Cullin 3 mutant causing familial hyperkalemic hypertension lacks normal activity in the kidney

Abstract

Mutations in the ubiquitin ligase scaffold protein cullin 3 (CUL3) cause the disease familial hyperkalemic hypertension (FHHt). We recently reported that in the kidney, aberrant mutant CUL3 (CUL3-Δ9) activity lowers the abundance of CUL3-Δ9 and Kelch-like 3, the CUL3 substrate adaptor for with-no-lysine kinase 4 (WNK4) and that this is mechanistically important. However, whether CUL3-Δ9 exerts additional effects on other targets that may alter renal function is unclear. Here, we sought to determine 1) whether CUL3-Δ9 expression can rescue the phenotype of renal tubule-specific Cul3 knockout mice, and 2) whether CUL3-Δ9 expression affects other CUL3 substrates. Using an inducible renal tubule-specific system, we studied two CUL3-Δ9-expressing mouse models: Cul3 knockout (Cul3^–/–/Δ9) and Cul3 heterozygous background (Cul3^+/–/Δ9, FHHt model). The effects of CUL3-Δ9 in these mice were compared with Cul3^–/– and Cul3^+/– mice. Similar to Cul3^–/– mice, Cul3^–/–/Δ9 mice displayed polyuria with loss of aquaporin 2 and collecting duct injury; proximal tubule injury also occurred. CUL3-Δ9 did not promote degradation of two CUL3 targets that accumulate in the Cul3^–/– kidney: high-molecular-weight (HMW) cyclin E and NAD(P)H:quinone oxidoreductase 1 (NQO1) [a surrogate for the CUL3-Kelch-like ECH-associated protein 1 (KEAP1) substrate nuclear factor erythroid-2-related factor 2]. Since CUL3-Δ9 expression cannot rescue the Cul3^–/– phenotype, our data suggest that CUL3-Δ9 cannot normally function in ubiquitin ligase complexes. In Cul3^+/–/Δ9 mice, KEAP1 abundance did not differ but NQO1 abundance was higher, suggesting adaptor sequestration by CUL3-Δ9 in vivo. Together, our results provide evidence that in the kidney, CUL3-Δ9 completely lacks normal activity and can trap CUL3 substrate adaptors in inactive complexes.

NEW & NOTEWORTHY CUL3 mutation (CUL3-Δ9) causes familial hyperkalemic hypertension (FHHt) by reducing adaptor KLHL3, impairing substrate WNK4 degradation. Whether CUL3-Δ9 affects other targets in kidneys remains unclear. We found that CUL3-Δ9 cannot degrade two CUL3 targets, cyclin E and nuclear factor erythroid-2-related factor 2 (NRF2; using a surrogate marker NQO1), or rescue injury or polyuria caused by Cul3 disruption. In an FHHt model, CUL3-Δ9 impaired NRF2 degradation without reduction of its adaptor KEAP1. Our data provide additional insights into CUL3-Δ9 function in the kidney.

INTRODUCTION

Cullin 3 (CUL3) is a central component of a wide array of ubiquitin ligase complexes. It serves as a scaffold that interacts with over 120 different substrate adaptor proteins (1) that determine the specific target of ubiquitination and a RING ubiquitin ligase that links the ubiquitin moiety to the substrate. Ubiquitination of proteins by CUL3-containing ligase complexes (CRL3s) plays an important role in many important processes, including regulation of the cell cycle, protein trafficking, development, and oxidative stress responses. Consistent with its critical role in fundamental processes, global Cul3 knockout is embryonic lethal (2). Aberrant CRL3 activity results in a wide variety of pathological states including skeletal myopathies, neurological disorders, metabolic disorders, and cancers.

Mutations in CUL3 cause the most severe form of the disease familial hyperkalemic hypertension (FHHt’ also known as Gordon syndrome or pseudohypoaldosteronism type II). The majority of FHHt-causing mutations identified to date lead to missplicing of exon 9, resulting in the generation of a form of CUL3 with a 57-amino acid deletion (Δ403–459, referred to here as CUL3-Δ9). A novel mutation within the coding region of exon 10 (Δ474–477, CUL3^Δ474–477) was recently identified in a single patient with FHHt (3). CUL3 normally undergoes cycling of covalent addition of neural precursor cell expressed, developmentally downregulated 8 (NEDD8), catalyzed by NEDD8-activating enzyme (NAE) and removal of NEDD8, catalyzed by c-Jun activation domain-binding protein-1 (JAB1), as part of the constitutive photomorphogenesis 9 (COP9) signalosome complex. This cycling of NEDD8 addition/removal (“neddylation”/“deneddylation”) is critical for the stability and activity of CRL3. Both FHHt-causing CUL3 mutants have an impaired ability to interact with JAB1 leading to “hyperneddylation,” which results in dysfunctional CRL3.

FHHt initially presents with hyperkalemia with a later onset of hypertension. Hyperactivation of the NaCl cotransporter (NCC), expressed along the renal distal convoluted tubule (DCT), plays a central pathophysiological role (4–7), with effects on the vasculature exacerbating hypertension (8, 9). In the kidney, CUL3-Δ9-containing CRL3 indirectly hyperactivates NCC by impairing degradation of the CRL3 target with-no-lysine 4 (WNK4). WNK4 accumulates, leading to increased activation of downstream STE20/SPS1-related proline-alanine-rich kinase (SPAK), which then phosphorylates its substrate NCC in an uncontrolled manner (4, 6, 7). We recently reported that in the kidney, aberrant CUL3-Δ9 activity and acute JAB1 disruption cause a reduced abundance of CUL3-Δ9 and Kelch-like (KLHL)3, the CRL3 substrate adaptor for WNK4, and that this is a central mechanism in CUL3-Δ9-mediated FHHt (5).

