NUMB dysfunction defines a novel mechanism underlying hyperuricemia and gout

Abstract

Defective renal excretion and increased production of uric acid engender hyperuricemia that predisposes to gout. However, molecular mechanisms underlying defective uric acid excretion remain largely unknown. Here, we report a rare genetic variant of gout-unprecedented NUMB gene within a hereditary human gout family, which was identified by an unbiased genome-wide sequencing approach. This dysfunctional missense variant within the conserved region of the NUMB gene (NUMB^R630H) underwent intracellular redistribution and degradation through an autophagy-dependent mechanism. Mechanistically, we identified the uric acid transporter, ATP Binding Cassette Subfamily G Member 2 (ABCG2), as a novel NUMB-binding protein through its intracellular YxNxxF motif. In polarized renal tubular epithelial cells (RTECs), NUMB promoted ABCG2 trafficking towards the apical plasma membrane. Genetic loss-of-function of NUMB resulted in redistribution of ABCG2 in the basolateral domain and ultimately defective excretion of uric acid. To recapitulate the clinical situation in human gout patients, we generated a NUMB^R630H knock-in mouse strain, which showed marked increases of serum urate and decreased uric acid excretion. The NUMB^R630H knock-in mice exhibited clinically relevant hyperuricemia. In summary, we have uncovered a novel NUMB-mediated mechanism of uric acid excretion and a functional missense variant of NUMB in humans, which causes hyperuricemia and gout.

Fig. 1. Genetic alteration of NUMB gene in patients with inherent hyperuricemia and gout.

a The pedigree of family with inherent hyperuricemia and gout. The squares and the circles indicate males and females, respectively. The diagonal line indicates an individual who had died. Patients with both hyperuricemia and gout are indicated with black-filled shapes. The arrow indicates proband. b NUMB gene structure. The rectangles indicate exons and the lines indicate introns. The filled rectangles indicate CDS region. *Variant identified in the family with inherent hyperuricemia and gout. Representative electropherograms show that the c.1889 G > A NUMB variant (rs375597310) is found in a heterozygous state in the proband (red arrow head, lower plot), but not in the family member who does not have hyperuricemia or gout (red arrow head, upper plot). c NUMB rs375597310 and ABCG2 rs2231142 genotype of family members. d Distribution of the missense variant (R630H) identified in the family and NUMB protein ortholog alignments. The R630H variant is located at highly conserved residue across species. Red arrow head indicates R630H variant.

Fig. 2. NUMB interacts with ABCG2.

a Illustration of the three uric acid transporter proteins containing YxNxxF motif. b-c Co-immunoprecipitation of NUMB and ABCG2. ABCG2-FLAG plasmid or empty vector was transient-transfected into HEK293 cells. Immunoprecipitation was done with FLAG antibody, followed by detecting the co-precipitation of ABCG2-FLAG with endogenous NUMB and ABCG2-FLAG and NUMB in input b. Star indicates non-specific band. On the other hand, NUMB-FLAG plasmid or empty vector was co-transfected with ABCG2-eGFP plasmid. Immunoprecipitation was done with FLAG antibody, followed by detecting the co-precipitation of ABCG2-eGFP with NUMB-FLAG and ABCG2-eGFP and NUMB-FLAG in input c. d Immunofluorescence staining of NUMB and ABCG2 in HEK293 cells. Nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI). Fluorescent signal intensity along the dashed lines are shown in the bottom panel. e Immunofluorescence staining of NUMB and ABCG2 in polarized MDCK cells. ABCG2-eGFP plasmid was transfected into MDCK cells. After reaching 100% confluence, the cells were maintained for another 3–5 days to establish a polarized monolayer. ZO-1 (blue), Na/K ATPase (magenta) and NUMB (red) were stained. NUMB was immuno-stained with rabbit anti-NUMB antibody (Cell signaling technology). Fluorescent signal intensity along lines in the amplified regions numbered with 1–3 are shown in the bottom panel. f Comparison of ABCG2-NUMB co-distribution among apical, lateral and basal side with One-way ANOVA with Tukey’s Multiple Comparison Test (n = 10 per group). g Immunofluorescence staining of NUMB, ABCG2 and early endosome marker EEA1, recycling endosome marker RAB11, late endosome marker RAB7 in HEK293 cells. Fluorescent signal intensity along the dashed lines are shown in h. i The statistic comparison of the triple positive dots per cells in HEK293 cells were carried out with One-way ANOVA with Tukey’s Multiple Comparison Test (n = 5 per group). j Co-distribution of ABCG2-eGFP (green), NUMB (blue) and RAB11 (red) in polarized MDCK cells. ZO-1 (cyan) was stained as an apical marker. k NUMB was immuno-stained with mouse anti-NUMB antibody (Santa Cruz). Fluorescent signal intensity along lines in the amplified regions numbered with 1–3 are shown. l The statistical comparison of the triple positive dots among in apical, lateral and basal domain were carried out with One-way ANOVA with Tukey’s Multiple Comparison Test (n = 10 per group). The results are presented as mean ± SD.

