Deletion of heterogeneous nuclear ribonucleoprotein F in renal tubules downregulates SGLT2 expression and attenuates hyperfiltration and kidney injury in a mouse model of diabetes

Abstract

Aims/hypothesis

We previously reported that renal tubule-specific deletion of Heterogeneous nuclear ribonucleoprotein F (Hnrnpf) results in upregulation of renal angiotensinogen (Agt) and downregulation of sodium-glucose co-transporter 2 (Sglt2) in Hnrnpf^RT knockout (KO) mice. Non-diabetic Hnrnpf^RT KO mice develop hypertension, renal interstitial fibrosis and glycosuria with no renoprotective effect from downregulated Sglt2 expression. Here, we investigated the effect of renal tubular Hnrnpf deletion on hyperfiltration and kidney injury in Akita mice, a model of type 1 diabetes.

Methods

Akita Hnrnpf^RT KO mice were generated through crossbreeding tubule-specific (Pax8)-Cre mice with Akita floxed-Hnrnpf mice on a C57BL/6 background. Male non-diabetic control (Ctrl), Akita, and Akita Hnrnpf^RT KO mice were studied up to the age of 24 weeks (n=8/group).

Results

Akita mice exhibited elevated systolic blood pressure as compared with Ctrl mice, which was significantly higher in Akita Hnrnpf^RT KO mice than Akita mice. Compared with Akita mice, Akita Hnrnpf^RT KO mice had lower blood glucose levels with increased urinary glucose excretion. Akita mice developed kidney hypertrophy, glomerular hyperfiltration (increased glomerular filtration rate), glomerulomegaly, mesangial expansion, podocyte foot process effacement, thickened glomerular basement membranes, renal interstitial fibrosis and increased albuminuria. These abnormalities were attenuated in Akita Hnrnpf^RT KO mice. Treatment of Akita Hnrnpf^RT KO mice with a selective A1 adenosine receptor inhibitor resulted in an increase in glomerular filtration rate. Renal Agt expression was elevated in Akita mice and further increased in Akita Hnrnpf^RT KO mice. In contrast, Sglt2 expression was increased in Akita and decreased in Akita Hnrnpf^RT KO mice.

Conclusions/interpretation

The renoprotective effect of Sglt2 downregulation overcomes the renal injurious effect of Agt when these opposing factors coexist under diabetic conditions, at least partly via the activation of tubuloglomerular feedback.

Introduction

Glucose reabsorption in the kidney is mediated by sodium-glucose co-transporters (SGLTs). SGLT2 is located in the S1/S2 segment of the renal proximal tubule (RPT) and reabsorbs more than 90% of the filtered glucose [1]. SGLT1 is located in the S3 segment of the RPT and reabsorbs the remaining glucose under normoglycaemic conditions [2]. Results from recent clinical trials revealed the renoprotective action of SGLT2 inhibitors (SGLT2i) in chronic kidney disease (CKD) patients with or without type 2 diabetes [3–6]. These studies have consistently shown that individuals who take SGLT2i have an initial decrease in eGFR, with a preserved eGFR in the long-term, suggesting that SGLT2i works through renal haemodynamic changes. Moreover, it has been suggested that SGLT2 blockade increases NaCl delivery to the macula densa, which reduces GFR through the process of tubuloglomerular feedback (TGF) [7]. Other mechanisms, such as reduced oxygen consumption [8], lipid oxidation and relative hyperketonaemia [8], as well as direct anti-inflammatory and anti-fibrotic functions [9, 10], have also been proposed to contribute to this renoprotective action.

Intrarenal angiotensinogen (Agt) is predominantly expressed in the S3 segment of the RPT [11]. Agt is the sole precursor of all angiotensins in the renin-angiotensin system (RAS). It is now well established that all components of the RAS are expressed in the RPT, and intrarenal RAS affects systemic BP and electrolyte and water homeostasis independently from the circulating RAS. We have previously reported that a nuclear protein, heterogeneous nuclear ribonucleoprotein F (HNRNPF) binds to the Agt promoter and suppresses Agt transcription in renal proximal tubular cells [12]. We have further established that the renoprotective action of Hnrnpf is mediated via downregulation of renal Agt expression in Akita mice (model of type 1 diabetes) [13] and db/db mice (model of type 2 diabetes) [14]. More recently, we showed that non-diabetic mice with Hnrnpf deficiency in the renal tubules develop hypertension, albuminuria and tubulointerstitial fibrosis via increased Agt expression [15]. Intriguingly, renal tubular Hnrnpf knock out (Hnrnpf^RT KO) mice exhibited renal glycosuria, reduced expression of Sglt2 (also known as Slc5a2) without change in Sglt1(also known as Slc5a1) expression or signs of Fanconi syndrome [15]. Moreover, we did not observe renoprotective action of absent Hnrnpf^RT following downregulation of Sglt2 under non-diabetic conditions [15].