Questions remain, however, regarding the function of CUL3-Δ9 in vivo. CUL3-Δ9-mediated FHHt is an autosomal dominant form of the disease. Global constitutive Cul3 knockout mice and mice homozygous for CUL3-Δ9 are nonviable (6, 7), suggesting an essential developmental function of wild-type (WT) CUL3 that CUL3-Δ9 cannot substitute for. In renal epithelia, inducible Cul3 disruption leads to polyuria through an effect on aquaporin 2 (AQP2) and renal fibrosis by inducing proximal tubule injury (10, 11). Although the study of FHHt has demonstrated that CUL3-Δ9 expression disrupts its effects on KHLH3-mediated pathways, CUL3-Δ9 might form normally functioning CRL3s with other substrate adaptors. In contrast to the effects on KLHL3 (5, 7), Uchida and colleagues (7) reported that several other CUL3 substrate adaptors including Kelch-like ECH-associated protein 1 (KEAP1), KLHL2, and KLHL16, were not affected in mice heterozygous for CUL3-Δ9. However, we and others have shown that CUL3-Δ9 has increased affinity for several substrate adaptors (3, 6, 10, 12) and an impaired ability to interact with WT CUL3 (12). These observations raise the possibility that CUL3-Δ9 exerts additional effects on other CRL3 targets that may alter renal function. To gain a better understanding of CUL3-Δ9 activity in the kidney, we, therefore, used several mouse models to determine 1) whether CUL3-Δ9 expression could rescue the phenotype of renal tubule-specific Cul3 knockout mice and 2) whether CUL3-Δ9 expression affects other CRL3 substrates.

MATERIALS AND METHODS

Animals

Animal experiments were approved by the Oregon Health and Science University Institutional Animal Care and Use Committee (protocol IP00286). For all mouse strains, the doxycycline-inducible tubule-specific Pax8-rtTA/TRE-LC1 system was used (4, 5, 10, 13). CUL3-Δ9 was expressed from the Loxp-STOP-Loxp-Cul3-Δ9-IRES-tdTomato transgene (9). Cul3^wt/flox/CUL3-Δ9 (CUL3-Het/Δ9) mice were generated as previously described (4). Cul3 knockout mice (Cul3^−/−), littermates expressing CUL3-Δ9 on a Cul3^−/− background (Cul3^−/−/Δ9), Cul3 heterozygous mice (Cul3^+/−), Cul3 heterozygous mice also expressing CUL3-Δ9 (Cul3^+/−/Δ9), and Cul3 and Klhl3 heterozygous mice (Cul3^+/−/Klhl3^+/− mice) were generated as previously described (4, 5, 10). To induce recombination at floxed sites, 2 mg/mL doxycycline (Thermo Fisher Scientific, Waltham, MA) in 5% sucrose in drinking water was administered for 2 or 3 wk. After doxycycline treatment, mice were returned to regular drinking water for at least 2 wk before experiments were performed. Control mice given 5% sucrose drinking water were phenotypically equivalent to WT mice and were littermates of those that received doxycycline. Both male and female mice were used as indicated in figures, and all mice were on the mixed C57Bl/6J background.

PCR Genotyping

Genomic DNA extracts were prepared from tail snips by heating overnight at 55°C in 300 μL of digestion solution containing 5 mM EDTA, 200 mM NaCl, 100 mM Tris (pH 8.0), 0.2% SDS, and 0.4 mg/mL proteinase K, followed by ethanol precipitation. The following primers were used: Pax8, forward 5′- CCATGTCTAGACTGGACAAGA-3′ and reverse 5′- CAGAAAGTCTTGCCATGACT-3′; Cre recombinase (Cre), forward 5′- TTTCCCGCAGAACCTGAACCTGAAGAT-3′ and reverse 5′- TCACCGGCATCAACGTTTTCTT-3′; Cul3flox, forward 5′- CAGGTTGTATTTTAACTGCTTAAATGTCAAAACCT-3′ and reverse 5′- TTTGTCTGGACCAAATATGGCAGCCAAAACC-3′; and Cul3-Δ9, 5′- GGCGCGATTCTTACCAAGTCC-3′ and reverse 5′-GCGCATGAACTCTTTGATGACTT-3′. PCR genotyping for Klhl3^flox and confirmation of Flp recombinase was performed by quantitative PCR by Transnetyx (Cordova, TN).

Antibodies

Antibody sources, species, dilutions, and validation references are provided in Supplemental Table S1.

Western Blot Analysis

Harvested kidneys were cut in half, snap frozen in liquid nitrogen, and then stored at −80°C. A half kidney was homogenized using a Potter homogenizer in 1 mL of cold buffer containing 300 mM sucrose, 50 mM Tris·HCl (pH 7.4), 1 mM EDTA, 1 mM EGTA, 1 mM NaVO₄, 50 mM NaF, 1 mM dithiothreitol, 1 mM PMSF, 1 µg/mL aprotinin, 4 µg/mL leupeptin, and phosphatase inhibitor (PhosStop, Roche, Mannheim, Germany). Homogenates were centrifuged at 6,000 rpm for 15 min at 4°C, and supernatants were transferred to a new tube and then stored at −80°C. Protein loading was adjusted by densitometric quantitation of total protein after Coomassie staining (see Supplemental Fig. S1; 4, 5, 14, 15). Protein samples were separated by electrophoresis on 4–12% Criterion XT bis-Tris gels and 4–15% Criterion Tris-Glycine eXtended stain-free gels (Bio-Rad Laboratories, Hercules, CA) and transferred to low fluorescence PVDF membranes using the Trans-Blot Turbo transfer system (Bio-Rad Laboratories). Membranes were blocked with 5% nonfat milk in PBS-Tween (Thermo Fisher Scientific), followed by incubation with primary antibody overnight at 4°C. Anti-KLHL3 antibody was diluted in Can Get Signal (TOYOBO, Osaka, Japan). Appropriate horseradish peroxidase-conjugated secondary antibody in blocking buffer was added to membranes for 1 h at room temperature. Membranes were developed using enhanced chemiluminescence and Western Lightning Plus-ECL (Perkin-Elmer, Waltham, MA) and visualized using Pxi digital imaging system (Syngene, Frederick, MA). Densitometry was performed with ImageJ (http://rsbweb.nih.gov/ij/).

