Heterogeneous Nuclear Ribonucleoprotein K Modulates Angiotensinogen Gene Expression in Kidney Cells*

The present studies aimed to identify the 70-kDa nuclear protein that binds to an insulin-responsive element in the rat angiotensinogen gene promoter and to define its action on angiotensinogen gene expression. Nuclear proteins were isolated from rat kidney proximal tubular cells and subjected to two-dimensional electrophoresis. The 70-kDa nuclear protein was detected by Southwestern blotting and subsequently identified by mass spectrometry, which revealed that it was identical to 65-kDa heterogeneous nuclear ribonucleoprotein K (hnRNP K). hnRNP K bound to the insulin-responsive element of the rat angiotensinogen gene was revealed by a gel mobility shift assay and chromatin immunoprecipitation assay. hnRNP K inhibited angiotensinogen mRNA expression and promoter activity. In contrast, hnRNP K down-expression by small interference RNA enhanced angiotensinogen mRNA expression. Moreover, hnRNP K interacted with hnRNP F in pulldown and co-immunoprecipitation assays. Co-transfection of hnRNP K and hnRNP F further suppressed angiotensinogen mRNA expression. Finally, in vitro and in vivo studies demonstrated that high glucose increases and insulin inhibits hnRNP K expression in rat kidney proximal tubular cells. In conclusion, our experiments revealed that hnRNP K is a nuclear protein that binds to the insulin-responsive element of the rat angiotensinogen gene promoter and modulates angiotensinogen gene transcription in the kidney.

Diabetic nephropathy is a leading cause of end stage renal disease, accounting for 30 -50% of all new end stage renal disease cases in North America (1)(2)(3). Both clinical and animal studies indicate that intensive insulin therapy and prolonged treatment with angiotensin-converting enzyme inhibitors or angiotensin II-AT 1 receptor blockers delay the progression of nephropathy in diabetes, but neither strategy cures nephropa-thy (4 -12). Whereas such results support the concept that hyperglycemia and the renin-angiotensin system (RAS) 2 activation are involved in the development and progression of diabetic nephropathy, the molecular mechanism(s) linking hyperglycemia to RAS activation remain largely undefined.
Angiotensinogen (AGT), a glycoprotein consisting of 452 amino acid residues with an apparent molecular mass of 62-65 kDa, is the sole substrate in the RAS cascade (13,14). AGT is principally produced by the liver and cleaved by renin from the kidney to form angiotensin I, which is then further processed by angiotensin-converting enzyme to form angiotensin II. The existence of an intrarenal RAS is now generally accepted (15,16). Renal proximal tubules (RPTs) contain all components of the RAS, including messenger RNAs and proteins, such as AGT, renin, angiotensin-converting enzymes, and angiotensin II receptors (17)(18)(19)(20)(21)(22)(23)(24). Most recently, we reported that RAS blockade decreases blood pressure and proteinuria in transgenic mice overexpressing rat AGT (rAGT) gene in the kidney (25). These observations indicate that the local formation of angiotensin II may play an important role in the development of nephropathy in diabetes.
Our laboratory has established that high glucose (i.e. 25 mM) stimulates rAGT gene expression in IRPTCs (26 -29). RAS blockers and stable transfer of antisense rAGT cDNA into IRPTCs inhibit transforming growth factor-␤1 gene expression and cellular hypertrophy in high glucose (30,31). These investigations strongly indicate that rAGT and transforming growth factor-␤1 gene expression are essential for the high glucose effect on IRPTC hypertrophy and kidney injury. We have also established that insulin inhibits the stimulatory effect of high glucose levels on rAGT gene expression and the induction of hypertrophy in IRPTCs (32)(33)(34)(35)(36)(37)(38)(39)(40)(41)(42)(43). Moreover, a putative insulin-responsive element (IRE) containing nucleotides Ϫ878 to Ϫ864 (5Ј-CCT TCC CGC CCT TCA-3Ј) upstream of the transcription start site of the rAGT gene promoter has been identified, and it binds to two major nuclear proteins with apparent molecular mass of ϳ48 and 70 kDa from IRPTCs (34). We recently reported that the 48-kDa nuclear protein is identical to the 46-kDa heterogeneous nuclear ribonucleoprotein F (hnRNP F) (35). Furthermore, transient transfer of sense and antisense hnRNP F cDNA, respectively, inhibits and enhances rAGT gene expression in IRPTCs (35).