In the present study, we used a type 1 diabetes mouse model (Akita Hnrnpf^RT KO) to explore the impact of altered Sglt2 expression on glycaemic control and protection of kidneys from injury. We hypothesised that decreased Sglt2 expression and activity may exert a renoprotective action that overrides the injurious effect of the known increase in renal Agt and Angiotensin II (Ang II) that occurs in Akita Hnrnpf^RT KO mice.

Methods

Generation of Akita Hnrnpf^RT KO mice

Hnrnpf floxed (Hnrnpf^fl/fl) mice with loxP sites flanking exon 4 of the Hnrnpf gene were generated as previously described on a C57BL/6 background [15]. Tubule-specific (Pax8)-Cre mice (Ref: https://doi.org/10.1002/gene.20008) (Stock number: 028196; Jackson Laboratory, Bar Harbor, ME, USA) were crossed with Hnrnpf^fl/fl to generate Hnrnpf^RT KO (Hnrnpf^fl/fl /Cre⁺) mice. Heterozygous Akita mice (C57BL/6-Ins2^Akita/J, Jackson Laboratories), which spontaneously develop hyperglycaemia after 4 weeks of age and are known to be suitable as an experimental platform for studies of diabetic kidney diseases [16, 17], were crossed with Hnrnpf^fl/fl to generate Akita Hnrnpf^fl/fl mice (NB: the homozygous Akita mouse is infertile). In the present study, the experimental mice were generated from male Akita Hnrnpf^fl/fl mice cross-bred with female Hnrnpf^RT KO (Hnrnpf^fl/fl /Cre⁺) mice (Fig 1a). Akita Hnrnpf^fl/fl/Cre⁺ (Akita Hnrnpf^RT KO) mice, as well as non-diabetic control littermates Hnrnpf^fl/fl mice (Ctrl) and diabetic Akita Hnrnpf ^fl/fl mice (Akita), were studied. Since the phenotype of Hnrnpf^RT KO (Hnrnpf^fl/fl /Cre⁺) mice were characterised in the previous study [15], Hnrnpf^RT KO (Hnrnpf^fl/fl /Cre⁺) mice were not included in these experiments.

Fig. 1.

Generation of renal tubular Akita Hnrnpf^RT KO mice. (a) Schematic representation of the strategy of generating renal tubular Hnrnpf gene knockout mice. (b) Genotyping identification; the PCR bands of floxed (568 bp) and wild-type (508 bp) alleles of Hnrnpf, Cre (392 bp), mutated (280 bp) and normal (140 bp) insulin 2 (Ins2) gene are indicated. (c) Quantitative Hnrnpf mRNA expression level in the cortex of 24-week-old mice. . Rpl13α was used as a reference gene. ***p<0.001; n=6–8 per group. Values are means ± SEM. (d) Representative western blot and quantification of HNRNPF protein expression in 24-week-old mice. ***p<0.001, n= 6 per group. (e) Representative microphotographs of immunofluorescent renal sections of 24-week-old mice (original magnification ×600). Merge of anti-HNRNPF (red) and anti-LTL, a proximal tubular marker (green). Nuclear staining with DAPI (blue). Images are representative of six independent analyses. Scale bars, 50 μm. BG, Blood glucose; BW, Body weight; G, Glomerulus; Het, Heterozygous; PT, Proximal tubule; WT, Wild-type. Image Credit: White Animal Mouse Cute Rodent Mice from Vector.me (by lemmling).

All mice were maintained on a 12-h light-dark cycle and fed a standard rodent chow diet. Genotyping was performed by PCR amplification of ear DNA, using primers specific for Cre recombinase [15], 5′ loxP site [15], and mutated insulin 2 gene [18], as previously described (Fig. 1b, electronic supplementary material [ESM] Table 1). The studies were performed on male mice only because female Akita mice are known to have lower blood glucose levels with less pronounced kidney injury [16, 17].

All animal studies were performed in accordance with the Principles of Laboratory Animal Care (NIH Publication No. 85–23, revised 1985 [http://grants1.nih.gov/grants/olaw/references/phspol.htm]) and were approved by the CRCHUM Animal Care Committee (protocol number: CM16016JCs).

