Heterogeneous Nuclear Ribonucleoprotein F Mediates Insulin Inhibition of Bcl2-Modifying Factor Expression and Tubulopathy in Diabetic Kidney

We investigated the molecular mechanism(s) by which insulin prevents Bcl2-modifying factor (Bmf)-induced renal proximal tubular cell (RPTC) apoptosis and loss in diabetic mice. Transgenic mice (Tg) mice specifically overexpressing human BMF in RPTCs and non-Tg littermates were studied at 10 to 20 weeks of age. Non-diabetic littermates, diabetic Akita mice +/− insulin implant, Akita Tg mice specifically overexpressing heterogeneous nuclear ribonucleoprotein F (hnRNP F) in their RPTCs and immortalized rat renal proximal tubular cells (IRPTCs) were also studied. BMF-Tg mice exhibited higher systolic blood pressure, urinary albumin/creatinine ratio, RPTC apoptosis and urinary RPTCs than non-Tg mice. Insulin treatment in Akita mice and Akita mice overexpressing hnRNP F suppressed Bmf expression and RPTC apoptosis. In hyperinsulinemic-euglycemic wild type mice, renal Bmf expression was down-regulated with up-regulation of hnRNP F. In vitro, insulin inhibited high glucose-stimulation of Bmf expression, predominantly via p44/42 mitogen-activated protein kinase (MAPK) signaling. Transfection of p44/42 MAPK or hnRNP F small interfering RNA (siRNA) prevented insulin inhibition of Bmf expression. HnRNP F inhibited Bmf transcription via hnRNP F-responsive element in the Bmf promoter. Our results demonstrate that hnRNP F suppression of Bmf transcription is an important mechanism by which insulin protects RPTCs from apoptosis in diabetes.

apoptosis in type 1 diabetic (T1D) Agt-Tg mice 18,19 . Conversely, catalase (CAT) overexpression attenuates RPTC apoptosis in T1D CAT-Tg mice 20 and T2D db/db CAT-Tg mice 21 , supporting the view that hyperglycemia via enhanced ROS generation plays a central role in RPTC apoptosis in diabetes.
We used microarray analysis to identify the downstream target genes of ROS, and noted elevated expression of Bcl2 (B-cell lymphoma 2)-modifying factor (Bmf) gene in the RPTCs of db/db mice; the elevated expression was, however, normalized in db/db CAT-Tg mice 22 . In vitro, Bmf overexpression enhanced and knockdown with small interference RNA (siRNA) decreased RPTC apoptosis in HG milieu. Furthermore, Bmf expression is markedly enhanced and localized to apoptotic RPTCs in human diabetic kidneys 22 . However, whether overexpression of Bmf would induce RPTC apoptosis and kidney injury in vivo has not been investigated.
Intensive insulin therapy has proven to be the most effective treatment for preventing nephropathy progression in T1D; however, the underlying mechanisms remain incompletely understood 23,24 . We previously reported that insulin inhibits high-glucose stimulation of renal rat Agt gene expression and RPTC hypertrophy through a putative insulin-responsive element (IRE) in the rat Agt gene promoter that interacts with 2 nuclear proteins, heterogeneous nuclear ribonucleoprotein F and K (hnRNP F/K) in vitro [25][26][27] . Overexpression of hnRNP F inhibited renal Agt gene expression and attenuated hypertension, kidney hypertrophy and RPTC apoptosis in Akita (T1D) and db/db Tg mice 28,29 . We further reported that hnRNP F/K mediate insulin inhibition of renal Agt gene expression and that insulin stimulates hnRNP F/K expression through p44/42 MAPK signaling pathway but not through the phosphatidylinositol 3-kinase (PI-3K) pathway in diabetic mice 30,31 . These findings suggest that insulin inhibition of renal Agt gene transcription and RPTC apoptosis occurs via hnRNP F/K in diabetes.
In the present study, we investigated the impact of Bmf overexpression on RPTC apoptosis in BMF-Tg mice and examined whether hnRNP F mediates, at least in part, insulin regulation of Bmf actions in Akita mice and in rat immortalized RPTCs (IRPTCs) cultured in HG milieu.

