protein expression in control and diabetic group (Fig. 5B).

Fig. 5

Histology of kidneys from humanized bacterial artificial chromosome transgenic mice (MT3-BACTg mice) with diabetic changes. A Compared with control MT3-BACTg mice (control, a), Periodic acid-Schiff (PAS) staining revealed that STZ-induced diabetic MT3-BACTg mice (STZ) at 15 weeks of age exhibited mesangial expansion and tubular dilation (b). Intriguingly, staining for the endothelial cell marker CD31 revealed peritubular capillary dilatation in STZ (e). In addition, PAS staining shows nodular glomerular sclerosis (black arrows) (g, h) and arteriolar and vascular pole hyalinosis (black asterisks) (i, j) in diabetic MT3-BACTg mice. Scale bars, 200 μm in lower magnified and 20 μm in higher magnified photomicrographs. B Compared with control mice, STZ-induced diabetic MT3-BACTg mice presented significantly increased protein expression of MT3 and its regulatory molecules. Statistical comparisons were performed via the Mann‒Whitney U test. *p < 0.05, **p < 0.01, ***p < 0.001. C Representative electronic microscopy images of the kidneys of streptozotocin-induced diabetic mice. Diabetes induced thickening of the glomerular basement, fusion of the podocyte foot processes, mitochondrial swelling, and a decrease in the number of mitochondrial cristae in both groups of eight-month-old mice. PAS staining presented more severe tubular injury in diabetic MT3-BACTg mice compared with those in wild-type (WT) mice. Scale bars: black, 5 μm; white, 1 μm, blue; 20 μm

MT3 aggravates diabetic renal injury associated with mitochondrial injury

To confirm the effects of MT3 on DN, we investigated the effects of MT3 on established STZ-induced DN via transmission electron microscopy (TEM) in wild-type littermates and MT3-BACTg mice at 8 months of age (Fig. 5C). Notably, the MT3-BACTg mice presented greater variations in mitochondrial size, the absence of basal infoldings, total disorientation of the mitochondria within the cell cytoplasm, and a decrease in the number of mitochondria compared with wild-type littermates (Fig. 5C), accompanied with more severe tubular injury shown in PAS staining. Thus, sustained higher level of MT3 might promote progression of DN.

STZ-induced MT3-BACTg mice present increased renal expression of MT3, not Mt3, accompanied by increased Cp, Cyrbd1, Fgfr2, and Kl

Next, we analyzed gene expression in the renal cortex of MT3-BACTg mice via qRT-PCR. MT3, Cp, Cybrd1, Fgfr2, and Kl expression was significantly increased in the STZ group compared with the control group (Fig. 6A, C–F). Mt3 expression did not significantly differ between the control group and the STZ group (p = 0.97) (Fig. 6B).

Fig. 6

The expression of MT3, Mt3, Cp, Crbrd1, Fgfr2, and Kl in the cortex of kidneys from humanized bacterial artificial chromosome transgenic mice (MT3-BACTg mice) as determined via quantitative reverse transcriptase PCR (RT-PCR). A MT3 expression was significantly increased in the streptozotocin (STZ)-induced diabetic group. B Mt3 expression was not different between the control group and the STZ-induced diabetic group. Compared with control mice, STZ-induced diabetic mice presented greater expression of Cp C, Crbrd1 D, Fgfr2 E, and Kl F. Statistical comparisons were performed via Student’s t test. *p < 0.05, **p < 0.01, ***p < 0.001 and ****p < 0.0001. Urinary albumin excretion was statistically correlated with renal cortical Mt3 H, Cp I, Cyrbd1 J, Fgfr2 K, and Kl L expression but not with MT3 expression G. Urinary NGAL excretion was statistically correlated with renal cortical Cp O, Cyrbd1 P, Fgfr2 Q, and Kl R expression but not with MT3 M or Mt3 N expression. Plasma zinc levels were statistically correlated with renal cortical MT3 S and urinary NGAL excretion V but not with Mt3 T or urinary albumin excretion U. Urinary zinc levels were statistically correlated with urinary albumin excretion W and urinary NGAL excretion X. The relationships between two variables were analyzed via simple linear regression. p values < 0.05 were considered significant. Black circles: control MT3-BACTg mice (n = 10). Black squares: STZ-induced diabetic MT3-BACTg mice (n = 7)

MT3 expression was not correlated with urinary excretion of albumin (Fig. 6G) or NGAL (Fig. 6M). Moreover, Mt3 expression was not correlated with urinary NGAL excretion (Fig. 6N), a marker of renal tubular injury. Nevertheless, Cp, Cybrd1, Fgfr2, and Kl expression was significantly correlated with urinary albumin excretion (Fig. 6H–L) and urinary NGAL excretion (Fig. 6O–R).

Plasma zinc levels are correlated with renal expression of MT but not Mt3

To determine the effects of MT3 on zinc metabolism in STZ-induced diabetic mice, we evaluated the correlation between zinc levels and renal function in control and STZ-induced diabetic MT3-BACTg mice. Plasma zinc levels were correlated with renal MT3 (p < 0.05, Fig. 6S) and urinary NGAL excretion (p < 0.001, Fig. 6V) but not with renal Mt3 (p = 0.2043, Fig. 6T) or urinary albumin excretion (p = 0.2197, Fig. 6U). In contrast, urinary zinc levels were not correlated with MT3 or Mt3 expression (Supplementary Fig. 3). Urinary zinc levels were significantly correlated with urinary albumin excretion (p < 0.05, Fig. 6W) and urinary NGAL excretion (p < 0.0001, Fig. 6X).