Physiological Studies

Body weight was monitored bi-weekly. Random blood glucose was measured bi-weekly with Accu-Chek Performa Glucometer (Roche Diagnostics, Laval, QC, Canada)[19]. Systolic BP (SBP) was measured with BP-2000 tail-cuff pressure machine (Visitech Systems, Apex, NC, USA) in the morning, three times per week for each animal as described previously [13–15, 18]. Each animal was habituated to the procedure for at least 15–20 min per day for 5 days before the first SBP measurement at week six. SBP values are presented as means ± SEM of eight mice per group.

At 24 weeks of age, 24 h before euthanasia, mice were housed individually in metabolic cages for urine collection. Food intake and water consumption were also recorded. GFR was determined by FITC-inulin kinetics in awake mice as recommended by the Animal Models of Diabetic Complications Consortium (http://www.diacomp.org/) with slight modifications [15, 20]. Immediately after the GFR measurements, mice were euthanised by an overdose of pentobarbital. Blood was collected by cardiac puncture and centrifuged for serum. The kidneys and heart were removed and weighed. The left kidneys were processed for histology and immunostaining studies, and the cortex of right kidneys were used for RNA isolation and western blotting.

Treatment with selective A1 adenosine receptor inhibitor

To investigate the role of TGF in the GFR changes, Akita Hnrnpf^RT KO mice at 15 weeks of age were treated with a selective A1 adenosine receptor inhibitor (A1aRi, 8-cyclopentyl-1,3-dimethylxanthine, 1 mg kg^–1 d^–1, i.p. injection; Sigma-Aldrich (St. Louis, MO, USA)) once a day for a week (n=6) [21]. After 1 week, GFR was measured as described above and mice were euthanised by carbon dioxide asphyxiation. The kidneys were weighed and processed for histology. Blood was collected by cardiac puncture and centrifuged for serum. Separate groups of Ctrl (n=7), Akita (n=7), and Akita Hnrnpf^RT KO (n=6) mice treated with vehicle (154 mmol/l NaCl [saline], i.p. injection daily for 7 days) served as controls.

Studies on blood and urine samples

Haematocrit was determined using glass microcapillaries at the first time point of inulin clearance. Urinary albumin concentration was measured by ELISA (Albuwell and Creatinine Companion, Exocell, Philadelphia, PA, USA)[13–15, 18, 22], and 24-h urinary albumin excretion was calculated with the urine volume. Serum and urine glucose concentrations were determined using a colorimetric-enzymatic method (Autokit Glucose, Wako Diagnostics, Richmond, VA, USA). Fractional reabsorption of glucose (FR-glucose) was calculated using the serum and urine glucose concentration, urine volume, and GFR of each mouse [23]. Urinary Ang II levels were measured by a commercial ELISA kit (Bachem America, Torrence, CA, USA) according to the protocol III as described previously [15, 18]. Urinary adenosine was assayed by fluorometric method (Abcam (Cambridge, MA, USA), catalogue no. ab211094). The urine was pre-treated with catalase beads (Abcam, catalogue no. 218718) and the fluorescence signal was detected by excitation 535nm and emission 587nm.

Histologic studies

Freshly dissected kidneys were fixed in 4% formaldehyde in PBS, embedded in paraffin, and sectioned at 3 μm. Kidney sections were stained with periodic acid–Schiff using standard histologic procedures. Mean glomerular tuft volumes of more than 30 randomly-selected glomeruli per mouse were determined by the methods of Weibel and Gomez [19, 24] by using an image analysis software (Motics Images Plus 2.0; Motic, Richmond, BC, Canada). Mesangial expansion score was graded using a semi-quantitative scale of 0–3 (0, normal: mesangial matrix occupies <10% of glomerular tuft area; 1, mild: 10–25%; 2, moderate: >25–50%; and 3, severe: >50%) [16, 25]. Kidney interstitial fibrosis was evaluated by Sirius red staining and the collected images were analysed and semi-quantified with 5–6 mouse kidneys per group using National Institutes of Health ImageJ software version 1.52q (http://rsb.info.nih.gov/ij/).

Immunohistochemistry and immunofluorescence staining was performed using standard protocols, as described elsewhere [26, 27].The sources of antibodies used are listed in ESM Table 2. Sections were examined and photographed with Nikon ECLIPSE Ti microscope (Nikon, Melville, NY, USA) at the same time under identical conditions.