RPTC-Specific Expression of Human Bcl2 Modifying Factor (hBMF) transgene in transgenic Mice.
hBMF-Tg mice were generated by inserting myc-hBMF cDNA with the stop codon into a pKAP2 plasmid containing the kidney-specific androgen-regulated protein (KAP) promoter (Fig. 1a). Southern blot analysis revealed the presence of the transgene in heterozygote and homozygote animals from line 148 (Fig. 1b). Tissue-specific analysis by RT-PCR confirmed hBMF mRNA expression in the kidney cortex and RPTs isolated from male hBMF-Tg mice, as well as in the kidney cortex of female hBMF-Tg mice implanted with testosterone pellet but not in other tissues from both male and female mice (Fig. 1c). hBMF transgene was detected in RPTs of male hBMF-Tg mice but not in non-Tg mice and can be differentiated from endogenous mouse Bmf using primers specific for hBMF and mouse Bmf in RT-PCR, respectively (Fig. 1d). WB of isolated RPTs with anti-Bmf or anti-cMyc antibody (Fig. 1e) and immunostaining of kidney sections with an anti-Bmf antibody (that recognizes both human and mouse Bmf) (Fig. 1f) confirmed significantly higher BMF expression in RPTCs from male hBMF-Tg mice than in non-Tg mice. Furthermore, immunofluorescence staining with anti-Bmf and anti-aquaporin-1 (APQ1, a proximal tubule marker) revealed RPTC-specific BMF expression and its co-localization with AQP1positive RPTCs of male hBMF-Tg mice ( Fig. 1g and Supplementary Fig. 1c). These data demonstrate that hBMF expression is RPTC-specific in hBMF-Tg mice.

Overexpression of hBMF Increases Systolic Blood Pressure and Kidney Injury in Tg Mice.
Longitudinal systolic blood pressure (SBP) measurements revealed consistently higher SBP in hBMF-Tg mice than in non-Tg mice from week 11 to 20 (Fig. 2a). Cross-sectional SBP measurements at week 20 showed significantly elevated SBP (on average by ~8 mm Hg) in hBMF-Tg mice as compared with non-Tg mice (Table 1).
Marked increases in urinary albumin/creatinine ratio (ACR) were observed in hBMF-Tg mice at 20 weeks of age ( Fig. 2b and Table 1). Blood glucose, body weight (BW), kidney weight (KW), tibial length (TL), BW/KW, or BW/TL did not differ significantly between hBMF-Tg mice and non-Tg mice (Table 1). Periodic acid Schiff (PAS) staining of kidney sections revealed mild structural changes in hBMF-Tg compared to non-Tg mice and yielded higher statistically significant tubular injury scores (Fig. 2c). Moreover, more pronounced Masson's Trichrome staining and TGFβ1 immunostaining were detected in glomerulo-tubular areas of hBMF-Tg mice as compared to non-Tg controls and confirmed by semi-quantification (Fig. 2d,e, respectively). RT-qPCR showed significantly increased mRNA levels of TGFβ1 (Fig. 2f), fibronectin (Fn1) (Fig. 2g) and collagen 1α (Col 1α) (Fig. 2h) in RPTs of hBMF-Tg mice compared to non-Tg mice, indicating that BMF overexpression induces tubulointerstitial fibrosis in kidneys of hBMF-Tg mice. Moreover, morphological measurements revealed significant increases in glomerular tuft volume and RPTC volume in hBMF-Tg mice as compared to non-Tg mice ( Table 1).

Overexpression of BMF Stimulates RPTC Apoptosis and Increases Urinary RPTCs.
The percentage of TUNEL-positive cells was significantly higher in hBMF-Tg mice as compared to non-Tg mice (Fig. 3a,b). WB analysis showed increases in Bax and cleaved caspase-3 expression (Fig. 3c,d) without detectable changes in Bcl2 and caspase-3 expression in RPT extracts of hBMF-Tg mice as compared to non-Tg mice. Furthermore, co-immunoprecipitation (co-IP) experiments revealed more BMF co-IP with Bcl2 whereas less Bcl2 co-IP with Bax in RPTs of hBMF-Tg mice than non-Tg mice (Fig. 3e). These data would indicate that in hBMF-Tg mice, BMF predominantly binds Bcl2 and dissociates it from Bax, thereby tipping the balance of the Bax/Bcl2 ratio towards caspase-3 activation and subsequently RPTC apoptosis.