Fig. 3. NUMB-knockout abolished the apical distribution of ABCG2 which affects uric acid excretion.

a Distribution of ABCG2 in polarized NUMB-knockout MDCK cells and control cells. ABCG2-eGFP plasmid was transfected into NUMB-knockout (KO) or control (VE) MDCK cells. After reaching 100% confluence, the cells were maintained for another 3–5 days to establish a polarized monolayer. ZO-1 (blue), Na/K ATPase (magenta) and NUMB (red) were immuno-stained. The Z projection, orthogonal views and the 3D projection images are shown. N: nuclear; Ap: apical side; Ba: basolateral side. b Statistical comparison of the percentage of apical and basolateral ABCG2 between in NUMB-knockout and in control MDCK cells (n = 12 per group). c Co-distribution of ABCG2 (green) and RAB11 (red) in polarized NUMB-knockout MDCK cells and control cells. ZO-1 (cyan) was stained as an apical marker. Fluorescent signal intensity along lines in the amplified regions numbered with 1–6 are shown in the bottom panel. d Statistical comparison of ABCG2/RAB11-double positive signal in each cellular domain between in NUMB-knockout and in control MDCK cells (n = 10 per group). e Detection of NUMB protein in NUMB-knockout (KO) or control (VE) MDCK cells. The representative images of western blotting are shown in the upper panel. β-ACTIN was detected as the loading control. The detection was repeated four times. The statistic result is shown in the bottom panel (n = 4 per group). f Illustration of trans-cellular uric acid transport assay. g–i Statistical comparison of the excretion rate (g, n = 4), reabsorption rate of uric acid (h, n = 3) and the intracellular uric acid (i, n = 3) between in NUMB-knockout (KO) and in control (VE) MDCK cells. j Statistical comparison of uric acid excretion rate among NUMB-knockout (KO) and control (VE) MDCK cells treated with ABCG2 inhibitor or solvent DMSO (n = 3 per group). k Illustration of how NUMB direct ABCG2 to the apical membrane of RTECs. After synthesis, ABCG2 proteins are transported to the cell surface. ABCG2 proteins on the basolateral membrane are endocytosed and encapsulated within BEE, from where the encapsulated ABCG2 were sorted and transported to CRE. NUMB recognizes and binds to ABCG2 in CRE and directs ABCG2 to ARE. From ARE, ABCG2 is delivered to the apical surface. Statistics was calculated using two-tailed unpaired Student’s t-test (e, g–i) or Two-way ANOVA with Tukey’s Multiple Comparison Test (b, d, j). The results are presented as mean ± SD.