Electron microscopy

Kidney tissues fixed in 3% glutaraldehyde were post-fixed in OsO₄ and embedded in epoxy resin. Ultrathin sections were collected on carbon-coated formvar grids and stained with uranyl acetate and lead citrate, as previously described [28]. The images were obtained with Hitachi 7700 transmission electron microscope (Hitachi, Santa Clara, CA, USA). The thickness of glomerular basement membranes (GBMs) was measured at least 100 points per mouse and expressed as the harmonic mean by a pathologist blinded to genotype. The foot process effacement of podocytes was evaluated by a board-certified renal pathologist (M.Yamashita) in a blinded manner, and expressed as the mean (%) of (effaced podocyte foot processes)/(underlying GBM) in at least ten glomerular capillary loops per mouse.

Real time-quantitative PCR

The mRNA levels of various genes in the kidney cortex were quantified by real time-quantitative PCR (qRT-PCR) with the forward and reverse primers listed in ESM Table 1 [27, 29].

Western blot analysis

Western blot was performed for the estimation of protein levels using the mouse kidney cortex as described previously [30]. The relative densities of HNRNPF and β-actin bands were quantified by computerised laser densitometry (ImageQuant software, version 5.1; Molecular Dynamics, Sunnyvale, CA, USA). Details of the sources of antibodies and working dilutions are listed in ESM Table 2.

Statistical analysis

All values are expressed as the mean ± SEM. Statistical comparisons were made using GraphPad Prism Version 8.0 software (GraphPad Software, San Diego, USA). Comparisons between the three experimental groups were made by one-way ANOVA, followed by Bonferroni’s post hoc test. For all analyses, p values <0.05 were considered statistically significant.

Results

Renal tubule-specific deletion of Hnrnpf in Akita mice

Akita Hnrnpf^RT KO mice were generated by using Pax8-Cre/lox recombination strategy (Fig. 1a). PCR analysis of genomic DNA extracted from ear punch tissues to distinguish the genotype of floxed (568 bp), WT (508 bp), Cre (392 bp), normal Ins2 (140 bp) and mutated Ins2 (280 bp) are shown in Fig. 1b. Hnrnpf mRNA (Fig. 1c) and protein (Fig. 1d) expressions were significantly reduced in the cortex of Akita Hnrnpf^RT KO mice at 24 weeks of age. Double-staining of kidney sections (Fig. 1e) with anti-HNRNPF antibody and anti-Lotus tetragonolobus lectin (LTL, a marker of RPT [31]) antibody (Vector Labs, Burlingame, CA, USA) confirmed renal tubule-specific deletion of HNRNPF in Akita Hnrnpf^RT KO mice. Kidneys from Akita Hnrnpf^RT KO mice had positive HNRNPF staining in glomeruli, but not in renal tubules (Fig. 1e).

Physiological variables

Consistent with previous reports including ours [18, 20, 32], Akita mice had lower body weight than Ctrl mice but did not differ from Akita Hnrnpf^RT KO mice (Fig. 2a). SBP was significantly higher in Akita mice than Ctrl, and it was significantly higher in Akita Hnrnpf^RT KO mice than Akita mice from the age of 16 to 24 weeks (Fig. 2b).

Fig. 2.

Characterisation of Akita Hnrnpf^RT KO mice. Longitudinal changes in (a) body weight, (b) mean SBP (performed three times per mouse per week in the morning), and (c) non-fasting blood glucose. * p<0.05 Ctrl vs Akita, ** p<0.01 Ctrl vs Akita, *** p<0.001 Ctrl vs Akita, † p<0.05 Akita vs Akita Hnrnpf^RT KO, ‡ p<0.001 Akita vs Akita Hnrnpf^RT KO; means ± SEM (n=8 per group). Dotted line represents the upper detection limit of the glucometer (~33 mmol/l). (d) Kidney weight (KW)/tibia length (TL), (e) GFR/body weight (BW), and (f) 24-h urine albumin excretion of 24-week-old mice. Values are means ± SEM (n=8 per group). *p<0.05, **p<0.01, ***p<0.001

Blood glucose levels were markedly higher in Akita mice than in Ctrl, whereas Akita Hnrnpf^RT KO mice had consistently lower blood glucose levels than Akita mice throughout the entire period of study (Fig. 2c). Since some Akita mice had higher blood glucose levels than the upper detection limit of the glucometer (~33 mmol/l), we used a colorimetric assay to measure the glucose levels in serum obtained by cardiac puncture at 24 weeks of age (Table 1). These measurements confirmed that serum glucose levels of Akita Hnrnpf^RT KO mice were significantly higher than in Ctrl mice, but significantly lower than in Akita mice. Moreover, Akita Hnrnpf^RT KO mice had significantly higher urinary glucose excretion than Akita mice (Table 1). Calculated FR-glucose of Akita Hnrnpf^RT KO mice was lower than that of Akita mice (Table 1). These findings indicate that decreased serum glucose levels in Akita Hnrnpf^RT KO mice can be attributed to increased urinary glucose excretion.

