Male C57BL/6J mice were purchased from Shanghai Model Organisms Center, Inc. (Shanghai, China) and housed at Fudan University Medical Animal Center with a 12-hour light/dark cycle and provided with standard chow. All animal studies were approved by the Institutional Animal Care and Use Committee of Fudan University. 8-week-old male mice with a C57BL/6J background were intraperitoneally injected with streptozotocin (STZ) (Sigma-Aldrich, 50mg/kg, diluted in citrate buffer, pH = 4.5) for five consecutive days to establish STZ induced diabetic model until 28 weeks after STZ administration when they were sacrificed. Littermates were conducted with the same volume of buffer as control. Fasting plasma glucose (FPG) levels were measured via tail-vein blood using an ACCU-CHEK PERFORMA II glucometer after a 7-hour fast. Mice exhibiting FPG ≥ 200 mg/dL 1week post-STZ injection were considered diabetic. Spot urine samples were used to quantify albumin and creatinine with a commercial ELISA kit (Exocell, Inc. USA), following the manufacturer’s protocol. The urinary albumin to creatinine ratio (UACR) was determined by albumin level and creatinine level from the same samples. Blood urea nitrogen (BUN) was measured using the UREA KIT (liquid; UV-GLDH method; Shanghai Kehua Bioengineering Co., Ltd.). Kidney samples were collected when mice were sacrificed. Blood pressure was measured with the tail-cuff method using a noninvasive automatic blood pressure analyzer (BP-2000 Blood Pressure Analysis System, Visitech Systems, USA) according to the manufacturer’s instructions.

Kidney tissue was extracted and then measured using an NADH/NAD quantification kit (Biovision, USA) according to the manufacturer’s instructions. In brief, ~ 20 mg kidney was washed with PBS, pelleted, and extracted with 400 μL NADH/NAD extraction buffer. 50 μL of extracted samples were used for total NADt detection. Samples were decomposed at 60°C for 30 minutes, cooled, and finally 50 μL transferred to a 96-well plate to measure NADH. A standard curve was prepared with 0, 20, 40, 60, 80, and 100 pmol/well of standard NADH, with the final volume adjusted to 50 μL with NADH/NAD extraction buffer. NAD Cycling Mix composed of NAD cycling buffer and NAD cycling enzyme mix was added to each well at room temperature for 5 minutes to convert NAD to NADH. NADH developer was then added, and the plate was cycled at room temperature for 1–4 hours. OD 450 nm readings were taken, and reactions were stopped with a stop solution. The NADt amount in a sample well was calculated with the equation:

Kidneys were fixed in a 4% paraformaldehyde solution, embedded in paraffin, cut into sections with a thickness of 3 μm and then subjected to periodic acid–schiff (PAS) staining. In brief, kidney sections were dewaxed in water, oxidized in 1% periodic acid solution (BaSO, China) for 10 minutes, placed in Schiff’s reagent (BaSO, China) for 20 minutes and then counterstained in Mayer’s hematoxylin for 10 seconds (BaSO, China). The area of mesangium and glomeruli was calculated by pixel counts on a minimum of 10 randomly selected glomeruli per kidney section by ImagePro Plus software in a single-blind fashion.

Total protein was extracted from the frozen quartered kidney tissue using RIPA lysis buffer (Beyotime, China) with PMSF (Sigma-Aldrich, USA), protease inhibitor cocktail (Roche, USA) and phosphatase inhibitor cocktail (Roche, USA). Protein concentrations were determined using BCA protein assay kit (Beyotime, China). Equal amount of protein (~20 μg per lane) was loaded in a 7.5% or 10% SDS-PAGE mini-gel and transferred to a PVDF membrane (Millipore, USA). The membrane was blocked with 5% BSA in TBST buffer (100 mM TBS, pH 7.5, 0.1% Tween-20) or 5% nonfat dry milk dissolved in TBST buffer and then incubated in primary antibody overnight at 4°C. The primary antibodies included: anti-SIRT1 antibody (Abcam, 1:1000), anti-CD38 antibody (Proteintech, 1:1000), anti-PARP1 antibody (Proteintech, 1:1000), anti-KMO antibody (Proteintech, 1:1000), anti-QPRT antibody (Proteintech, 1:1000), anti-NAMPT antibody (Millipore, 1:5000), anti-GAPDH antibody (Proteintech, 1:8000). Membranes were then incubated with appropriate secondary antibodies for one hour at room temperature and subjected to chemiluminescence detection using ECL Reagent (Millipore,USA).