As Bmf is known to enhance anoikis 32,33 , we investigated whether overexpression of BMF would increase the number of RPTCs in the urine as an indirect indicator of loss of RPTCs. Using flow cytometry, we detected a greater percentage of cells in the urine that stained positive for prominin-1, a proximal tubule marker 34 in hBMF-Tg than in non-Tg mice (Fig. 3f,g). These data indicate that overexpression of BMF induces RPTC loss, followed by shedding into the urine in hBMF-Tg mice. The isolated 17-kb KAP2-myc-hBMF transgene (digested with SpeI and NdeI) was microinjected into 1-cell fertilized mouse embryos obtained from super ovulated C57Bl6 C3H mice. (b) Southern blotting of genomic DNA for founders with biotin-labeled BMF probe. Heterozygous and homozygous F 1 , F 2 and F 3 were screened by PCR with specific primers (Table 1) www.nature.com/scientificreports www.nature.com/scientificreports/ Insulin Inhibits Bmf Expression and RPTC Apoptosis in Akita Mice. We next tested whether insulin reduction of RPTC apoptosis in Akita mice is mediated, at least in part, via suppression of Bmf expression. As anticipated, Bmf expression was higher in RPTs of Akita mice than in WT mice and was inhibited by insulin (Fig. 4a). WB for Bmf protein expression in isolated RPTs confirmed these findings (Fig. 4b,c). Co-immunostaining studies showed Bmf localization to TUNEL-positive RPTCs of Akita mice, which was normalized by insulin treatment (Fig. 4d). The percentage of TUNEL-positive RPTCs was significantly higher in Akita than in WT mice and was normalized by insulin treatment (Fig. 4e). RT-qPCR analysis revealed increases in Bmf (Fig. 4f) and Bax mRNA expression (Fig. 4g), with decreases in Bcl-2 mRNA (Fig. 4h) in Akita mice. These changes in the Akita mice were normalized by insulin treatment with the exception of Bax mRNA. Consistently, the Bax/Bcl-2 mRNA ratio was significantly increased in Akita mice and was reversed by insulin treatment (Fig. 4i).

Insulin Inhibits Bmf Expression Independent of its Glucose Lowering Effect. We previously
reported that hyperinsulinemia up-regulated hnRNP F expression in RPTCs independent of its glucose lowering effect 31 . To investigate whether insulin could also inhibit renal Bmf expression independent of its effect on lowering systemic blood glucose in vivo, hyperinsulinemic-euglycemic clamp experiments were performed on non-diabetic WT mice. Consistent with the previous report 31

HnRNP F Overexpression Inhibits Bmf Expression in Akita hnRNP F-tg Mice.
To explore the functional relationship between hnRNP F and Bmf expression in diabetes, we compared Akita hnRNP F-Tg mice to non-Tg mice and Akita mice 28 . Significant decreases in hnRNP F immunostaining ( Supplementary Fig. 1b) and increases in Bmf immunostaining were observed in kidneys of Akita mice as compared with non-Akita WT mice, whereas Bmf immunostaining was reduced in Akita hnRNP F-Tg mice (Fig. 5d). WB of Bmf protein expression ( Fig. 5e) and RT-qPCR of Bmf mRNA expression ( Fig. 5f) confirmed these observations. These data would imply a role for hnRNP F in mediating insulin suppression of Bmf expression.

Insulin Inhibits Bmf expression via HnRNp F in IRptCs.
Confirming our previous findings 22 , we found that HG stimulated Bmf mRNA expression in cultured IRPTCs and that insulin reversed this finding (Fig. 6a). Insulin treatment also prevented the stimulatory effect of HG on rat Bmf promoter (N-1370/+102) activity (Fig. 6b). Pharmacological blockade of p44/42 MAPK (with PD98059 and U0126) effectively prevented insulin inhibition of Bmf promoter activity, whereas blockade of PI-3-Kinase with wortmannin was without effect (Fig. 6b). Moreover, transfection with siRNA of p44 MAPK or p42MAPK also abolished insulin inhibition of Bmf promoter activity in HG, whereas scrambled (Scr) siRNA had no effect (Fig. 6c).