Table 1.

Physiological variables of mice at 24 weeks of age

	Ctrl	Akita	Akita HnrnpfRT KO
SBP (week 24) (mmHg)	118.8 ± 1.3	133.6 ± 2.5**	143.7 ± 3.7***†
Urine volume (ml/day)	0.75 ± 0.07	7.0 ± 0.4***	10.4 ± 1.4***‡
Food intake (g/day)	0.41 ± 0.07	1.36 ± 0.11***	1.28 ± 0.21***
Water intake (ml/day)	1.71 ± 0.13	7.98 ± 0.49***	11.70 ± 1.62***†
Serum glucose (mmol/l)	12.0 ± 0.6	59.9 ± 5.3***	45.8 ± 2.8***‡
Urine glucose (μmol/day)	1.43 ± 0.66	4717.9 ± 274.5***	6484.5 ± 657.3***†
FR-glucose (%)	99.77 ± 0.18	85.56 ± 3.05**	66.23 ± 4.20***§
Urinary Ang II/Cr (ng/μmol Cr)	5.0 ± 0.4	9.6 ± 0.8*	11.4 ± 1.8*
Haematocrit (%)	49.5 ± 1.2	50.5 ± 1.0	52.3 ± 1.1

n=8/group. Values are means ± SEM

Cr, Creatinine

^*p<0.05 vs Ctrl,

^**p<0.01 vs Ctrl,

^***p<0.001 vs Ctrl,

^†p< 0.05 vs Akita,

^‡p< 0.01 vs Akita,

^§p< 0.001 vs Akita

Both urine volume and water intake were increased in Akita mice compared with Ctrl and were further increased in Akita Hnrnpf^RT KO mice (Table 1). We did not observe significant differences in haematocrit between the groups (Table 1) or apparent signs of dehydration in any of the groups studied.

Akita mice developed significant kidney hypertrophy as assessed by the kidney weight/tibia length ratio (Fig. 2d), and glomerular hyperfiltration assessed by the GFR/body weight ratio (Fig. 2e). Both kidney hypertrophy and glomerular hyperfiltration were attenuated in Akita Hnrnpf^RT KO mice. Furthermore, urinary albumin excretion was increased in Akita mice compared with Ctrl, whereas it was lower in Akita Hnrnpf^RT KO mice (Fig. 2f). Urinary Ang II/creatinine ratio was higher in Akita than in Ctrl mice, and Akita Hnrnpf^RT KO mice had even higher levels than Akita mice, indicating elevated renal Agt expression in Akita Hnrnpf^RT KO mice (Table 1).

Renal pathology and fibrosis in Akita Hnrnpf^RT KO mice

Akita mice developed glomerulomegaly, mesangial expansion, and glomerular and tubulointerstitial fibrosis, which were less pronounced in Akita Hnrnpf^RT KO mice (Fig. 3a–d). Furthermore, electron microscopic examination of the glomeruli showed thickened GBMs and foot process effacement in Akita mice, and these features were ameliorated in Akita Hnrnpf^RT KO mice (Figs. 3a, e, f).

Fig. 3.

Histological analyses of mouse kidneys at 24 weeks. (a) Representative images of periodic acid–Schiff (PAS) with lower (×100) and higher (×600) magnification, Sirius red staining (×100 and ×600), and electron microscopy (EM). Black scale bars, 50 μm (PAS and Sirius red) and white scale bar, 1 μm (EM). Black arrows indicate tubulointerstitial fibrosis. Red arrows indicate foot processes. (b) Mean glomerular tuft volume and (c) mesangial expansion score. Each dot represents the mean of 30–40 glomeruli in a mouse, n=6 per group. (d) Semi-quantification of Sirius red positive area, n=5–6 per group. (e) GBM thickness: each dot represents the harmonic mean of at least 100 points per one mouse. (f) Foot process effacement quantification of the EM pictures: each dot represents the mean of at least ten glomerular capillary loops per one mouse. n=4 for Ctrl, n=5 for Akita, and n=3 for Akita Hnrnpf^RT KO. qRT-PCR of fibronectin (Fn1) (g) and α-smooth muscle actin (α-SMA) (h) in kidney cortex of 24-week-old mice. Rpl13α was used as a reference gene. Values are means ± SEM. n=8 for Ctrl, n=6 for Akita and Akita Hnrnpf^RT KO. *p<0.05, **p<0.01, ***p<0.001

Next, we examined the extent of renal fibrosis by determining the expression of profibrotic genes. Renal cortices from Akita mice displayed increased mRNA expressions for fibronectin (Fn1) and α-Smooth muscle actin (α-SMA, also known as Acta2), while these mRNA expressions were similar in Akita Hnrnpf^RT KO and Ctrl mice (Fig. 3g, h).