The present studies aimed to identify the 70-kDa nuclear protein and to investigate its action on rAGT gene expression. We identified the 70-kDa nuclear protein as 65-kDa heterogeneous nuclear ribonucleoprotein K (hnRNP K) by two-dimensional electrophoresis and mass spectrometry (MS). Recombinant hnRNP K bound to rAGT-IRE, as shown by a gel mobility shift assay (GMSA) and chromatin immunoprecipitation assays. Overexpression of hnRNP K attenuated rAGT mRNA expression and rAGT gene promoter activity in IRPTCs. In contrast, down-expression of hnRNP K by small interference RNA enhanced rAGT gene expression. Moreover, hnRNP K was pulled down and co-immunoprecipitated with hnRNP F. Co-transfection of hnRNP K and hnRNP F further suppressed rAGT mRNA expression. Finally, in vitro and in vivo studies revealed that high glucose or hyperglycemia increased and insulin inhibited hnRNP K expression in rat kidney proximal tubular cells. These experiments demonstrated that FIGURE 1. Rat AGT IRE-BPs detected by two-dimensional electrophoresis and Southwestern and Western blottings. A, rat IRPTC nuclear extracts were subjected to two-dimensional electrophoresis and then stained with Coomassie Brilliant Blue R-250. M, Amersham Biosciences rainbow molecular mass markers. N.E., rat IRPTC nuclear extracts without isoelectrofocusing. B, Southwestern blotting analysis of AGT IRE-BPs from IRPTC nuclear proteins after two-dimensional electrophoresis. After two-dimensional electrophoresis, the nuclear proteins were transferred onto a Hybond C-extra membrane, hybridized with radioactively labeled 3ϫ AGT-IRE (Ϫ878/Ϫ864), washed, and subjected to autoradiography. The arrowheads indicate the proteins that were determined to be hnRNP K by later MS. The rectangle denotes the 70-kDa proteins that were subjected to MS. Their identities were determined as 65-kDa hnRNP K. C, Western blotting of the two-dimensional electrophoresis membrane from B using anti-hnRNP K antibody. FIGURE 2. MS analysis of proteins detected by Southwestern blotting. Spot 1 was isolated from twodimensional gel and then subjected to trypsin digestion and MALDI-MS analysis. The homology of peptide sequences with hnRNP K identified by MALDI-MS and data base search is shown in boldface type. The sequence coverage of hnRNP K reached 14%. Similar results were obtained from spot 2 (not shown). 65-kDa hnRNP K is a nuclear protein that binds to the rat AGT gene promoter and modulates AGT gene expression in the kidneys.
The plasmid containing the concatemeric wide type and mutant rAGT-IRE motif DNAs were constructed by inserting the double-stranded concatemeric wide type or mutant rAGT-IRE motif oligonucleotide with the NotI enzyme restriction site added on both termini into the polyclonal site of pcDNA 3.1 by conventional methodology. The double-stranded concatemeric wide type or mutant rAGT-IRE motif DNA fragment was then excised from the plasmid and treated with alkaline phosphatase and used for labeling as probe.
Two-dimensional Electrophoresis-Two-dimensional electrophoresis was carried out with the IPGphor isoelectric focusing unit (Amersham Biosciences) as previously described (35). For two-dimensional separation, the IPG strips were placed above 10% polyacrylamide gel containing SDS and electrophoresed (SDS-PAGE) (35). Amersham Biosciences rainbow markers served as molecular weight markers. IRPTC nuclear extracts (100 g) were run on the same 10% SDS-PAGE as the controls. Each sample was divided into two strips for two-dimensional electrophoresis. One gel was stained with Coomassie Brilliant Blue R-250 (Amresco Inc., Solon, OH) to visualize proteins. The other was electrotransferred to a Hybond C-extra membrane (Amersham Biosciences) for Southwestern blotting.
SouthwesternBlotting-Southwesternblottingwasperformed according to the procedure of Kwast-Welfeld et al. (43) with slight modifications (34,35). Briefly, IRPTC nuclear proteins (200 g) were resolved on a 4 -20% SDS-PAGE gradient or on 10% SDS-PAGE (44) and then electrotransferred to a Hybond C-extra mem-brane. The membrane was in-cubated with 10% (w/v) nonfat milk proteins and then washed at least twice with binding buffer containing 0.25% nonfat milk proteins. Subsequently, it was hybridized overnight with 32 P-labeled rAGT-IRE DNA (ϳ1.0 -2.0 pmol; 10 6 cpm/ml) in binding buffer containing 0.25% nonfat milk proteins and 300 g/ml nondenatured herring sperm DNA at 4°C. The membrane was finally washed, air-dried, and exposed for autoradiography. Expression of Recombinant hnRNP K-Murine hnRNP K cDNA (36), with the NotI enzyme restriction site added on the 5Ј and 3Ј ends of sense and antisense primers, respectively, was subcloned at the polyclonal site (NotI) of the bacterial expression vector pGex 4T-3 by conventional methodology. E. coli BL-21 cells (Amersham Biosciences) were transformed by pGex 4T-3 containing hnRNP K cDNA. Expression of the fusion protein (GST fused with hnRNP K (GST-hnRNP K)) in BL-21 cells was induced by the addition of 1 mM isopropylthiogalactoside in the culture medium with incubation for 4 h at 37°C. The bacteria were then harvested, and GST-hnRNP K fusion proteins were purified from the bacterial extracts by GST affinity column chromatography according to the manufacturer's protocol (Amersham Biosciences). The purified GST-hnRNP K fusion proteins were tested in GMSAs.