Furthermore, siRNA of hnRNP F, but not Scr siRNA abolished insulin inhibition of Bmf promoter activity in HG (Fig. 6d). Consistently, we detected 10-fold increase and 70% decrease of hnRNP F mRNA expression when IRPTCs were transfected with pCMV-hnRNP F-HA plasmid and siRNA of hnRNP F, respectively (Fig. 6e). In contrast, Bmf mRNA expression decreased by 60% and increased 175% when IRPTCs were transfected with pCMV-hnRNP F-HA and hnRNP F siRNA as compared to controls, respectively (Fig. 6f).  www.nature.com/scientificreports www.nature.com/scientificreports/      (Fig. 7c). In contrast, Bmf promoter with deletion of N-997 to N-991 and N-402 to N-395 did not alter the response to insulin (Fig. 7c). Furthermore, pGL4.20-Bmf promoter (N-1370/N + 102) activity was significantly suppressed when co-transfected with pCMV-hnRNP F-HA, whereas it had no detectable effect on pGL4.20-Bmf promoter (N-1370/N + 102 ∆ N-1086/N-1081) in NG (Fig. 7d). Moreover, Bmf promoter with deletion of nucleotides N-997 to N-991 and N-402 to N-395 did not affect the inhibitory effect of pCMV-hnRNP F-HA on Bmf promoter activity. These results point toward that nucleotides N-1086 to N-1081 is a putative core RE responding to insulin and hnRNP F.

Identification of hnRNP F-Response Element (RE) in Rat
Indeed, EMSA showed that the WT double-strand DNA fragment (N-1093 to N-1072) binds to IRPTC nuclear proteins, which could be displaced by the WT DNA fragment, but not by mutated DNA fragments (M1, M2 and M3) (Fig. 7e). Moreover, incubation with an anti-hnRNP F antibody, but not with rabbit IgG (control) induced a supershift (SS) of the hnRNP F-RE with nuclear proteins (Fig. 7e).  www.nature.com/scientificreports www.nature.com/scientificreports/

Discussion
The present study demonstrates that overexpression of hBMF specifically in the RPTCs enhances RPTC apoptosis and loss into the urine in hBMF-Tg mice, indicating a critical role for Bmf in mediating tubular cell apoptosis and loss. Our findings also indicate that insulin treatment prevents RPTC apoptosis and inhibits renal Bmf transcription via a novel putative insulin-response element (IRE) in the Bmf gene promoter that interacts with hnRNP F. These findings identify a novel mechanism by which insulin may prevent nephropathy progression in diabetes.
Bmf is a member of the BH3-only protein family 32 . Bmf binds to cytoskeletal structures and is sequestered to myosin V motors through association with dynein light chain 2. Certain damage signals, such as the detachment of adherent cells (anoikis) from their substratum trigger the release of Bmf, which through binding to pro-survival Bcl-2 proteins induces dissociation of the Bcl-2/Bax dimer, thereby allowing the pro-apoptotic action of Bax on mitochondria 32,33,35,36 . Bmf is expressed in various organs, including the kidney 33 . Mice deficient in Bmf −/− , however, do not display any obvious phenotypic abnormalities 37,38 . Recent studies by Pfeiffer et al. demonstrated that Bmf not only plays a major role in progressive pancreatic β-cell death in HNF1α-MODY, but also contributes to the pancreatic beta-cell function to maintain glucose homeostasis, independent of cell death signaling 39 . However, little is known about the role of Bmf in the kidney or in the pathogenesis of diabetes. We previously reported the presence of increased Bmf expression in apoptotic RPTCs in db/db mice as well as in human diabetic kidneys 22 .
To provide direct evidence for Bmf-mediated RPTC apoptosis in vivo, we generated Tg mice specifically overexpressing hBMF in their RPTCs. Our data demonstrate that overexpression of hBMF in RPTCs indeed induces RPTC apoptosis and loss into the urine in hBMF-Tg mice. Furthermore, overexpression of hBMF increased expression of Bax and cleaved (active) caspase-3 and enhanced Bax binding to Bcl2, consistent with Bax-mediated activation of the intrinsic (mitochondrial) pathway of apoptosis.