Expression of Agt and SGLT2 in Akita Hnrnpf^RT KO mice

Consistent with previous reports [19, 33], we detected increased expression of Agt protein and mRNA in the renal cortex of Akita mice (Figs. 4a, b). Significantly increased Agt (Fig. 4a) and Ang II (Fig. 4a) immunostaining was detectable in Akita Hnrnpf^RT KO mice as compared with Akita mice and Ctrl mice, which was further confirmed by quantification of Agt mRNA expression in isolated RPTs by qRT-PCR (Fig. 4b). Expression of SGLT2 protein (Fig. 4c) and mRNA (Fig. 4d) were higher in Akita mice than in Ctrl mice and were markedly lower in Akita Hnrnpf^RT KO mice as compared with Akita mice. Recent studies have suggested that the reduction of Na⁺/H⁺ exchanger isoform 3 (NHE3) activity is the determinant of natriuretic effect of SGLT2i [34–36]. Therefore, we also evaluated the Nhe3 (also known as Slc9a3) expression in Akita Hnrnpf^RT KO mice. As shown in Fig. 4e, unexpectedly, Nhe3 mRNA expression was higher in Akita Hnrnpf^RT KO mice than Akita mice.

Fig. 4.

Renal Agt and Sglt2 expression in Akita Hnrnpf^RT KO mice. (a) Representative microphotographs of immunostainings of 24-week-old mice, anti-AGT immunohistochemistry (magnification ×200) and anti-Ang II immunohistochemistry (magnification ×200). Scale bars, 50 μm. (b) qRT-PCR of Agt/Rpl13α mRNA ratio in kidney cortex of 24-week-old mice. (c) anti-SGLT2 (red), anti-LTL (a proximal tubular marker; green), and merge of the SGLT2 and LTL immunofluorescence (magnification ×100). Scale bars, 50 μm. (d) qRT-PCR of Sglt2 (Slc5a2)/Rpl13α mRNA ratio in kidney cortex of 24-week-old mice. (e) qRT-PCR of NHE3 (Slc9ac)/Rpl13α mRNA ratio in kidney cortex of 24-week-old mice. Rpl13α was used as a reference gene. Values are means ± SEM (n=8 for Ctrl, n=6 for Akita and Akita Hnrnpf^RT KO). *p<0.05, **p<0.01, ***p<0.001

Glomerular haemodynamic changes in mice by A1 adenosine receptor inhibitor

Next, we investigated whether reductions in GFR in Akita Hnrnpf^RT KO mice relative to those in Akita mice, which reflect attenuated glomerular hyperfiltration, can be attributed to improved blood glucose or to restoration of TGF via increased sodium delivery to the macula densa. Adenosine is widely considered to be a mediator of TGF [37]. As shown in Fig. 5a, urinary adenosine excretion was significantly increased in Akita mice than Ctrl mice. It was slightly further increased in Akita Hnrnpf^RT KO mice, though not statistically significant (p=0.11, Akita vs Akita Hnrnpf^RT KO). Therefore, we compared the GFR of Akita Hnrnpf^RT KO mice with or without treatment with selective A1aRi, which has been shown to block constriction in the afferent arteriole induced by high sodium concentration at the macula densa [21, 38].

Fig.5.

Effect of treatment of Akita Hnrnpf^RT KO mice with selective A1aRi. (a) Urinary adenosine excretion of 24-week-old mice (n=6/group). (b) GFR/body weight (BW), (c) serum glucose, (d) kidney weight (KW)/tibia length (TL) of 16-week-old mice. Akita Hnrnpf^RT KO mice treated with i.p. injection of A1aRi daily for 7 days (Akita Hnrnpf^RT KO A1aRi, n=6) were compared with Ctrl (n=7), Akita (n=7), and Akita Hnrnpf^RT KO (n=6) mice which received the same volume of i.p. saline. (e) Representative images of merge of SGLT2 (red) and LTL (green) immunofluorescence (magnification ×100) and (f) representative images of periodic acid–Schiff (PAS) staining (magnification ×600). Scale bars, 50 μm. (g) Semi-quantification of SGLT2/LTL ratio. (h) Mean glomerular tuft volume. Values are mean ± SEM. *p<0.05, **p<0.01, ***p<0.001.