Immunofluorescence and Histology

For immunolocalization, three mice per group were analyzed to confirm findings; representative images are shown. Mice were anesthetized with a ketamine-xylazine-acepromazine cocktail, and kidneys were removed and placed in 10% buffered formalin phosphate (Fisher Scientific, Fair Lawn, NJ) for 24 h. The kidneys were then transferred to a 70% ethanol solution for at least 24 h. The Oregon Health and Science University Histopathology Core Facility embedded the kidney in paraffin and prepared sections. Sections were used after deparaffinization and rehydration. Antigen retrieval was performed by microwave heating in antigen unmasking solution (Vector Laboratories, Burlingame, CA) or by incubating for 1 h at ∼100°C. Sections were then blocked with 5% normal goat serum and 5% donkey serum in PBS for 30 min, followed by incubation with primary antibody, diluted in 5% BSA, overnight at 4°C. Goat anti-rabbit Fab fragments (1:50, No. 111-007-003, Jackson ImmunoResearch, West Grove, PA) were used for coimmunofluorescence with multiple antibodies raised in the rabbit. Sections were incubated with fluorescent dye-conjugated secondary antibody, diluted in blocking buffer, for 1 h at room temperature before being mounted with Prolong Diamond Antifade Mountant (Invitrogen, Carlsbad, CA). Fluorescence images were acquired using a Keyence BZ-X810 fluorescence microscope (Osaka, Japan). For histology, sections were stained with hematoxylin-eosin by the Oregon Health and Science University Histopathology Core.

Blood Analysis and Urinalysis

Blood was collected via cardiac puncture under isoflurane anesthesia and transferred into heparinized tubes; 95 μL were loaded into a Chem8+ or CG8+ cartridge for electrolyte measurement by an i-STAT analyzer (Abbot Point of Care, Princeton, NJ). The remainder was centrifuged at 2,000 g for 5 min at room temperature, and plasma was removed and stored at −80°C. Plasma phosphate concentration was determined by a Malachite Green Phosphate Assay Kit (Cayman Chemical, Ann Arbor, MI). Mice were acclimated to metabolic cages for 2 days before urine collection. Urine was collected under water-saturated light mineral oil after 24 h, and water consumption was measured by weighing containers.

Statistics

The null hypothesis was tested using two-tailed unpaired t tests and one-way ANOVA using GraphPad Prism 9 as indicated in the figures. Post hoc analysis was performed using unpaired t tests with the Bonferroni correction. All data are plotted as means ± SE. Significant P values are indicated as *P < 0.05, **P < 0.01, and ***P < 0.001 in the figures.

RESULTS

CUL3-Δ9 Cannot Prevent Polyuria in Renal Tubule-Specific Cul3 Knockout Mice

Induced renal tubule-specific CUL3 deletion in adult mice causes polyuria associated with loss of glycosylated AQP2 (10, 11). We first determined whether CUL3-Δ9 expression could rescue this phenotype by expressing it on the CUL3 knockout background (Fig. 1A). To express CUL3-Δ9, we used a transgenic line containing the CUL3-Δ9 cDNA downstream from a floxed transcriptional blocker (9). The transgene also drives expression of the fluorescent reporter tdTomato from the same mRNA via an internal ribosomal entry site. Mice carrying two floxed Cul3 alleles, either without the CUL3-Δ9 transgene (Cul3^−/−) or with it (Cul3^−/−/Δ9), were generated. In both lines, the Pax8-rtTA-LC1 Cre system permits recombination at the floxed sites in response to doxycycline administration. As previously shown (5), compared with uninduced controls, CUL3 abundance was significantly lower in both Cul3^−/− and Cul3^−/−/Δ9 mice, and expression of CUL3-Δ9 and tdTomato protein was detected in Cul3^−/−/Δ9 mice (Fig. 1B). Western blot analysis showed that glycosylated AQP2 was almost completely absent in Cul3^−/− mice compared with control mice, but abundances did not differ between Cul3^−/− and Cul3^−/−/Δ9 mice (Fig. 1C). Immunofluorescence confirmed that the AQP2 signal was reduced in both Cul3^−/− and Cul3^−/−/Δ9 mice (Fig. 1D). Consistent with this, CUL3 disruption caused polyuria and increased water intake that were not prevented by CUL3-Δ9 expression (Fig. 1E). Body weight was lower, and plasma Na⁺ concentration and Cl⁻ concentration were higher in Cul3^−/− and Cul3^−/−/Δ9 mice compared with control mice but did not differ between Cul3^−/− and Cul3^−/−/Δ9 (Fig. 1, F–H). No obvious sex differences were observed for water balance, plasma Na⁺ concentration, or plasma Cl⁻ concentration, but there was a sex difference for body weight (P < 0.05, control males vs. control females and Cul3^−/−/Δ9 males vs. Cul3^−/−/Δ9 females; trending P = 0.11, Cul3^−/− males vs. Cul3^−/− females).

Figure 1.