The isolation of RNA and real-time PCR were performed as previously described (26). In brief, frozen quartered kidney tissue were processed with TRIzol (Thermo Fisher, USA), chloroform, isopropanol, 70% ethanol, and finally RNA suspension in RNase-free water. PrimeScript™ RT reagent Kit (Takara, Japan) was used to perform cDNA synthesis according to the manufacturer’s protocol. Levels of mRNA were determined by real time qPCR using SYBR Premix Ex Taq Kit (Takara, Japan). The expression levels of the target genes were normalized to GAPDH using the 2^−ΔΔCt method.

The primer sets used are shown in Table 1 .

Kidneys were fixed in 4% paraformaldehyde solution, embedded in paraffin, and cut into sections with a thickness of 3 μm. Sections were dewaxed in water and placed in an EDTA antigen retrieval buffer (pH 8.0) for thermal repair. They were then incubated in a 3% H2O2 solution for 30 minutes. For blocking, 3% bovine serum albumin in PBS was used, followed by overnight incubation with primary antibodies at 4°C. Next, a secondary antibody (HRP) was added, and the sections were incubated at room temperature for 30 minutes before diaminobenzidine (DAB) staining controlled under a microscope. The positive staining appeared brownish-yellow in the first round. The second round of immunohistochemistry staining followed the same procedure: antigen retrieval, endogenous peroxidase blocking, and BSA blocking. Target primary antibodies were incubated at 4°C overnight followed by secondary antibody incubation at room temperature for 30 minutes. Finally, a red staining solution working solution (Shanghai RecordbioTechnology Co. Ltd, PB: CU: Red dye: AC= 860:40:1:100) was added and controlled under the microscope. Positive staining in the second round appeared red. Immunofluorescence was conducted on paraffin-embedded kidney sections as previously described (27). The primary antibodies used in immunohistochemistry are as follows: anti-KMO antibody (Proteintech, 1:200), anti-QPRT antibody (Proteintech, 1:200), anti-NCC antibody (Abcam, 1:1000), anti-THP antibody (Santa Cruz, 1:200), anti-AQP2 antibody (Santa Cruz, 1:200). Primary antibody used in immunofluorescence are as follows: anti-KMO antibody (Proteintech, 1:200), anti-QPRT antibody (Proteintech, 1:200), anti-LTL antibody (Vector Laboratories, 1:200).

Human urinary KYN was quantified using LC-MS. The procedure and specifics of LC-MS conditions were detailed as described in reference (24).

The NephroSeq database (www.nephroseq.org) was used to examine the mRNA level of KMO gene. The Lindenmeyer Normal Tissue Panel dataset was employed to discern its distribution in renal compartments among healthy kidney transplant donors. Three datasets were available for the analysis of KMO expression levels between healthy kidney transplant donors and DKD patients. These datasets include Woroniecka Diabetes Glom Dataset, Woroniecka Diabetes TubInt Dataset, and Schmid Diabetes TubInt Dataset.

All data were presented as the mean ± SEM. Statistical comparisons were performed using unpaired Student’s t test or ANOVA. GraphPad Prism 8.0 software was used for statistical analysis. A value of P< 0.05 indicated a statistically significant difference.

To delineate alterations in NAD+ levels within diabetic kidneys, we successfully established a diabetic model of type 1 diabetes induced by STZ ( Figures 1B, C ). Compared to control littermates, NAD+ levels were significantly reduced in diabetic mice kidneys 29 weeks post-STZ administration ( Figure 1D ). Renal impairment was evaluated by ACR of spot urine, BUN and histological examination. As depicted in Figures 1E, F , diabetic mice showed increased urine ACR, a sensitive and early indicator of DKD, independent of changes in blood pressure. Although the kidney function biomarker, BUN, showed an increasing trend in the diabetic mice ( Figure 1G ), no statistically significant difference was noted. Histological lesions examined by PAS staining showed more mesangial matrix expansion in diabetic group ( Figure 1H ). Our data suggested mild and early renal impairment was induced by STZ in C57 mice, and NAD+ levels was reduced in these diabetic kidneys.