One-week treatment with daily A1aRi administration increased GFR in Akita Hnrnpf^RT KO mice (Fig. 5b), without affecting blood glucose levels (Fig. 5c), and kidney weight/tibia length (Fig. 5d), SGLT2 expression (Fig. 5e, g) or renal morphology (Fig. 5f, h) as compared with non-treated Akita Hnrnpf^RT KO mice. These results imply that reduced glomerular hyperfiltration in Akita Hnrnpf^RT KO was mediated, at least in part, through activation of TGF.

Discussion

Accumulating evidence indicates that SGLT2 inhibition is renoprotective in patients with diabetic kidney disease, independent of its blood glucose lowering effect [3–5]. However, many questions regarding SGLT2 and SGLT2i remain unanswered; for instance, little is known about the molecular mechanisms regulating SGLT2 expression, the renal outcome of increased or decreased endogenous SGLT2 expression, and the long-standing renoprotective action of SGLT2i. The Hnrnpf^RT KO mouse model with enhanced Agt and reduced Sglt2 expression [15], provides a unique opportunity to study the effects of long-term alteration in Sglt2 expression with and without diabetes.

In the present study, we demonstrated that Akita mice with Hnrnpf^RT KO had augmented systemic hypertension as compared with Akita mice, likely due to increased Agt expression and subsequently enhanced Ang II production (Fig.6). Furthermore, Akita Hnrnpf^RT KO mice exhibited improved glucose control, likely from increased glycosuria, associated with downregulation of Sglt2 expression. These findings are consistent with, and expand upon, those about non-diabetic Hnrnpf^RT KO mice [15]. A novel finding in the present study is that Akita Hnrnpf^RT KO mice showed less pronounced kidney hypertrophy, glomerular hyperfiltration and renal fibrosis than Akita mice. Moreover, the decrease in glomerular hyperfiltration led to less glomerular injury, observed as attenuated GBM thickness, foot process effacement and mesangial expansion, and to less urinary albumin excretion. These findings indicate that despite increased renal Agt and Ang II expression, the renoprotective action of downregulated Sglt2 appears to override the deleterious effect of Ang II within the diabetic kidney. Considering the fact that non-diabetic Hnrnpf^RT KO mice showed more pronounced signs of renal injury, as indicated by increased albuminuria and renal fibrosis, than control mice [15], the status of diabetes may influence the outcomes of Sglt2 expression changes (Fig.6).

Fig. 6.

Summary of the phenotypes of non-diabetic and Akita Hnrnpf^RT KO mice. ACR, albumin creatinine ratio; WT, Wild-type

Since Akita Hnrnpf^RT KO had lower blood glucose levels than Akita mice, we cannot exclude the possibility that better glucose control contributed to the improvements in renal pathophysiological measures. However, as shown in Figs 2 and Table 1, the blood glucose levels of Akita Hnrnpf^RT KO mice were still much higher than that of Ctrl mice, while renal fibrosis, expression of profibrotic genes, and mean urinary albumin excretion were comparable to those in Ctrl mice. Therefore, it appears unlikely that renal improvements observed in Akita Hnrnpf^RT KO mice can simply be attributed to lower blood glucose. To address this issue, we measured urinary adenosine excretion and GFR after treatment with A1aRi in Akita Hnrnpf^RT KO mice. Urinary adenosine was not significantly increased in Akita Hnrnpf^RT KO mice compared with Akita mice. However, A1aRi-treated Akita Hnrnpf^RT KO mice had GFR comparable to that in untreated Akita mice of the same age, while blood glucose and renal SGLT2 expression were unchanged compared with Akita Hnrnpf^RT KO mice without A1aRi. We interpret these observations to indicate that downregulation of Sglt2 in Akita Hnrnpf^RT KO mice likely contributes to the slowing of disease progression by decreasing intraglomerular pressure via TGF in addition to lowering blood glucose levels, similar to Akita mice treated with SGLT2i [21, 39].

We have found that Nhe3 mRNA expression was increased in Akita Hnrnpf^RT KO mice. This contrasts with a previous report where SGLT2i decreased the tubular protein expression of NHE3 in rats with induced diabetes [40]. This difference may be related to increased Ang II in Akita Hnrnpf^RT KO mice since Ang II has been shown to stimulate NHE3 expression [41]. Moreover, studies have demonstrated a discrepancy between NHE3 mRNA expression, protein expression and transporter activity, and post-transcriptional NHE3 phosphorylation is also an important factor [36, 42]. Therefore, it is still unclear if changes in NHE3 activity contributes to TGF and natriuresis in Akita Hnrnpf^RT KO mice. More studies are needed along these lines.