Matrix-assisted Laser Desorption/ Ionization-MassSpectrometry(MALDI-MS)-Spots
GMSAs-These assays were performed according to the methodology described elsewhere (34,35), employing labeled rAGT-IRE DNA as probe. Briefly, the rAGT-IRE DNA frag- Complexes were eluted, cross-linking was reversed, and purified DNA was used as a template in PCR with primers specific to the rat AGT gene promoter and ␤-actin gene. DNA was separated by agarose gel electrophoresis and visualized. Then the DNA was transferred onto a Hybond XL nylon membrane and hybridized with a digoxigenin-labeled internal DNA probe.

FIGURE 5. Effect of transient transfection of hnRNP K cDNA on AGT mRNA expression in IRPTCs.
A, Western blotting of hnRNP K protein in IRPTCs transiently transfected with pRC/RSV or pRSV/hnRNP K. B, RT-PCR analysis of endogenous rAGT and ␤-actin mRNA expression in IRPTCs transiently transfected with pRC/RSV or pRSV/hnRNP K. 24 h after gene transfection, the cells were cultured for 24 h in 5 mM Dulbecco's modified Eagle's medium containing 5% fetal bovine serum. Then the cells were harvested and assayed for hnRNP K and rAGT mRNA by Western blotting and RT-PCR assay, respectively. The relative levels of hnRNP K protein and AGT mRNA were compared with ␤-actin protein and ␤-actin mRNA, respectively. The hnRNP K and AGT mRNA level in pRC/RSV-transfected cells represents the control level (100%). The results were expressed as means Ϯ S.D. of three independent experiments, and each treatment group was assayed in triplicate (***, p Յ 0.005). ment was 5Ј-end-labeled with [␥-32 P]ATP by T 4 polynucleotide kinase. Purified GST-hnRNP K fusion proteins (1.5 g) or GST (5 g) in the presence of 1 g of poly(dI/dC) in 20 mM Hepes (pH 7.6), 1 mM EDTA, 50 mM KCl, 2 mM spermidine, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, and 10% glycerol (v/v) were incubated for 30 min on ice. Subsequently, the 5Ј-labeled probe (ϳ0.1 pmol) was added and further incubated for 45 min on ice. The mixture was run on 6% (w/v) nondenaturing PAGE and exposed for autoradiography. In competition assays, a 100-fold molar excess of unlabeled DNA fragments was added to the reaction mixture and incubated for 30 min on ice before incubation with the labeled probe.
Chromatin Immunoprecipitation (ChIP)-ChIP analysis was performed according to the methodology of Kuo and Allis (45) with slight modifications (46). Briefly, 0.4 ml of 37% formaldehyde was added to 10 ml of overlaying medium of IRPTC culture for 15 min at 4°C. After cross-linking, the cells were harvested, washed twice with 1 ml of phosphate-buffered saline in Eppendorf tubes, and then lysed with 0.5 ml of immunoprecipitation buffer (150 mM NaCl, 5 mM EDTA, 1% Triton X-100, 0.5% Nonidet P-40, 50 mM Tris-HCl, pH 7.5, 0.5 mM dithiothreitol) containing the following inhibitors: 10 g/ml leupeptin, 0.5 mM phenylmethylsulfonyl fluoride, 30 mM p-nitrophenol phosphate, 10 mM NaF, 0.1 mM Na 3 VO 4 , and 10 mM ␤-glycerophosphate. After one wash with immunoprecipitation buffer, the pellet was suspended in 1 ml of immunoprecipitation buffer and sheared in a Branson sonicator with 10-s cycles, 1 pulsed and 1 continuous, for 10 min at an output 3 and 80% duty cycle. Pulldowns were done using anti-K protein antibody with or without blocking peptide (100 M) and protein A beads (Amersham Biosciences). The beads were washed five times with 1 ml of immunoprecipitation buffer without inhibitors. DNA was eluted twice from the beads with 250 l of elution buffer (1% SDS, 0.1 M NaHCO 3 ) for 15 min with periodic vortexing (at room temperature). Cross-linking was reversed by adding 1 l of 10 mg/ml RNase and 5 M NaCl to a final concentration of 0.3 M and incubating the tubes at 65°C for 4 h. After adding 2.5 volumes of 100% ethanol, the pellet was precipitated overnight at Ϫ20°C. DNA was pelleted and resuspended in 100 l of Tris-EDTA buffer, pH 8.0. Then 11 l of 10ϫ protein K buffer (0.1 M Tris, pH 7.8, 50 mM EDTA, 5% SDS) and 1 l of 20 g/l proteinase K were added and incubated at 45°C for 2 h. DNA was extracted with phenol/chloroform and precipitated with ethanol, and the final DNA pellet was dissolved in 20 l of Tris-EDTA buffer.