CUL3-Δ9 expression cannot rescue polyuria caused by cullin 3 (Cul3) deletion. A: in Cul3^−/− and Cul3^−/−/Δ9 mice, predicted wild-type (WT) CUL3 abundance was reduced to ∼0%, but CUL3 was only detected in Cul3^−/−/Δ9 mice. Control mice were Cul3^fl/fl or Cul3^fl/fl/Δ9 mice carrying Pax8rt-TA/LC1-Cre transgenes administered vehicle (5% sucrose in drinking water) only. B: Western blot analysis showed that both doxycycline-induced Cul3^−/− and Cul3^−/−/Δ9 mice displayed significantly lower CUL3 abundance than controls. CUL3 expression did not significantly differ between Cul3^−/− and Cul3^−/−/Δ9 mice. CUL3-Δ9 and tdTomato expression were only observed in Cul3^−/−/Δ9 mice. C: Western blot analysis revealed that glycosylated (glyc) aquaporin 2 (AQP2) abundance was significantly lower in the kidney in both Cul3^−/− mice and Cul3^−/−/Δ9 mice. D: immunofluorescence staining showed that the AQP2 signal was reduced in the renal cortex and medulla of Cul3^−/− and Cul3^−/−/Δ9 mice. Cul3^−/−/Δ9 [male (M)], littermate Cul3^–/– [female (F)], and control mice (M) were used for representative images. Scale bars = 50 µm. E: metabolic cage analysis showed higher urine volume (left) and water intake (right) in Cul3^−/− and Cul3^−/−/Δ9 mice. F–H: body weight (BW) was significantly lower and plasma Na⁺ concentration and plasma Cl⁻ concentration were significantly higher in Cul3^−/− and Cul3^−/−/Δ9 mice compared with controls but did not differ from each other. Individual values and means ± SE are shown; values in parentheses are numbers of mice (n). Statistical differences were examined by one-way ANOVA, followed by post hoc unpaired t tests with the Bonferroni correction (B, C, E, and F). *P < 0.05; **P < 0.01; ***P < 0.001. NS, P > 0.05. NEDD8, neural precursor cell expressed, developmentally downregulated 8. NS, not significant.

CUL3-Δ9 Cannot Rescue Proximal and Collecting Duct Injury in Cul3 Knockout Mice

Inducible renal tubule-specific CUL3 disruption results in sustained proximal tubule injury associated with cell cycle dysregulation (10, 11). To determine whether CUL3-Δ9 displays activities that can prevent renal injury, we again used Cul3^−/− and Cul3^−/−/Δ9 mice. We performed Western blot analysis and immunofluorescence for kidney injury molecule-1 (KIM-1), Lotus tetragonolobus lectin (LTL), neutrophil gelatinase-associated lipocalin (NGAL), and Ki-67, markers of proximal tubule injury, distal tubule injury, and proliferation, respectively. KIM-1 expression was not detected in control mice and was markedly higher in both Cul3^−/− and Cul3^−/−/Δ9 mice (Fig. 2A and Supplemental Fig. S2A). Immunofluorescence showed loss of LTL, a marker of the proximal tubule brush border (Fig. 2B), which colocalized with KIM-1 in both Cul3^−/− and Cul3^−/−/Δ9 mice, indicating proximal tubule injury (Fig. 2B). Similarly, NGAL expression was higher in both Cul3^−/− and Cul3^−/−/Δ9 mice compared with controls (Fig. 2A and Supplemental Fig. S2B). Circulating NGAL filtered by the glomerulus is reabsorbed along the proximal tubule, whereas during injury NGAL expression is induced along the distal tubules (16, 17). In both Cul3^−/− and Cul3^−/−/Δ9 mice, immunofluorescence showed no NGAL upregulation along the thick ascending limb (Na⁺-K⁺-2Cl⁻ cotransporter-positive tubules) and DCT/connecting tubule (CNT; calbindin-positive tubules; Supplemental Fig. S3, B and C) but strong upregulation along both cortical and medullary collecting ducts in AQP2-positive cells (Fig. 2, C and D, and Supplemental Fig. S3D). Consistent with an earlier study examining NGAL localization in the proximal tubule (16), NGAL vesicles were observed in the proximal tubule, particularly along the apical surface, in Cul3^−/− and Cul3^−/−/Δ9 mice (Supplemental Fig. S3, A and C). A greater number of Ki-67-positive proliferating cells was observed along proximal and distal tubules in both Cul3^−/− and Cul3^−/−/Δ9 mice (Supplemental Fig. S4). We previously reported progression of injury to severe kidney damage 8 wk after the induction of CUL3 disruption, with only focal regions of damage observed after 4 wk (11). Similarly, both Cul3^−/− and Cul3^−/−/Δ9 mice displayed focal regions of injury 5 wk after the initiation of doxycycline administration, which progressed to extensive damage by 8 wk (Supplemental Fig. S5). Plasma phosphate was significantly higher in Cul3^−/− mice but did not differ between Cul3^−/− and Cul3^−/−/Δ9 mice (Supplemental Fig. S6A), and plasma [K⁺] trended higher in both Cul3^−/− and Cul3^−/−/Δ9 mice (Supplemental Fig. S6B), reflecting diffuse tubular injury, although blood urea nitrogen did not differ at this time point (Supplemental Fig. S6C). The difference between plasma [Na⁺] and [Cl⁻] was significantly lower, and the total CO₂ level was not significantly but lower in Cul3^−/− and Cul3^−/−/Δ9 mice, indicating mild metabolic acidosis (Supplemental Fig. S6, D and E). Together, these results suggest no significant effect of CUL3-Δ9 on the Cul3^−/− injury phenotype.

Figure 2.