The level of NAD+ is contingent upon a delicate balance between synthesis and consumption (28). In order to explore the underlying causes of the NAD+ decline in diabetic kidneys, we initially examined the levels of NAD+-consuming enzymes. In our study, although a decreasing trend of SIRT1 and PARP1 protein levels was observed in western blot analysis, no significance was calculated. Similarly, there was no change of CD38 levels in diabetic conditions ( Figure 2 ). NAD+-dependent deacetylase SIRT1 was reported to be reduced in high energy environment as diabetic kidneys, leading to transcription factor acetylation and kidney lesion (29). SIRT1 has a high Km value for NAD+ compared to other NAD+ consuming enzymes, meaning SIRT1’s activity is highly dependent on NAD+ availability but contributes little to NAD+ consumption (30). Since indifferent expression of NAD+-consuming enzymes could not lead to NAD+ deficiency, then, we assessed the changes of key enzymes in different NAD+ synthesis pathways in diabetic kidneys by mRNA quantification. Compared to the control, the mRNA levels of KMO, a rate-limiting enzyme that transfer KYN into 3-HK, and quinolinate phosphoribosyl transferase (QPRT), an enzyme that catalyze quinolinic acid (QA) into NMN in the de novo synthesis pathway, were decreased after STZ administration ( Figure 3A ). In the salvage pathway, no significant decrease of NAMPT nor NMNAT1 were observed in diabetic kidneys, with the exception of upregulation of NMNAT3. These findings suggested a contributory role of de novo pathway in NAD+ dynamics when exposed to diabetes. WB analysis further confirmed the reduction of KMO protein in the kidneys of the diabetic group ( Figure 3B ). Taken together, our data implied that the downregulation of KMO in de novo pathway might lead to a decrease in NAD+ levels.

Considering the potential value of KMO in DKD, we further investigated the localization of KMO within the mouse kidney. Immunohistochemical co-staining results revealed that KMO did not colocalize with aquaporin 2 (AQP2), a recognized marker of the collecting duct, nor with sodium-chloride cotransporter (NCC), a marker for the distal convoluted tubule, or with Tamm-Horsfall protein (THP), a marker of thick ascending limb of the renal medulla ( Figure 4A ). KMO was predominantly expressed in the cytoplasm of the proximal tubule, as evidenced by immunofluorescence analysis with Lotus tetragonolobus lectin (LTL) ( Figure 4B ). Moreover, QPRT and 3-hydroxyanthranilic acid oxygenase (HAAO), two other pivotal enzymes with predictive value of AKI in the de novo pathway (21, 22, 24), were also present in the same pattern. These observations indicated that the proximal tubule may serve as the primary site for the de novo synthesis pathway of NAD+ in the mouse kidneys.

Our study suggested that the diminished NAD+ levels in diabetic mouse kidneys may be due to a reduction of KMO in the de novo synthesis pathway. To explore the distribution of KMO in human kidneys along with its alteration in DKD patients, we utilized the NephroSeq dataset to analyze the transcriptional level of KMO. A total of four datasets, the patient characteristics of which are detailed in Tables 2 – 5 , were available for analysis. As illustrated in Figure 5A , microarray data form the Lindenmeyer Normal Tissue Panel dataset revealed the expression of KMO in both glomeruli and tubulointerstitium, with a more pronounced expression observed in the tubulointerstitium, suggesting the tubulointerstitium as a pivotal site for NAD+ de novo synthesis in human kidney align with the mouse data. A significant decrease in KMO was found in DKD patients when compared to healthy living donors in isolated glomeruli ( Figure 5B ). However, in the tubulointerstitium region, although a downward trend in KMO expression was observed across two datasets, the changes were not statistically significant, likely attributed to the limited sample size ( Figures 5C, D ). Analysis of KMO according to eGFR level from Schmid Diabetes TubInt dataset showed a KMO decreased as eGFR fell ( Figure 5E ). Furthermore, an accumulation of urinary KYN, a metabolite upstream of KMO, was observed in diabetic population by LC-MS ( Figure 5F ). These data altogether indicated that renal KMO was decreased in diabetic kidney and played a potential role in pathogenesis of DKD.