PCR amplifications were done in 50 l of 1ϫ PCR buffer containing DNA, 0.5 M primers (forward primer (5Ј-CCT TGA TGC CTC CAA CAA CT-3Ј) and backward primer (5Ј-GGT GGG AGC TGA GAA GAC AG-3Ј), corresponding to nucleotides Ϫ1043 to Ϫ1026 and Ϫ718 to Ϫ698 of the rAGT gene promoter (47), or forward primer (5Ј-TCC TGT GGC ATC CAT GAA ACT AC-3Ј) and backward primer (5Ј-AGC ATT TGC GGT GCA CGA TGG AG-3Ј), corresponding to nucleotides ϩ808 to ϩ830 and ϩ1120 to ϩ1098 of the rat ␤-actin (48), respectively), a 40 M concentration of each deoxynucleotide triphosphate, 1.5 mM MgCl 2 , and 1 unit of Taq DNA polymerase (Invitrogen). The PCR products were resolved on 2% agarose gel and transferred onto a Hybond XL nylon membrane (Amersham Biosciences). Digoxigenin-labeled oligonucleotide 5Ј-CCT CCC TTC CCG CCC TTC ACT TTC TAG T-3Ј and 5Ј-CCA ACT CTC TTG GCT TAA GGA-3Ј, corresponding to nucleotides Ϫ882 to Ϫ855 of the rAGT gene promoter (47) and intron 4 nucleotides ϩ56 to ϩ77 of the ␤-actin gene (48), respectively, prepared with a digoxigenin oligonucleotide 3Ј-end labeling kit (La Roche Biochemicals), served to hybridize the PCR products on the membrane. After stringent washing, the membrane was detected with a digoxigenin luminescence kit (La Roche Biochemicals) and exposed to Eastman Kodak Co. BMR film.
Mammalian Expression of Recombinant hnRNP K-Murine hnRNP K cDNA with FLAG tag at the N terminus in mammalian expression vector pRC/RSV (36) was transfected into IRPTCs with Lipofectamine according to the instruction manual provided by the supplier (Invitrogen). We optimized the DNA concentration for gene transfection at 2-3 g per 0.5-1 ϫ 10 6 cells. 48 h after transfection, total RNAs and nuclear proteins were isolated from IRPTCs and assayed for rAGT mRNA by RT-PCR (34,35) or for hnRNP K protein by Western blotting, respectively. Western Blotting for hnRNP K-Briefly, the cell pellets were lysed in 100 l of RIPA buffer and centrifuged, and 30 g of the supernatants were subjected to 10% SDS-PAGE (44) and then transferred onto a polyvinylidene difluoride membrane (Hybond-P; Amersham Biosciences). The membrane was initially blotted for anti-hnRNP K antibody (1:4,000 dilutions) and then reblotted for anti-␤-actin antibody (1:10,000 dilutions) and chemiluminescent developing reagent (La Roche Biochemicals). The relative densities of the hnRNP K and ␤-actin bands were quantified by computerized laser densitometry (ImageQuant software (version 5.1); Amersham Biosciences).
Pulldown Assays-The cellular extracts were incubated with glutathione-Sepharose 4B beads with or without bound GST-hnRNP K or GST-hnRNP F (35) for 4 h at 4°C. The beads were washed five times with 1 ml of RIPA buffer, and the proteins were resolved by 10% SDS-PAGE (44) as described above. The membrane was blotted with anti-hnRNP F (1:10,000 dilutions) or anti-hnRNP K antibody (1:4,000 dilutions) or anti-TBP antibody (1:1,000 dilutions) and developed with chemiluminescent developing reagent.
Co-immunoprecipitation Assays-IRPTC nuclear proteins were incubated with rabbit anti-hnRNP F polyclonal antibody (10 g) or normal rabbit IgG antibody (10 g) for 4 h in RIPA buffer at 4°C. Then protein A-Sepharose beads were added and incubated for 1 h at 4°C. The beads were washed five times with RIPA buffer. The retained proteins were eluted with 2ϫ loading buffer and subjected to SDS-PAGE and Western blotting with anti-hnRNP K antibody. Chloramphenicol Acetyltransferase (CAT) Assay-The method of construction of the rAGT-CAT fusion gene, pOCAT/ rAGT Ϫ1498/ϩ18 (fusion gene containing 1,498 nucleotides upstream of the transcription start site and 18 nucleotides of exon I fused with the CAT reporter gene), and mutant pOCAT/rAGT Ϫ1498/ϩ18 with mutated IRE has been described previously (33,35). Control plasmid or fusion gene was transfected into IRPTCs using Lipofectamine (Invitrogen) according to methods described previously (33,35). 48 h after transfection, the cells were harvested and assayed for CAT activity (33,35).
To normalize the efficiency of transfection, 0.5 g of pRSV/ ␤-Gal (a vector with the Rous sarcoma viral enhancer/promoter fused to the bacterial ␤-galactosidase gene) was co-transfected with pOCAT/rAGT Ϫ1498/ϩ18 (33,35). The plasmid pRSV/CAT (a vector with the Rous sarcoma viral enhancer/promoter fused to the CAT reporter gene) served as a positive control to monitor the efficiency of transfection of rAGT-CAT fusion gene. The level of transfection efficiency for FIGURE 7. Effect of siRNA of hnRNP F and hnRNP K on rAGT mRNA expression in IRPTCs. 48 h after transfection, the cells were harvested and assayed for hnRNP F or hnRNP K protein (A) and AGT mRNA expression (B) by Western blotting and RT-PCR, respectively. The level of hnRNP F or hnRNP K and AGT mRNA expression was compared with ␤-actin protein and mRNA, respectively. The results were expressed as means Ϯ S.D. of three independent experiments, and each treatment group was assayed in duplicate (**, p Յ 0.01; ***, p Յ 0.005; N.S., not significant). pRSV/CAT in IRPTCs ranged from 60 to 90% (i.e. the percentage of conversion of 14 C-chloramphenicol to mono-and diacetyl chloramphenicol). The transfection efficiency of pOCAT/ rAGT Ϫ1498/ϩ18 in IRPTCs ranged from 25 to 35% compared with pRSV-CAT. The inter-and intra-assay coefficient variations of transfection for pOCAT/rAGT Ϫ1498/ϩ18 in IRPTCs were 25 and 12% (n ϭ 10), respectively.