An important difference between Akita Hnrnpf^RT KO mice and Akita mice treated with SGLT2i is higher expression of Agt in the kidneys of Akita Hnrnpf^RT KO mice. Previous studies have established that activation of the intrarenal RAS evokes haemodynamic, profibrotic and proinflammatory changes, ultimately leading to deterioration of renal function [11, 13, 14, 18, 43]. The effect of SGLT2i on intrarenal RAS has been reported with contradictory results; SGLT2i increases the circulating and urinary RAS mediators in individuals with type 1 diabetes [7, 44], while in animal models of non-diabetic CKD and type 2 diabetes, SGLT2i did not change or even suppress intrarenal RAS [45, 46]. We have also recently reported that treatment of Akita mice with SGLT2i did not affect intrarenal Agt expression, whereas blood glucose was lowered and renal injury was significantly ameliorated [47]. Accordingly, increased Agt expression in Akita Hnrnpf^RT KO mice was unlikely due to attenuation of Sglt2 expression, rather it was likely to be a direct consequence of Hnrnpf deletion, which also directly de-suppresses Agt gene transcription [12–14].

At present, it is still unclear how Hnrnpf modulates Sglt2 expression. Though they are complicated by simultaneously elevated Agt expression, Hnrnpf^RT KO mice could be a unique preclinical tool to study Sglt2 regulation. Identifying the molecular mechanisms governing endogenous Sglt2 transcription and expression could lead to the identification of potential targets for developing novel treatment strategies to protect the kidney from both diabetes-associated and other types of injury.

In conclusion, our present findings, together with our work on non-diabetic Hnrnpf^RT KO mice [15], would indicate that the renoprotective action of reducing Sglt2 expression overrides the deleterious effects of upregulated renal Agt/Ang II in the presence of diabetes. Our present study raises the possibility that SGLT2i or familial renal glycosuria may have different effects in individuals with diabetes and individuals without diabetes. More detailed mechanistic studies are needed to address how Hnrnpf modulates Sglt2 expression in renal tubules.

Research in context.

What is already known about this subject?

• Heterogeneous nuclear ribonucleoprotein F (Hnrnpf) modulates angiotensinogen (Agt) gene transcription in renal proximal tubules

• Renal tubule-specific Hnrnpf knockout (Hnrnpf^RT KO) mice exhibit increased expression of Agt and reduced expression of sodium-glucose co-transporter 2 (Sglt2)

• Non-diabetic Hnrnpf^RT KO mice exhibit hypertension, albuminuria, tubulointerstitial fibrosis and glycosuria

What is the key question?

• What is the impact of Hnrnpf deletion on glomerular hyperfiltration and renal injury in the Akita type 1 diabetes mouse model?

What are the new findings?

• Glomerular hyperfiltration and fibrosis are attenuated in Akita Hnrnpf^RT KO mice, likely via the activation of tubuloglomerular feedback due to reduced Sglt2 expression

• Akita Hnrnpf^RT KO mice have increased renal Agt expression and systolic blood pressure

• The renoprotective effect of Sglt2 downregulation overrides the deleterious effects of Agt/Ang II when these opposing factors are present in diabetic conditions

How might this impact on clinical practice in the foreseeable future?

• Our study suggests there are long-term renoprotective effects of Sglt2 downregulation. Identifying the molecular mechanisms governing endogenous Sglt2 expression could lead to the identification of potential targets for developing novel treatment strategies to protect the kidney from diabetic injury

Abbreviations

A1aRi

A1 adenosine receptor inhibitor

AGT

Angiotensinogen

Ang II

Angiotensin II

CKD

Chronic kidney disease

Ctrl

Control

FR-glucose

Fractional reabsorption of glucose

GBM

Glomerular basement membrane

HNRNPF

Heterogeneous nuclear ribonucleoprotein F

KO

Knock out

NHE3

Na⁺/H⁺ exchanger isoform 3

RAS

Renin-angiotensin system

RPT

Renal proximal tubule

SBP

Systolic blood pressure

SGLT1

Sodium-glucose co-transporter 1

SGLT2

Sodium-glucose co-transporter 2

SGLT2i

SGLT2 inhibitor

TGF

Tubuloglomerular feedback

Data availability

The datasets generated and analysed during the current study are available from the corresponding authors on reasonable request.