CUL3-Δ9 cannot rescue proximal tubule and distal tubule injury caused by cullin 3 (Cul3) deletion. A: Western blot analysis of the whole kidney lysate revealed that abundances of kidney injury molecule-1 (KIM-1) and neutrophil gelatinase-associated lipocalin (NGAL), markers of proximal tubule injury and distal tubule injury, respectively, were higher in both Cul3^−/− and Cul3^−/−/Δ9 mice. The white arrowhead indicates a nonspecific band. B: immunofluorescence showed KIM-1-positive proximal tubules with loss of the Lotus tetragonolobus lectin (LTL) signal (white) in both Cul3^−/− and Cul3^−/−/Δ9 mice. C and D: immunofluorescence showed NGAL-positive proximal tubules and distal tubules (Supplemental Fig. S2), with a high signal in cortical (Supplemental Fig. S2D) and medullary collecting duct in cells with a low aquaporin-2 (AQP2) signal (D, green) in both Cul3^−/− and Cul3^−/−/Δ9 mice. For A, individual values and means ± SE are shown; values in parentheses are numbers of mice (n). Statistical differences were examined by one-way ANOVA, followed by post hoc unpaired t tests with the Bonferroni correction. **P < 0.01; ***P < 0.001. NS, P > 0.05. For B and D, scale bars = 50 µm; representative images are shown of 2 mice/group. Cul3^−/−/Δ9 [male (M)], littermate Cul3^–/– [female (F)], and control mice (M) were used for representative images. NS, not significant.

CUL3-Δ9 Has No Activity Toward CRL3 Substrates in Cul3^−/− Mice

Examination of several other CRL3 adaptors revealed that only KLHL3 abundance is reduced by CUL3-Δ9 expression in vivo (7). However, in vitro studies have shown that can CUL3-Δ9 bind to substrate adaptor proteins with higher affinity than WT CUL3, resulting in varied effects on target protein degradation (10, 12). CUL3-Δ9 might therefore affect substrate degradation through effects on CRL3 function, without effects on adaptor abundance, in vivo. Therefore, we examined the effects of CUL3-Δ9 on two other CUL3 substrates. Nuclear factor erythroid-2-related factor 2 (NRF2) is a substrate for the KEAP1-CUL3 CRL3 complex (Supplemental Fig. S7A; 18). Cyclin E is a target that binds directly to CUL3 without a substrate adaptor (2, 19). The NH₂-terminus of cyclin E directly interacts with CUL3, resulting in preferential degradation of high-molecular-weight (HMW) cyclin E over low-molecular-weight (LMW) cyclin E, which has an NH₂-terminal truncation (Supplemental Fig. S7B). Consistent with a CUL3 regulatory effect on NFR2 and cyclin E, Cul3^−/− mice display higher abundance of NAD(P)H:quinone oxidoreductase 1 (NQO1), a surrogate marker of NRF2 activity, and HMW cyclin E (11). KEAP1 abundance did not differ between control, Cul3^−/−, or Cul3^−/−/Δ9 mice, but NQO1 was greatly increased in both Cul3^−/− and Cul3^−/−/Δ9 mice (Fig. 3A). Similarly, HMW cyclin E abundance was higher and LMW cyclin E abundance was lower in Cul3^−/− and Cul3^−/−/Δ9 mice (Fig. 4A), consistent with our previous results in Cul3^−/− mice (see Fig. 8D in Ref. 10). Immunofluorescence showed that NQO1 was mainly expressed along the proximal tubule (LTL-positive tubules) in control mice, but NQO1 was increased mainly in distal tubules, including the DCT/CNT (calbindin-positive tubules) and collecting duct (AQP2-positive tubules), in both Cul3^−/− and Cul3^−/−/Δ9 mice (Fig. 3, B–E). Greater numbers of nuclear cyclin E-positive cells were observed in both Cul3^−/− and Cul3^−/−/Δ9 mice along the proximal tubule, DCT/CNT, and collecting duct, particularly in the medulla (Fig. 4, B–E). Both a time-course experiment and shorter-term (4 days) doxycycline treatment confirmed that CUL3 deletion and transgene induction were concurrent, as previously shown (5), and showed similar but less striking effects on the abundance of NQO1 (Supplemental Figs. S8 and S9); no significant effects on LMW or HMW cyclin E were observed with shorter-term CUL3 deletion, consistent with our previous data (see Fig. 2, A and B in Ref. 11). These data suggest that CUL3-Δ9 alone cannot form normally functioning CRL3s since NQO1 and HMW cyclin E abundances in Cul3^−/−/Δ9 mice are similar to those observed in Cul3^−/− mice.

Figure 3.

CUL3-Δ9 has no activity toward the cullin 3 (CUL3)-Kelch-like ECH-associated protein 1 (KEAP1) substrate nuclear factor erythroid-2-related factor 2 (NRF2) when wild-type CUL3 is deficient. A: Western blot analysis revealed that the abundance of the CUL3 substrate adaptor KEAP1 was similar in control, Cul3^−/−, and Cul3^−/−/Δ9 mice, but the abundance of NAD(P)H:quinone oxidoreductase 1 (NQO1), a surrogate marker for activity of the CUL3-KEAP1 substrate NRF2, was significantly higher in both Cul3^−/− and Cul3^−/−/Δ9 mice. B–E: immunofluorescence staining showed that NQO1 (red) was mainly expressed along the proximal tubule [Lotus tetragonolobus lectin (LTL); C, white] in control mice, but in both Cul3^−/− and Cul3^−/−/Δ9 mice NQO1 was also expressed at high levels along distal segments, including the late distal convoluted tubule/connecting tubule [calbindin (CB); D, green] and collecting duct [aquaporin-2 (AQP2); E, green). Cul3^−/−/Δ9 [male (M)], littermate Cul3^–/– (female), and control mice (M) were used for representative images. Scale bars = 50 µm. For A, individual values and means ± SE are shown; values in parentheses are numbers of mice (n). Statistical differences were examined by one-way ANOVA, followed by post hoc unpaired t tests with the Bonferroni correction. ***P < 0.001. NS, P > 0.05. NS, not significant.