Fig. 4. Abnormal intercellular distribution and excessive clearance of NUMB^R630H via autophagy.

a Expression of NUMB^WT and NUMB^R630Hprotein. WT NUMB-3FLAG or NUMB^R630H-3FLAG plasmid was co-transfected with a vector expressing GFP into HEK293 cells, followed by detecting NUMB^WT-FLAG and NUMB^R630H-FLAG proteins. GFP and β-ACTIN were detected as the transfection and loading control, respectively. The detection was repeated three times. The statistic result is shown in the bottom panel (n = 3 per group). b The distribution of NUMB^WT and NUMB^R630H protein. HEK293 cells were transfected with NUMB^WT-3FLAG or NUMB^R630H-3FLAG plasmid, followed by immuno-fluorescence staining with FLAG tag antibody. The white arrow heads indicate the granule-like structures found in cells expressing NUMB^R630H. c, d Immuno-fluorescence staining of NUMB^WT and NUMB^R630H and LC3. NUMB^WT-3FLAG or NUMB^R630H-3FLAG plasmid was co-transfected with GFP-LC3 plasmid into HEK293 cells, followed by immuno-fluorescence staining with FLAG tag antibody and GFP antibody. The white arrow heads indicate the granule-like structures of NUMB^R630H co-stained with LC3. Fluorescent signal intensity along the dashed lines are shown in d. e Comparison of the percentage of NUMB punctate or granule structure co-localized with LC3 between in cells expressing NUMB^WT and in cells expressing NUMB^R630H (n = 10 per group). f Detection of NUMB^WT and NUMB^R630H protein in cells treated with autophage inhibitor 3-MA. NUMB^WT-3FLAG or NUMB^R630H-3FLAG plasmid was co-transfected with vector expressing GFP. Treatment with 3-MA was started 36 h after transfection and last for 12 h. The expression of NUMB^WT and NUMB^R630H protein was detected using FLAG tag antibody and GFP was detected as a transfection control. The detection was repeated three times. The statistic results are shown in the right panel (n = 3 per group). g, h Interactome of NUMB^WT and NUMB^R630H protein. NUMB^WT and NUMB^R630H-3FLAG plasmids were transfected into HEK293 cells. The co-immunoprecipitation was done using FLAG tag antibody. The precipitated proteins were separated by SDS-PAGE electrophoresis. The representative sliver staining image is shown in g upper panel. The immunobloting of immunoprecipitated NUMB^WT and NUMB^R630H protein are shown in g bottom panel. The Venn diagram of interacting proteins of NUMB^WT and NUMB^R630H protein is shown in h. i Immuno-fluorescence staining of NUMB^WT, NUMB^R630H and AP2. HEK293 cells were transfected as mentioned above, followed by immuno-staining with FLAG tag antibody and anti-adaptin α antibody. j Fluorescent signal intensity along the dashed lines. k Statistical comparison of the percentage of NUMB punctate structure co-stained with AP2 between in cells expressing NUMB^WT-FLAG and in cells expressing NUMB^R630H-FLAG (n = 10 per group). l The scatterplot images of co-localization analysis between AP2 and NUMB^WT or NUMB^R630H. m Statistical comparison of the excretion rate of uric acid among control MDCK cells (n = 4) and MDCK cells stably expressing NUMB^WT (n = 4) and NUMB^R630H (n = 3). n The expression of NUMB^WT-FLAG and NUMB^R630H-FLAG. MDCK cells were infected with control lentivirus (control, VE), lentivirus/ NUMB^WT-FLAG or entivirus/NUMB^R630H-FLAG, followed by selection with puromycin for 3 weeks. GFP and GAPDH were detected as the infection control and loading control, respectively. Statistics was calculated using two-tailed unpaired Student’s t-test (a, e, k), Two-way ANOVA with Tukey’s Multiple Comparison Test (f) or One-way ANOVA with Tukey’s Multiple Comparison Test (m). The result is presented as mean ± SD.