All animals were allowed free access to rat chow and water. All methods of animal care and sacrifice were approved by the Animal Care Committee of the Centre Hospitalier de l'Université de Montréal.
Isolation of Rat RPTs-Two weeks after the induction of diabetes, the rats were anesthetized and euthanized (control and treated rats were euthanized at the same time point). Kidneys were removed immediately for proximal tubule isolation. The renal cortex was separated from the medulla and minced under sterile conditions. Proximal tubules were isolated by Percoll gradient (50) with slight modifications (49). Proximal tubular cells were characterized by their histological appearance (50). A highly purified preparation of proximal tubules (Ͼ97% by microscopy) with Ͼ95% viability (determined by trypan blue exclusion) was obtained. Aliquots of freshly isolated proximal tubules from individual rats were immediately used for total RNA and protein isolation.
Statistical Analysis-3-5 separate independent experiments were performed per protocol, and each treatment group was run in duplicate. The data were analyzed by one-way analysis of variance and the Bonferroni test. A probability level of p Յ 0.05 was regarded as statistically significant.

Identification of 70-kDa IRE-binding Proteins (IRE-BPs) in
IRPTCs- Fig. 1A shows the staining of nuclear proteins after two-dimensional electrophoresis. Southwestern blotting of IRE-BPs after two-dimensional electrophoresis is displayed in Fig. 1B. Positive spots with apparent molecular mass of 70 and 48 kDa were cut out and subjected to MALDI-MS. The MS results of 70-kDa proteins are displayed in Fig. 2. The two spots with an apparent molecular mass of 70 kDa were identified as a common protein (accession number NM_057141). Data base analysis revealed that they are identical to the rat hnRNP K cDNA sequence reported by Ito et al. (51). The two spots with an apparent molecular mass of 48 kDa were identified as 46-kDa hnRNP F (35) (data not shown).
To confirm the authenticity of rat hnRNP K revealed by Southwestern blotting, we stripped the radioactivity from the membrane and reblotted it with rabbit polyclonal antiserum against hnRNP K, as shown in Fig. 1C. It is apparent that the proteins interacting with anti-hnRNP K were superimposable, with positive signals detected by Southwestern blotting, as seen in Fig. 1B. These data confirm that the 70-kDa proteins that interact with AGT-IRE are identical to hnRNP K.
GMSA of Radioactively Labeled rAGT-IRE DNA Fragment with GST-hnRNP K Fusion Protein(s)-Bacterially expressed recombinant hnRNP K proteins were employed to study the interaction of rAGT-IRE with hnRNP K. One major band appeared with retarded mobility with labeled rAGT-IRE by employing GST-hnRNP K fusion protein (Fig. 3A). No slowly migrating band was observed when the labeled DNA was incubated with GST (bacterial extract of empty vector pGex 4T-3). The addition of an unlabeled rAGT-IRE was effective in competing with the binding of labeled rAGT-IRE DNA to the fusion proteins(s) (100-fold molar excess of unlabeled DNA fragment) but not the unlabeled DNA fragment of hGAPDH-IRE, rat glucagon-IRE, and the Sp1 consensus sequence (Fig. 3A) or mutants (M1 and M2) of rAGT-IRE (Fig. 3B). GST-hnRNP K bound specifically to wild type rAGT-IRE but not to mutant rAGT-IRE (Fig. 3C).
ChIP Analysis of hnRNP K Interactions with Gene Loci-ChIP assays were used to test if hnRNP K interacts with the IRE of the rAGT gene promoter in vivo. Fig. 4 displays the PCR product of pulled down DNA with primers specific to rAGT and the ␤-actin gene. A ϳ300-bp DNA fragment was apparent in input control cells and cells that were treated with the cross-linking agent and without blocking peptide 48 h after transfection of pRSV/CAT, wild-type pOCAT/rAGT Ϫ1498/ϩ18, or mutant pOCAT/rAGT Ϫ1498/ϩ18 with or without hnRNP F or hnRNP K cDNA, the cells were harvested and assayed for CAT activity. Relative activity in cells transfected with 2.5 g of rAGT Ϫ1498/ϩ18 or its mutant was given a relative value of 100% (control). Each point represents the mean Ϯ S.D. of three independent experiments, and each treatment group was assayed in duplicate (***, p Յ 0.005; N.S., not significant). 2 and 4). In contrast, no PCR product was generated in cells without treatment with hnRNP K antibody (lane 3) or in the presence of blocking peptide (lane 5). PCR product specific to ␤-actin was also apparent in input control cells (lane 2) but not in cells that were treated with or without hnRNP K antibody in the absence or presence of hnRNP K blocking peptide. These results validate the interaction of hnRNP K with rAGT gene loci in vivo.