Figure 4.

CUL3-Δ9 has no activity toward cyclin E when wild-type cullin 3 (CUL3) is deficient. A: Western blot analysis revealed higher abundance of high-molecular-weight (HMW) cyclin E, which is directly bound by CUL3, and lower abundance of low-molecular-weight (LMW) cyclin, which cannot be bound by CUL3, in both Cul3^−/− and Cul3^−/−/Δ9 mice. B–E: immunofluorescence showed more cyclin E-positive cells (red) along the proximal tubule [(Lotus tetragonolobus lectin (LTL); C, white] and late distal convoluted tubule (DCT2)/connecting tubule [calbindin (CB); D, green] and collecting duct [aquaporin-2 (AQP2); E, green] in both Cul3^−/− and Cul3^−/−/Δ9 mice. Cul3^−/−/Δ9 [male (M)], littermate Cul3^–/– [female (F)], and control mice (M) were used for representative images. Scale bars = 50 µm. For A, individual values and means ± SE are shown; values in parentheses are numbers of mice (n). Statistical differences were examined by one-way ANOVA, followed by post hoc unpaired t tests with the Bonferroni correction. ***P < 0.001. NS, P > 0.05. NS, not significant.

CUL3-Δ9 Impairs CUL3-KEAP1 CRL3 Activity Toward NRF2 but Not Cyclin E

Since CUL3-Δ9-mediated FHHt is an autosomal dominant form of the disease, we next evaluated the effect of CUL3-Δ9 on KEAP1, NQO1, and cyclin E in Cul3^+/−/Δ9 mice, which phenocopy the disease (4). Since CUL3-Δ9 induces its own degradation in the kidney, resulting in lower total CUL3 expression (to ∼50% of WT levels; 4, 6), we also used Cul3^+/− mice to determine specific effects of CUL3-Δ9 expression (Fig. 5, A–D). KEAP1 abundance did not differ compared with controls in Cul3^+/− or Cul3^+/−/Δ9 mice (Fig. 5, E and F). The abundance of NQO1 was significantly higher in both Cul3^+/− and Cul3^+/−/Δ9 mice compared with controls, but to a greater degree in Cul3^+/−/Δ9 mice (15% higher vs. 27% when directly compared; see Fig. 5G). In contrast, HMW cyclin E and LMW cyclin E abundances did not differ in Cul3^+/−/Δ9 and Cul3^+/− mice compared with control mice (Supplemental Fig. S10). We determined whether the effect on NQO1 was secondary to the FHHt phenotype or resulted from a specific effect of CUL3-Δ9 using Cul3^+/−/Klhl3^+/− mice that display an FHHt-like phenotype in the absence of CUL3-Δ9 expression (5). The abundances of NQO1 and cyclin E were similar in control and Cul3^+/−/Klhl3^+/− mice (Supplemental Fig. S11). Finally, the abundances of glycosylated AQP2, NGAL, and KIM-1 were similar in controls and Cul3^+/−/Δ9 mice (Supplemental Fig. S12). Together, these data suggest that CUL3-Δ9 impairs CUL3-KEAP1 CRL3 activity toward NRF2 in a mouse model of FHHt.

Figure 5.

CUL3-Δ9 impairs cullin 3 (CUL3)-containing ligase complex (CRL3)-Kelch-like ECH-associated protein 1 (KEAP1) activity toward nuclear factor erythroid-2-related factor 2 (NRF2) in a mouse model of familial hyperkalemic hypertension (FHHt). A–D: in Cul3^+/− and Cul3^+/−/Δ9 mice [a model of FHHt (4)], the predicted wild-type (WT) CUL3 abundance was reduced to ∼50%, but CUL3-Δ9 was only detected in Cul3^+/−/Δ9 mice. Control mice were genetically identical to either Cul3^+/− or Cul3^+/−/Δ9 mice and were administered vehicle (5% sucrose in drinking water) only. E–G: Western blot analysis revealed that the abundance of KEAP1 was similar in control, Cul3^+/−, and Cul3^+/−/Δ9 mice. The abundance of NAD(P)H:quinone oxidoreductase 1 (NQO1) was significantly higher in Cul3^+/− and Cul3^+/−/Δ9 mice than in controls, but the difference was higher in Cul3^+/−/Δ9 mice when they were directly compared (G). Individual values and means ± SE are shown; numbers in parentheses are numbers of mice (n). Statistical differences were examined by two-tailed unpaired t tests (B–G). **P < 0.01; ***P < 0.001. NS, P > 0.05. F, female; M, male; NEDD8, neural precursor cell expressed, developmentally downregulated 8. NS, not significant.