The mechanisms underlying increased SBP and ACR in hBMF-Tg mice are incompletely understood. The possibility that up-regulation of TGFβ1 gene expression in RPTCs and its downstream targets Fn1 and Col 1α, leading to higher tubulointerstitial fibrosis that facilitates the development of hypertension has received considerable attention. Indeed, a strong correlation was found between interstitial fibrosis and the development of SBP via a loss or rarefaction of capillaries around the tubules 40,41 . Our observations of increased glomerular tuft volume, RPTC volume and GFR in hBMF-Tg mice lend additional support to this notion. Furthermore, the observed increases in ACR can be explained, at least in part, by a combination of loss of RPTCs and increased GFR 42 .
The Akita mouse is an autosomal dominant spontaneous diabetic mouse model (mutation in insulin2) that exhibits many features resembling changes in T1D patients including hypoinsulinemia, hyperglycemia, hypertension and renal dysfunction 43,44 .
Insulin suppressed Bmf and stimulated hnRNP F expression in RPTCs of hyperinsulinemic-euglycemic mice after 3 h of hyperinsulinemia as compared to Akita mice after 4 weeks of insulin implantation. This rapid transcription is in agreement with previous reports that up-regulated and down-regulated genes in the skeletal muscle and liver were observed within 2 to 4 h under hyperinsulinemic-euglycemic condition 45,46 . These data demonstrate that insulin impacts RPT Bmf and hnRNP F expression, independent of its glucose-lowering effect.
Our data also revealed significantly higher expression of Bmf protein and mRNA in RPTs of 16 weeks old Akita than WT mice. Overexpression of hnRNP F significantly down-regulated Bmf expression in Akita hnRNP F-Tg mice, suggesting a role for hnRNP F in mediating insulin inhibition of Bmf expression in Akita mice. However, it is noteworthy that we could not exclude the off-target effects of random transgene insertion in mediating insulin suppression of Bmf expression in our hnRNP F-Tg mice. Addressing this issue would require mapping transgene insertion sites which is outside the scope of this present study. Generation of currently unavailable RPTC-specific hnRNP F knockout mice would provide more direct evidence of the role of hnRNP F on Bmf expression. We are now working on developing such a model for future work.
Combining studies with pharmacological inhibitors and siRNAs, we identified roles for the p44/42 MAPK signaling pathway and hnRNP F in mediating insulin suppression of renal Bmf gene transcription. These findings clearly link hnRNP F to mediating insulin inhibition of Bmf gene expression in the diabetic mouse kidney. Nevertheless, additional studies employing RPTC-specific hnRNP F knockout mice are needed to firmly establish this pathway.
The mechanism by which hnRNP F down-regulates renal Bmf gene expression remains to be investigated. A likely mechanism is that hnRNP F suppresses Bmf transcription via binding to a putative IRE in the Bmf promoter. This is supported by the findings that transfection of hnRNP F cDNA considerably decreases Bmf promoter activity, whereas transfection of hnRNP F siRNA reverses insulin effect. DNA sequence analysis revealed 3 GC-rich regions in the Bmf promoter including nucleotides N-1086 to N-1081 (5′-AGGGGG-3′), N-997 to N-991 (5′-GAGGGGC-3′) and N-402 to N-395 (5′-CCCCCGC-3′). Nucleotides N-1086 to N-1081 contain the sequence 5′-AGGGGG-3′ which is homologous to the IRE sequence of N-402 to N-398 (5′-AGGGGG-3′) and N-974 to N-969 (5′-AGGGGG-3′) in the rat Sirtuin-1 and Ace2 promoter, respectively 29,47 . Deletion of N-1086 to N-1081 in the Bmf promoter markedly reduced insulin-and hnRNP F-down-regulation of Bmf promoter activity in IRPTCs whereas deletion of N-997 to N-991 and N-402 to N-395 had no effect, indicating that the 5′-and 3′-flanking sequences of the core sequence might be important for hnRNP F binding. Moreover, biotinylated IRE (N-1093 to N-1072) binds to nuclear proteins of RPTCs. Addition of anti-hnRNP F antibody yielded a supershift of biotinylated IRE complex with nuclear proteins on EMSA. These findings identify N-1093 to N-1072 as a putative IRE that interacts with hnRNP F and inhibits Bmf transcription.