(lanes
Effect of hnRNP K on rAGT mRNA Expression in IRPTCs-It is apparent that hnRNP K protein levels in IRPTCs transiently transfected with pRSV/hnRNP K were significantly higher (Ͼ2-fold increase, p Ͻ 0.005) than those in control pRC/RSV-transfected IRPTCs (Fig. 5A) by Western blotting. In contrast, rAGT mRNA expression was significantly lower (50% decrease, p Ͻ 0.005) in pRSV/hnRNP K-transfected cells than those in pRC/RSV-transfected cells analyzed by RT-PCR (Fig. 5B). These results demonstrated that hnRNP K inhibits rAGT mRNA expression in IRPTCs.
High glucose stimulated (Ͼ3-fold increase, p Ͻ 0.01) and insulin inhibited rAGT mRNA expression in IRPTCs, as shown in Fig. 6A. In contrast, hnRNP K overexpression prevented the stimulatory effect of high glucose on rAGT mRNA expression in IRPTCs (Fig. 6B). These data suggest that hnRNP K modulates high glucose stimulation of rAGT gene expression in IRPTCs.
Effect of siRNA of hnRNP K on rAGT mRNA Expression in IRPTCs-Transient transfer of siRNA of hnRNP K and hnRNP F suppressed respective hnRNP K and hnRNP F expression (Fig. 7A) but enhanced rAGT mRNA expression in IRPTCs (Fig. 7B). In contrast, transient transfer of negative scrambled silencer had no effect on hnRNP K or hnRNP F protein and rAGT mRNA expression in IRPTCs. These data further support that both hnNRP F and hnRNP K modulate rAGT mRNA expression in IRPTCs.
Effect of hnRNP K on rAGT Gene Promoter Activity-Like hnRNP F (35), co-transfection with hnRNP K significantly suppressed pOCAT-rAGT fusion gene promoter activity ( Fig. 8) but had no effect on mutant pOCAT-rAGT fusion gene promoter activity. These studies further confirm the notion that hnRNP K modulates rAGT gene expression at the transcriptional level via binding to IRE.
Interaction of hnRNP K with hnRNP F in Pulldown and Co-immunoprecipitation Assays- Fig. 9A shows that GST-hnRNP K was pulled down with hnRNP F from IRPTCs. Vice versa, GST-hnRNP F was pulled down with hnRNP K from IRPTCs (Fig. 9B). GST by itself had no effect. We also confirmed the interaction between hnRNP F and hnRNP K by co-immunoprecipitation (Fig. 9C). These data demonstrate that hnRNP K could form a heterodimer with hnRNP F. Fig. 10A shows that the ATG mRNA expression is significantly lowered in IRPTCs that had been stably transfected with pcDNA 3.1/hnRNP F (35) as compared with cells that had been stably transfected with control plasmid pcDNA 3.1. Transient transfection with hnRNP K further suppressed rAGT mRNA expression in a dose-dependent manner in IRPTCs that had been stably transfected with pcDNA 3.1/hnRNP F (Fig. 10B). These studies support the notion that hnRNP K and hnRNP F could act additively to modulate endogenous rAGT gene expression in IRPTCs. FIGURE 9. Interaction of hnRNP K with hnRNP F. Equal amounts of purified GST or GST-hnRNP K (A) or GST-hnRNP F (B) bound to glutathione beads were mixed with IRPTC extract and incubated for 4 h at 4°C. After extensive washes, the proteins were resolved by 10% SDS-PAGE, transferred to membrane, and blotted with anti-hnRNP F antibody (A) or anti-hnRNP K (B). IRPTC nuclear proteins (N.P.) were incubated with 10 g of anti-hnRNP F antibody (anti-F) or normal rabbit IgG. The immunocomplex was precipitated by protein-A-Sepharose and washed five times with RIPA buffer. The proteins were eluted and subjected to 10% SDS-PAGE, transferred to membrane, and blotted with anti-hnRNP K antibody (C). Similar results were obtained from two other experiments. W.B., Western blotting; I.P., immunoprecipitation. Vivo-Fig. 11, A and B, shows the results of respective Southwestern and Western blot analysis of IRPTC nuclear extracts for hnRNP K incubated in normal glucose (5 mM D-glucose plus 20 mM D-mannitol) or high glucose (25 mM D-glucose) medium in the absence or presence of insulin (10 Ϫ7 M). It is apparent that high glucose levels enhanced and insulin suppressed hnRNP K expression in IRPTCs in vitro.

Effect of High Glucose and Insulin on hnRNP K Expression in Vitro and in
Similarly, studies in vivo revealed that hyperglycemia up-regulated hnRNP K expression in diabetic rat RPTs, and insulin treatment normalized hnRNP K to nondiabetic levels (Fig. 12). These data demonstrate that hyperglycemia and insulin regulate hnRNP K expression in diabetic rat RPTs.