DISCUSSION

In vitro studies have identified functional defects in CRL3s containing CUL3-Δ9 including auto-ubiquitination (6), altered affinity for substrate adaptors that sequester them in inactive CRL3s (3, 6, 10, 12, 20), anomalous degradation of KLHL3 (10, 21), and reduced interaction with WT CUL3 (12). The observations that CUL3-Δ9 binds substrate adaptors more tightly than WT CUL3 and that neddylation of WT CUL3 is reduced in the presence of CUL3-Δ9 suggest a strong dominant function of CUL3-Δ9 (12). Here, using several mouse models, we provide additional insight into CUL3-Δ9 function in the kidney. Using Cul3^−/−/Δ9 mice, we determined whether the dominant activities of CUL3-Δ9 were sufficient for normal CRL3 function. Expression of CUL3-Δ9 on the Cul3^−/− background did not prevent the development of polyuria and kidney injury observed in Cul3^−/− mice (10, 11). At the level of specific CUL3 targets, CUL3-Δ9 expression did not prevent the accumulation of NQO1 (a surrogate for NRF2 accumulation and hence activity) or HMW cyclin E. Since CUL3-Δ9 levels were extremely low, presumably due to auto-ubiquitination, and those of CUL3-Δ9-containing CRL3s remaining are functionally abnormal, the phenotype of Cul3^−/−/Δ9 mice is similar to that of Cul3^−/− mice. These data are consistent with previous observations showing that Cul3^Δex9/Δex9 knockin mice are nonviable (6, 7), but confirm them in the adult mouse kidney.

Our data provide further insight into renal injury, resulting from CUL3 disruption (11). NGAL was highly upregulated in both principal cells and intercalated cells in both Cul3^−/− and Cul3^−/−/Δ9 mice. This differs to what has been reported for the ischemia-reperfusion injury model, in which Ngal mRNA is profoundly upregulated in thick ascending limbs and intercalated cells (17). This suggests a specific role for CUL3 in normal collecting duct function. This raises the possibility that the loss of AQP2 leading to polyuria is secondary to injury, rather than a more direct effect of CUL3 on AQP2 expression. Indeed, in individual principal cells, immunofluorescence indicated that cells with lower NGAL expression displayed higher AQP2 expression (Supplemental Fig. S3). Further studies, such as time-course analysis and in vitro experiments exploring whether CUL3 directly affects AQP2 trafficking and expression, are needed to address this issue. NQO1 was also highly upregulated in principal cells, indicating NRF2 hyperactivation following CUL3 disruption. NRF2 hyperactivation might itself induce principal cell injury (Fig. 6), although data conflict regarding the renoprotective versus renoinjurious effects of NRF2 (22–24). Importantly in this regard, similar phenotypes of severe nephrogenic diabetes insipidus with a strong downregulation of AQP2 are observed in a nonlethal model of global genetic Keap1 knockout and in constitutive kidney-specific Keap1-deficient mice (25, 26). In these models, NRF2 is completely derepressed, but the mechanism by which AQP2 is downregulated is unknown; whether there is principal cell injury as shown by increased NGAL expression was not determined. Finally, dysregulation of the cell cycle due to reduced CUL3-mediated cyclin E regulation may also contribute to collecting duct injury. We have previously shown by Western blot analysis that increased cyclin E abundance precedes injury in Cul3^−/− mice (11); here, immunofluorescence revealed a higher number of cyclin E-positive cells along several segments including the collecting duct.

Figure 6.

Summary of CUL3-Δ9 function in the kidney. Left: in cullin 3 (Cul3) knockout (Cul3^−/−) mice, CUL3-containing ligase complex (CRL3) substrates, including with-no-lysine kinase 4 (WNK4; 5), nuclear factor erythroid-2-related factor 2 (NRF2) [NAD(P)H:quinone oxidoreductase 1 (NQO1) was used as a surrogate marker], and high-molecular-weight (HMW) cyclin E, are dramatically increased. These mice exhibit kidney injury along both the proximal tubule and distal tubule, leading to polyuria with aquaporin-2 loss. Middle: in Cul3 knockout mice expressing CUL3-Δ9 (Cul3^−/−/Δ9), CUL3-Δ9 alone can degrade the distal convoluted tubule-specific adaptor Kelch-like 3 (KLHL3; 5) but cannot degrade CRL3 substrates and rescue the phenotype, suggesting an absence of normal activity. Right: in Cul3 heterozygous mice expressing CUL3-Δ9 (Cul3^+/−/Δ9), a familial hyperkalemic hypertension (FHHt) model, WNK4 is increased by abnormal degradation of KLHL3 (5), whereas NQO1 is modestly increased without a difference in KEAP1 abundance, suggesting a sequestration effect of CUL3-Δ9. HMW cyclin E abundance is not increased, consistent with a lack of tubule injury and aquaporin-2 loss.

Recent data suggest that a low level of NRF2 activation can promote urine concentration. Graded NRF2 activation in Keap1 hypomorphic mice with only 20% of WT KEAP1 abundance levels (27) can concentrate their urine after water deprivation and are protected against lithium-induced nephrogenic diabetes insipidus (27). However, the mechanism was proposed to be AQP2 independent since NRF2 activation did not alter the abundance of glycosylated AQP2. Our data from Cul3^−/− and Cul3^−/−/Δ9 mice are similar to the effect of Keap1 disruption and suggest that CUL3-Δ9 cannot form normally functioning CRL3s that modulate NRF2 activity in principal cells. However, the more modest effect on NRF2 activity in Cul3^+/−/Δ9 raises the possibility that CUL3-Δ9 expression may protect against lithium-induced nephrogenic diabetes insipidus.

In addition, the modest NRF2 activation observed may influence the FHHt phenotype. A recent study showed that activation of NRF2 either pharmacologically with bardoxolone methyl or in Keap1 hypomorphic mice reduced phosphorylated NCC abundance (27). This might mitigate the effects of increased WNK4-SPAK on NCC hyperactivation. On the other hand, expression of the cyclooxygenase (COX) enzymes COX-1 and COX-2 was also lower with NRF2 activation in these models. Prostaglandins maintain renal blood flow and glomerular filtration rate and stimulate renin secretion, leading to aldosterone synthesis and K⁺ excretion. Therefore, inhibition of COX enzymes might worsen hyperkalemia and metabolic acidosis, as occurs with nonsteroidal anti-inflammatory drugs (28). Indeed, patients with FHHt with CUL3 mutations have the most severe hyperkalemia and metabolic acidosis and a trend to lower plasma renin levels compared with patients with FHHt carrying other mutations (29, 30).