Aged MT3-BACTg mice present renal diabetic lesions, as observed in diabetic subjects

Despite its role as an inducer of KL, the MT3 transgene showed no antiaging effects on MT3-BACTg mice (Supplementary Fig.  4 A). To confirm the effects of the MT3 transgene on aged kidneys without diabetes, we studied MT3-BACTg mice at age 2 years. Intriguingly, aged MT3-BACTg mice presented no significant difference in glucose metabolism compared to aged wild-type mice (Supplementary Fig. 4B, C), which was accompanied by glomerular nodules such as those associated with DN (Fig. 7A). The glomerular nodules in MT3-BACTg mice were negative for phosphotungstic acid-hematoxylin stain (PTAH), which indicates fibrin deposition (Fig. 7A). Periodic acid methenamine silver stain (PAM) presented mesangiolytic nodular lesions [23] in the kidneys of aged MT3BACTg mice (Fig. 7A). The mesangium was severely expanded with nodule formation, but no electron-dense deposits were observed in the glomerular nodules (Fig. 7B, a). The number of mesh-like endoplasmic reticula was increased in the proximal tubules, indicating cellular injury [24] (Fig. 7B, c). In addition, TEM revealed the obstruction of downstream peritubular capillaries (Fig. 7B, d), which leads to glomerular hypertension. Continuous MT3 expression might present more severe DN via retrograde glomerular hypertension.

Fig. 7

Aging induces severe kidney injury in MT3-BACTg mice. A PAS staining revealed that compared with diffuse mesangial expansion in the kidney of wild-type littermates (a, b), aged male MT3-BACTg mice with MT3 overexpression in renal tubules presented nodular glomerulosclerosis, such as Kimmelstiel–Wilson lesions (asterisk) and doughnut lesions (arrows) (c, d). Scale bars: 300 μm in a, c, 10 μm in b, d. MT3 immunopositive stains were present in the nodular lesion (e, arrow) and in the mesangial lesions (f). Scale bars: 20 μm. The glomerular nodules in MT3-BACTg mice were negative for PTAH (g) and positive for PAM (h, arrow). Scale bars: 50 μm in a, 10 μm in c. B Micrographs of MT3-BACTg glomeruli and proximal tubules at 2 years of age obtained via electron microscopy. a The mesangium was severely expanded with nodule formation. No electron-dense deposits were found. b Glomerular basement membranes (GBMs) were often severely thickened. c The number of endoplasmic reticula increased in the proximal tubules. d The swelling of endothelial cells caused narrowing of the capillary lumen. e, f The proximal tubules presented fewer abnormally shaped mitochondria, accompanied by the ubiquitous presence of the endoplasmic reticulum. f Magnified image of the decrease in the number and size of mitochondria in aged MT3-BACTg proximal tubules. Representative transmission electron micrograph. Scale bars: black, 5 μm; white, 2 μm; gray, 500 nm. PAS, periodic acid–Schiff. PTAH, phosphotungstic acid-hematoxylin. PAM, periodic acid methenamine

MT3 inhibited HIF-1α expression in renal proximal tubular cells in vitro and in vivo

The rat remnant kidney model identified the characteristic peritubular capillary rarefaction along with decreased interstitial VEGF expression. VEGF is one of HIF-1-targeted genes. To clarify the mechanism which induced the obstruction of peritubular capillary observed in TEM study, we studied the effects of MT3 on HIF-1α expression using HRPTECs. MT3siRNA significantly increased hypoxia-induced HIF-1α proteins accompanied with HIF1A increment (Fig. 8A, B and C). To validate the effects of MT3 on HIF-1α protein in vivo, we used proximal tubule-specific overexpressed human MT3 transgenic mice (MT3Tg). We found no overlap protein expression of HIF-1α and MT3 in the proximal tubules of diabetic MT3Tg (Supplementary Fig. 5). Overexpression of MT3 also failed to prevent diabetic nephropathy in STZ-induced MT3Tg mice as seen in MT3-BACTg mice (Supplementary Table 7).

Fig. 8

Human MT3 inhibits hypoxia-induced HIF-1α protein expression in HRPTECs. A MT3 siRNA increased hypoxia-induced HIF-1α protein expression. B Densitometric analysis showed that hypoxia significantly induced HIF-1α protein expression compared with that in a negative control (nega) siRNA in normoxia and that MT3 siRNA significantly increased hypoxia-induced HIF-1α protein expression to that of a negative control (nega) siRNA in hypoxia. C Hypoxia significantly inhibited HIF1A compared with that in normoxia. MT3siRNA significantly increased HIF1A in normoxia and hypoxia compared with that in the control. One-way repeated-measures ANOVA with Bonferroni's multiple comparison post hoc tests, **p < 0.01, ****p < 0.0001. D Interplay between MT3 and HIF-1α in human diabetic nephropathy. Chronic hypoxia-induced MT3 expression might inhibit HIF-1α/VEGF expression in renal proximal tubules, resulting in peritubular capillary injury which causes retrograde glomerular hypertension. E HIF-1α expressions were inversely correlated with MT3 expressions in renal proximal tubular cells in both MT3-BACTg and MT3Tg mice

Relationship between the GFR and MT3 gene expression

To explore the clinical significance of MT3 in DN, we accessed the Nephroseq database (https://www.nephroseq.org/) and obtained relevant microarray data from the ‘Woroniecka Diabetes Tubint’, which includes 22 human diabetic kidney disease samples. In DN patients, correlation analysis revealed a positive correlation between MT3 mRNA expression in the renal tubulointerstitium and the glomerular filtration rate (GFR) (r = 0.650, p = 0.001; Supplementary Fig. 6), revealing that MT3 may increase in the early stage of DN accompanied with relative hypoxia, which is revealed by increased oxygen consumption.