It is noteworthy that hnRNP F not only regulates Bmf transcription as it also modulates the transcription of Agt 28 , Sirtuin-1 29 , Nrf2 31 , Ace2 47 and possibly other genes 48 . Thus, the hnRNP F effect on Bmf expression might also be mediated by other (indirect) mechanisms. Furthermore, the role of hnRNP F in human pathophysiology remains to be explored. More studies are definitely needed along these lines. (2019) 9:6687 | https://doi.org/10.1038/s41598-019-43218-2 www.nature.com/scientificreports www.nature.com/scientificreports/ In summary, our results demonstrate that insulin inhibits Bmf transcription and prevents RPTC apoptosis and loss via p44/42 MAPK signaling and hnRNP F expression (Fig. 8) These findings would imply that Bmf activation could aggravate tubulopathy via inducing RPTC loss in the diabetic kidney. However, it remains to be seen whether renal hnRNP F might be a potential target for the prevention of tubulopathy in the diabetic kidney. Generation of human BMF-Tg and akita hnRNP F-Tg mice. Tg mice specifically overexpressing human myc-BMF in their RPTCs were generated using a similar strategy to that described previously 20,28,49 . In brief, full-length human BMF cDNA fused with Myc-tag was inserted into pKAP2 plasmid at the NotI site. The plasmid pKAP2 containing the KAP promoter that is responsive to androgen was a gift from Dr. Curt D. Sigmund (University of Iowa, Iowa City, IA) 50 .

Methods
Akita hnRNP F-Tg mice were generated by cross-breeding hnRNP F-Tg mice with heterozygous Akita (C57BL/6-Ins2 Akita/J) mice (Jackson Laboratory, Ann Harbor, ME) as previously described 28 . Physiological studies. Adult male non-Tg littermates (wild type (WT)) and myc-hBMF-Tg mice were studied at the age of 10 to 20 weeks. Akita mice and Akita hnRNP F-Tg mice were studied at the age of 10 to 16 weeks. Male non-Akita littermates (controls) and Akita mice at the age of 10 weeks treated ± insulin implants and studied from week 12 until week 16 41 . All animals had access to water and standard mouse chow ad libitum. www.nature.com/scientificreports www.nature.com/scientificreports/ Animal care and procedures followed the Principles of Laboratory Animal Care [National Institutes of Health (NIH) publication no. 85-23, revised 1985: http://grants1.nih.gov/grants/olaw/references/phspol.htm] and were approved by the Centre de recherche du centre hospitalier de l'Université de Montréal (CRCHUM) Animal Care Committee.
Mouse blood glucose levels after 4 to 5 h of fasting were measured by Accu-Chek Performa (Roche Diagnostics, Laval, QC, Canada). Morning SBP was monitored at least 2-3 times per week in each animal for 6 or 10 weeks with Visitech Systems BP-2000 tail-cuff (Apex, NC) 28,30,31,47,51 . The mice were acclimatized to the procedure for at least 15-20 min per day for 5 days before the first SBP measurements.
All animals were individually housed in metabolic cages for 8 h during day time for urine collection and blood was collected by cardiac puncture (centrifuged to obtain serum) before euthanasia at 16 or 20 weeks old. Urine samples were assayed for ACR by enzyme-linked immunosorbent assay (ELISA) (Albuwell and Creatinine Companion, Exocell, Inc., Philadelphia, PA). Immediately after euthanasia, the kidneys were removed, decapsulated, and weighed. Left kidneys were processed for histology and immunostaining and right kidneys for RPT isolation by Percoll gradient 28,30,31,47,51 . Aliquots of freshly isolated RPTs from individual mice were processed for isolation of total RNA and protein.