Finally, there are conflicting data regarding the renoprotective effects of NRF2 in kidney injury models (22–24). Increased NRF2 activity may protect against nonglomerular, nonproteinuric kidney injury, including ischemia-reperfusion injury and unilateral ureteral obstruction in animal models (22, 24), but has also been reported to promote podocyte injury and exacerbate proteinuria in chronic kidney disease (23). Our data suggest increased NRF2 activity in FHHt, which may lead to as-yet-unidentified renal phenotypes, but further studies, such as lithium-induced nephrogenic diabetes insipidus, acute kidney injury model, or pharmacological NRF2 activation with bardoxolone methyl, will be required.

Our data from Cul3^+/−/Δ9 mice support the concept that CUL3-Δ9 can sequester and trap CUL3 substrate adaptors in inactive CRL3s, as has been demonstrated in vitro (3, 6, 10, 12, 20). In both Cul3^+/− and Cul3^+/−/Δ9 mice, KEAP1 was unchanged compared with controls, but NQO1 abundance was significantly higher, consistent with higher NRF2 abundance. However, the effect on NQO1 was greater in Cul3^+/−/Δ9 mice, suggesting KEAP1 sequestration. Although some studies have reported a lack of effect of CUL3-Δ9 on KLHL3 abundance in vitro (3, 20), this has been demonstrated in several mouse models using a KLHL3 antibody validated in Klhl3^−/− mice (5, 7). For example, we found that KLHL3 abundance was about fourfold higher in Cul3^−/− mice than in controls, but in Cul3^−/−/Δ9 mice was even lower (by ∼50%) than in controls, confirming that CUL3-Δ9 potently degrades KLHL3 in vivo (5). Finally, HMW cyclin E abundance did not differ between control and Cul3^+/−/Δ9 mice. This suggests that for a substrate that does not require an adaptor (19), CUL3-Δ9 is unable to interfere, and sufficient WT CUL3 remains to degrade it; CUL1-mediated cyclin E degradation may also contribute (31–33). Together, these data suggest that CUL3-Δ9 exerts different effects on different adaptors and substrates, and the mechanisms involved can also differ (substrate sequestration for KEAP1-Nrf2 vs. adaptor degradation for KLHL3-WNK4). CUL3 can interact with over >120 substrate adaptors that contain a broad-complex, Tramtrack, and Bric à brac (BTB) domain (1). These adaptors determine the substrate bound to the CRL3 (e.g., KEAP1 mediates NRF2 binding). Two recent extensive in vitro studies have shown that CUL3-Δ9 displays enhanced interactions with numerous (>30) CRL3 adaptors (3, 20). This raises the possibility that other CRL3 targets are dysregulated in CUL3-mediated FHHt, the most severe form of the disease (29). In support of this, analysis of single cell RNA-sequencing data indicates segment-specific expression of many CRL3 adaptors in the kidney (34). Changes in the abundance of other CRL3-regulated proteins might therefore contribute to the severe FHHt phenotype or have other effects on renal function. However, studies to explore this will be complicated by a lack of knowledge of targets for the full spectrum of renal CRL3s.

Perspectives and Significance

Mice carrying two CUL3-Δ9 alleles globally display embryonic lethality, precluding study of the full spectrum of its tissue-specific activities. Our data demonstrate that CUL3-Δ9 is unable to form normally functioning CRL3s in the adult mouse kidney, since it cannot rescue the injury and polyuria observed in Cul3^−/− mice. Our data also suggest that loss of AQP2 resulting from Cul3 disruption may reflect collecting duct injury rather than more direct CUL3-dependent AQP2 regulation, but further studies are needed to confirm this. Several in vitro studies have suggested that sequestration of CUL3 substrate adaptors via enhanced binding is a mechanism by which CUL3-Δ9 limits substrate ubiquitination and degradation. Our data showing that abundance of NQO1 is higher in Cul3^+/−/Δ9 mice than in Cul3^+/− mice provide in vivo evidence for adaptor sequestration by CUL3-Δ9. This raises the possibility of additional renal defects beyond dysregulation of NCC-mediated Na⁺ and K⁺ handling in CUL3-mediated FHHt.

SUPPLEMENTAL DATA

Supplemental Table S1: https://doi.org/10.6084/m9.figshare.19985459.v1.

Supplemental Figs. S1–S12: https://doi.org/10.6084/m9.figshare.19985348.v1.

GRANTS

Y.M. received a postdoctoral award from the Uehara Foundation. R.J.C. is funded by National Institute of Diabetes and Digestive and Kidney Diseases Grant Mentored Research Scientist Career Development Award DK120790. M.Z.F. was funded by American Heart Association Postdoctoral Fellowship 17POST33670206 for this work. J.A.M. is funded by National Institute of Diabetes and Digestive and Kidney Diseases Grant DK098141.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

M.Z.F. and J.A.M. conceived and designed research; Y.M., R.J.C., M.Z.F., A.S., L.T.N., and J.A.M. performed experiments; Y.M. and M.Z.F. analyzed data; Y.M., R.J.C., M.Z.F., A.S., L.T.N., and J.A.M. interpreted results of experiments; Y.M. prepared figures; Y.M. and J.A.M. drafted manuscript; Y.M. and J.A.M. edited and revised manuscript; Y.M., R.J.C., M.Z.F., A.S., L.T.N., and J.A.M. approved final version of manuscript.