DISCUSSION
The present studies identified hnRNP K as one of the nuclear proteins that binds to IRE of the rAGT gene promoter and inhibits rAGT gene expression in IRPTCs. We reported recently that by employing a combination of proteomics and Southwestern blotting, 46-kDa hnRNP F was identified as the 48-kDa nuclear protein that binds to IRE of the rAGT gene promoter and inhibits rAGT gene expression in IRPTCs (35).
The present studies aimed to identify the molecular structure of the 70-kDa nuclear protein with the same approach as with 46-kDa hnRNP F. It is apparent that multiple IRPTC proteins are resolved by two-dimensional electrophoresis. A second gel run simultaneously with the first gel for Southwestern blotting demonstrated two positive spots with an apparent molecular mass of 70 kDa and pI of 5.0 -6.0, matching the stained protein spots on the first gel. After tryptic digestion and MALDI-MS, these two spots were found to match the partial amino acid sequence deduced from the hnRNP K sequence as reported by Ito et al. (51) (accession number NM_057141). Using specific rabbit antiserum against hnRNP K, we confirmed the identity of hnRNP K on the membrane by Southwestern blotting. Western blotting revealed positive signals superimposed on the positions of the two spots detected by Southwestern blotting. The reason why hnRNP K is present in two different forms (same apparent molecular mass but different pI) is presently unclear. One possibility is that these proteins might be isoforms or variants with different phosphorylated forms. We have further confirmed that the 48-kDa species by MALDI-MS analysis is 46-kDa hnRNP F, as we have reported previously (35) (data not shown).
hnRNP K was identified as one of these hnRNPs that binds cytidine-rich elements (52)(53)(54). hnRNP K is encoded by one gene and can be alternatively spliced to at least four isoforms with deduced molecular mass in the range of 50 -51 kDa, but in FIGURE 10. Effect of co-transfection of hnRNP K and hnRNP F on rAGT mRNA expression in IRPTCs. A, comparison of the AGT mRNA level in IRPTCs that had been stably transfected with plasmid pcDNA 3.1 or pcDNA 3.1/hnRNP F (35). B, pRSV/hnRNP K was transiently transfected into IRPTCs that had been stably transfected with pcDNA 3.1/hnRNP F. 24 h after transfection, the cells were harvested and assayed for rAGT mRNA by RT-PCR. The relative levels of AGT mRNA were compared with ␤-actin mRNA. The AGT mRNA level in IRPTCs that had been transiently transfected with 2 g of control plasmid pRC/RSV represents the control level (100%). The results were expressed as means Ϯ S.D. of three independent experiments, and each treatment group was assayed in duplicate (**, p Յ 0.01; ***, p Յ 0.005, N.S., not significant).
SDS-PAGE, hnRNP K has an apparent molecular mass of 65 kDa (55). hnRNP K has been localized in the nucleus, cytoplasm, and mitochondria and implicated in chromatin remodeling, transcription, splicing, and translation processes (see reviews by Bomsztyk et al. (56,57)). hnRNP K binds single-stranded and double-stranded DNA motifs (CT element, 5Ј-dTCCC) within the promoter of c-Myc, c-Src, and c-Fos genes and complexes with Sp1 and TATA-binding protein (TBP) to modulate gene transcription (58 -61). hnRNP K also represses the transcription of thymidine kinase (62), the neuronal nicotinic acetylcholine receptor ␤4 subunit (63), and osteocalcin (64). Thus, hnRNP K is a multifunctional protein that interacts with DNA, RNA, and transcriptional or translational molecules to alter the in vivo rate of gene transcription and translation (either stimulation or repression).
To demonstrate that hnRNP K interacts with rAGT-IRE, we expressed hnRNP K in a bacterial system. Our GMSAs revealed that labeled rAGT-IRE binds to GST-hnRNP K fusion protein. The addition of unlabeled rAGT-IRE DNA effectively displaced labeled rAGT-IRE at or greater than a 100-fold molar excess of unlabeled DNA, whereas mutant rAGT-IRE was not effective. Similarly, unlabeled hGAPDH-IRE, rat glucagon-IRE, and the Sp1 consensus sequence were not effective in displacing labeled rAGT-IRE. Moreover, GST-hnRNP K did not bind mutant rAGT-IRE. These studies indicate that rAGT-IRE binds specifically to hnRNP K. We did not observe, however, a supershift of labeled rAGT-IRE binding with hnRNP K fusion protein by several different batches of anti-hnRNP K antibodies generated in our laboratory (K. Bomsztyk) or obtained from Dr. Gideon Dreyfuss (data not shown). At present, we have no explanation for this observation. One possibility is that the binding affinity of antibodies is not high enough. The second possibility is that the structure of hnRNP FIGURE 11. Effect of high glucose and insulin on hnRNP K expression in IRPTCs analyzed by Southwestern and Western blotting. The cells were incubated in 5 mM plus 20 mM D-mannitol, 25 mM D-glucose, or 25 mM D-glucose plus insulin (10 Ϫ7 M) and 5% fetal bovine serum for 24 h. Then the cells were harvested, and nuclear proteins (100 g) were electrophoresed on SDS-PAGE (4 -20% (w/v) gradient gel), transferred onto a polyvinylidene difluoride membrane, and analyzed by Southwestern blotting with radioactive rAGT-IRE probe (A) and then by Western blotting with rabbit polyclonal antibodies against hnRNP K (B) and ECL-chemiluminescent developing reagent. The same membrane was also hybridized with antibodies against ␤-actin control. The relative densities of the hnRNP K bands were compared with the ␤-actin control. The hnRNP K level expressed in IRPTCs in 5 mM glucose medium was considered to be the control (100%). Each bar represents the mean Ϯ S.D. of three independent experiments, and each treatment group was assayed in duplicate (*, p Յ 0.05; **, p Յ 0.01). FIGURE 12. hnRNP K expression in nondiabetic and diabetic rat RPTs. After 2 weeks of STZ-induced diabetes with or without insulin (Ins.) supplementation, RPTs were collected, extracted for proteins, and assayed for hnRNP K and ␤-actin levels by Western blotting. The relative levels of hnRNP K were normalized with ␤-actin. hnRNP K levels in nondiabetic rats were considered as 100%. Each point represents the mean Ϯ S.D. of five rat RPTs. ***, p Յ 0.001. K might hinder the binding with anti-hnRNP K antibodies. Clearly, more experiments are needed to clarify this observation.