In a separate series of studies, conscious male C57Bl/6 mice (aged 12 to14 weeks) after a 4-hour food restriction were used for hyperinsulinemic-euglycemic clamp experiments as previously described 31 . Morphologic studies. Kidney sections of 4 µm thick from five to six animals/group were assessed by standard periodic acid-Schiff (PAS) or Masson's trichrome staining. Immunohistochemistry staining (IHC) was performed by the standard avidin-biotin-peroxidase complex method (ABC Staining, Santa Cruz Biotechnology, Santa Cruz, CA). Semi-quantitation of the relative staining was done by NIH Image J software (http://rsb.info. nih.gov/ij/). Tubular injury score, mean glomerular tuft, tubular luminal area, and RPTC volumes were assessed on PASstained sections at ×200 magnification as described previously 28,30,31,47,51 . Briefly, PAS stained images (8-10 fields/ kidney) were assessed for tubular injury in the cortex area that displayed tubular dilation, tubular atrophy, cast formation, vacuolization, degeneration, and loss of the brush border. The tubules were evaluated according to the following injury grade (0-3): 0 = no tubular injury, 1 = less than 25% tubules injured, 2 = 25-50% injured, 3 = more than 50% tubules injured 52 . The score corresponding to tubular injury was calculated for each group by summing and then averaging the grades for each field.
The mean glomerular volume (V G ) was determined by the method of Weibel 53 with the aid of an image analysis software system (Motics Images Plus 2.0, Motic, Richmond, BC, Canada). The V G was estimated by the mean glomerular tuft area (A T ) derived from the light microscopic measurement of 30 random sectional profiles of glomeruli from each group (n = 6 animals per group) using the formula: V g = β/k x A T 1.5 , where β = 1.382 (shape coefficient for spheres) and k = 1.1 (size distribution coefficient).
Tubular luminal areas were measured on renal sections (six animals/group; 4 to 5 sections per kidney, 4 random fields per section, 50 tubules around the glomerulus per field) with the same Motics Image Plus 2.0 image analysis software.
Outer cortical RPTs with similar cross-sectional views and clear nuclear structure were selected for cell volume measurement. Mean cell volume from 100 RPTCs was estimated by the Nucleator method 54 .

Real-Time quantitative PCR.
Real-time quantitative (q) PCR was used to quantify the mRNA levels of various genes in RPTs with the forward and reverse primers listed in Supplementary Table 2 28,30,31,47,51 . Cell culture. Rat IRPTCs (passages 13 through 18) were used 55 . Rat Bmf gene promoter inserted into the plasmid pGL4.20 was stably transfected into IRPTCs as described previously 28,30,31,47,51 .
To investigate insulin effect, IRPTC stable transformants (75% to 85% confluence) were synchronized for 16 hrs in serum-free DMEM (5 mmol/L D-glucose). Then they were incubated in normal glucose (NG, 5 mmol/L D-glucose plus 20 mmol/L D-mannitol) or HG (25 mmol/L D-glucose) DMEM containing 1% depleted fetal bovine serum (FBS) and insulin (10 −7 mol/L or 573 ng/mL) for up to 24 hours ± PD98059 or U0126 or wortmannin 31 . The cells were then harvested, and the Bmf gene promoter activity was measured by luciferase assay 31 . IRPTCs stably transfected with pGL4.20 served as controls. In addition, scrambled siRNA or siRNAs of p44/42 MAPK or hnRNP F were transfected into IRPTC stable transformants and cultured for 24 hours, their effects on Bmf gene promoter activity and mRNA expression were assessed.
Flow cytometry analysis of urinary cells. Urine was collected for 6 hrs from individual non-Tg or myc-hBMF-Tg mice placed individually in metabolic cages. Urines from 5 mice in each group were collected, pooled and centrifuged at 500 × g for 10 minutes to pellet down the cells in the urine. Pellets were rinsed with PBS and passed through a 35 μm cell strainer (#352235, Corning Inc., Corning, USA). Each cell suspension was then blocked with 2% FBS in PBS for 10 minutes then stained with DRAQ7 and anti-prominin-1 antibody conjugated with PE-Vio770 (MiltenyiBiotec, BergischGladbach, Germany, 1:25 dilution) 34 for 30 minutes at room temperature in the dark, then washed with PBS; equal volumes of suspension from non-Tg or myc-hBMF-Tg mice were