Fig. 5. The defection in uric acid excretion and the abnormal distribution of NUMB and ABCG2 protein in renal tubular epithelia cells of NUMB^R632H/NUMB^R632H homozygous mice.

a Illustration of the establishment of orthologous NUMB^R632H mutation knock-in mice. b Representative electropherograms show the generation of mice heterozygous (WT/RH) or homozygous (RH/RH) for the introduced mutation (blue shadow). c Chi-square to compare observed and expected distributions of three genotype according to Mendel’s law. d–g Comparison of the ratio of urinary uric acid to urinary creatinine (d, WT/WT n = 8; RH/RH n = 10; t = 2.600, df = 16), the fractional excretion of uric acid (FEUA) (e, WT/WT n = 5; RH/RH n = 8; t = 2.570, df = 11), serum urate (f, WT/WT n = 6; RH/RH n = 8; t = 2.272, df = 12) and creatinine clearance (g, WT/WT n = 8; RH/RH n = 8; t = 0.5308, df = 14) between in WT/WT mice and in NUMB^R632H/NUMB^R632H (RH/RH) mice. h Representative images of immunofluorescence staining of NUMB in kidney tissue of WT/WT mice and NUMB^R632H /NUMB^R632H mice. The solid lines indicate the edge of renal tubules, and the dashed lines mark the lumen of renal tubules. T: renal tubules; Ap: apical side of renal tubular epithelia cells (RTECs); Ba: basolateral side of RTECs. White arrows indicate sporadic distribution of granule-like structure of NUMB^R632H protein in cytoplasm of RTECs. i The statistic comparison of NUMB protein (top panel), apical NUMB (middle panel) and NUMB aggregate-like structure (bottom panel) each renal tubule between in WT/WT mice (n = 7) and in NUMB^R632H/NUMB^R632H mice (n = 6). j Representative images of immunofluorescence staining of ABCG2 and basolateral marker Na/KATPase. The white arrow heads indicate the distribution of ABCG2 (red). In WT/WT mice, ABCG2 is distributed to the apical side just beneath the LTL signal. However, in NUMB^R632H /NUMB^R632H mice, ABCG2 located to the basolateral side with increased co-localization with Na/K ATPase. The illustration of ABCG2 distribution is shown in the bottom left, and the statistic result of the percentage of apical ABCG2 is shown in the bottom right (n = 6 per group, t = 19.34, df = 10). Statistics was calculated using two-tailed unpaired Student’s t test. The results are presented as mean ± SD.

Fig. 6. Chronic uric acid nephropathy in NUMB^R632H /NUMB^R632H homozygous mice.

a H&E staining of the kidney tissue of WT/WT mice and NUMB^R632H /NUMB^R632H mice. The stars indicate the proximal tubules (PTs), the red arrow heads indicate the detached RTECs and the black arrow heads indicate the inflammatory cells infiltration. b–d Statistic comparison of the height of RTECs in PTs, the width of lumen of PTs and the percentage of renal tubules (RTs) with RTECs detached from basement membrane between in WT/WT mice (n = 7) and NUMB^R632H /NUMB^R632H mice (n = 6). e, f TEM images of renal tubules of WT/WT mice and NUMB^R632H /NUMB^R632H mice. Red arrow heads indicate vacuoles with diameter over 100 nm. Nu: nuclear; Mv: microvilli; M: mitochondria; Va: vacuole; star: non-cellular structures in the lumen. The statistic result of the amount of vacuoles with diameter over 100 nm per 100 μm² is shown in e (n = 3, t = 3.228, df = 4). g Immunohistochemical staining of F4/80, CD169 and α-SMA. Red arrow heads indicate the positive cells. The statistic result of the percentage of F4/80⁺ (n = 6, t = 0.159, df = 11), CD169⁺ (n = 6, t = 8.428, df = 11) and α-SMA⁺ (n = 6, t = 5.870, df = 11) cells in total peritubular cells are shown in the bottom. h The representative gloss examination photos and H&E staining of renal tissue with severe inflammation in NUMB^R632H /NUMB^R632H mice. i Immunofluorescence staining of active Caspase-1. The tissues were also stained with LTL (red) and DAPI (blue). The statistic result of active Caspase-1 signal is show in the bottom panel (n = 6, t = 5.141, df = 10). Statistics was calculated using two-tailed unpaired Student’s t test (b–g, i). The results are presented as mean ± SD.