Most convincingly, our ChIP assays revealed that hnRNP K interacts with rAGT gene promoter loci. Taken together, these data unequivocally demonstrate that hnRNP K binds to rAGT-IRE.
Interestingly, transient transfection of hnRNP K inhibits rAGT gene expression in IRPTCs, and stable transfection of hnRNP K cDNA prevented the stimulatory effect of high glucose on rAGT mRNA expression in IRPTCs. In contrast, suppression of hnRNP K expression by siRNA enhances rAGT gene expression in IRPTCs. To the best of our knowledge, this is the first report that hnRNP K could modulate rAGT gene expression in kidney proximal tubular cells in vitro. The negative effect of hnRNP K is similar to that of hnRNP F on rAGT gene expression (35). At present, the molecular mechanism(s) of hnRNP K action on rAGT mRNA expression is not known. One possibility is that hnRNP K behaves like a negative transacting protein and inhibits the binding of other positive transacting factor(s) to TBP and RNA polymerase II, subsequently attenuating rAGT gene expression. This possibility is supported by findings, including ours, 3 that the hnRNP K molecule binds directly to TBP and, therefore, may inhibit the basal transcription machinery (37,53). The second possibility is that hnRNP K overexpression inhibits the formation of an activating transcriptional complex on the promoter and subsequently represses rAGT gene expression. This possibility is supported by the studies of Du et al. (63) and Da Silva et al. (65), showing that hnRNP K inhibits Sp1 binding to the promoter of the gene encoding the ␤ 4 subunit of the nicotinic acetylcholine receptor and CD 43 gene promoter, respectively. The third possibility is that hnRNP K could recruit unidentified repressor molecules and subsequently repress rAGT gene expression. This possibility is supported by previous studies demonstrating that hnRNP K could bind the murine repressor Zik1 (66). Finally, it is also possible that hnRNP K could form a heterodimer with hnRNP F (a hypothetical stable complex) and then bind to rAGT-IRE and subsequently attenuate rAGT gene expression. This possibility is supported by the findings that hnRNP K could form an heterodimer or trimer with hnRNP A1/or hnRNP C to inhibit human thymidine kinase promoter (62). Indeed, this possibility is further supported by our findings that hnRNP K was pulled down and co-immunoprecipitated with hnRNP F. Co-transfection of hnRNP K and hnRNP F further suppressed rAGT mRNA expression. Clearly, more work is needed along these lines to elucidate the mechanism(s) of action of hnRNP K on rAGT gene expression in IRPTCs.
Finally, our studies revealed that high glucose stimulated and insulin inhibited hnRNP K expression in IRPTCs in vitro and in diabetic rat kidneys in vivo. These data are consistent with our previous studies in IRPTCs in vitro (34). To the best of our knowledge, this is the first report that hnRNP K could be modulated by high glucose and insulin in kidney proximal tubular cells in vitro and in vivo. The exact physiological role(s) of hnRNP K in renal rAGT gene expression is unknown. Our preliminary data revealed that hnRNP K mRNA expression is significantly lower in RPTs of spontaneously diabetic BioBreeding (BB) rats compared with nondiabetic controls, whereas rAGT mRNA expression is significantly higher in RPTs of diabetic BB rats. 3 These studies indicate that hnRNP K is an endogenous suppressor and may play an important role in counterbalancing AGT gene overexpression in high glucose in vivo. Studies are ongoing in our laboratory along this line.
In summary, we have established, by a combination of Southwestern blotting and proteomics, that hnRNP K is a nuclear protein that binds to rAGT-IRE and inhibits rAGT gene expression in IRPTCs. It appears that high glucose and insulin regulate hnRNP K expression in kidney proximal tubular cells. Our studies raise the possibility that hnRNP K expression may play a role in counterbalancing high glucose stimulation of rAGT gene expression and in modulating local intrarenal RAS activation. Dysregulation of hnRNP K expression may contribute to renal injury in diabetes via altered local intrarenal RAS activation.