Targeted Inactivation of EGF Receptor Inhibits Renal Collecting Duct Development and Function

Abstract

The ureteric bud (UB) expresses high levels of the EGF receptor (EGFR) during kidney development, but its function in this setting is unclear. Here, Egfr mRNA was abundant in medullary portions of the UB trunk but absent from the branching UB tips during embryogenesis. Homozygous Egfr knockout did not affect the pattern of UB arborization, but renal papillae were hypoplastic and exhibited widespread apoptosis of tubular cells. Because these EGFR-deficient mice die within 1 week of life, we targeted Egfr inactivation to the renal collecting ducts using Cre-lox technology with a Hoxb7-Cre transgene. This targeted inactivation of Egfr led to a thin renal medulla, and at 7 weeks of age, the mice had moderate polyuria and reduced urine-concentrating ability. At 30 to 33 weeks, water deprivation demonstrated a continued urine-concentrating defect despite similar levels of vasopressin between knockout mice and littermate controls. Taken together, these results suggest that unlike other tyrosine kinases expressed at the UB tip, EGFR functions primarily to drive elongation of the emerging collecting ducts and to optimize urine-concentrating ability.

During development, the EGF receptor (EGFR) is highly expressed in developing organs, suggesting an important role in organogenesis. Egfr ligands are numerous, but all induce a molecular cascade involving receptor dimerization, tyrosine kinase activity, signaling via the mitogen-activated protein kinase pathway, and cell proliferation. Egfr is highly expressed in fetal kidney at the basolateral surface of emerging tubular cells but is rapidly downregulated in the postnatal period as nephrogenesis comes to an end¹; however, the precise role of EGFR in renal development has been difficult to unravel because homozygous knockout of the Egfr gene is lethal in early stages of embryogenesis for many mouse strains.^2,3 On the CD1 background, Egfr(−/−) mice which survive into the neonatal period exhibit progressive wasting until they die in the first 1 to 2 weeks of life. In this study, we report that targeted Egfr knockout has no effect on ureteric bud (UB) branching, but limits elongation of collecting ducts in the papilla and blunts renal concentrating ability in postnatal life.

To define the Egfr expression pattern during renal development, we performed in situ hybridization studies on cryosections of embryonic day 15.5 (E15.5; Figure 1A) and E18.5 (Figure 1, B and C) mouse kidney. Egfr is predominantly expressed in maturing collecting ducts of the renal medulla and is conspicuously absent in the outer nephrogenic zone, where the collecting duct system arborizes through branching at UB tips (Supplemental Figure S1). Our observations are consistent with the immunohistochemical studies of human fetal kidney reported by Bernardini et al.⁴ To assess the importance of Egfr expression for early development of the collecting duct system, we examined medullary structure of homozygous Egfr^neo/neo mice, previously reported by Threadgill et al.³ The mice were severely runted at birth (Figure 1D) and died within the first week of life. When examined on the third day of life, mutant kidneys were hypoplastic (Figure 1E) and showed striking atrophy of the renal papilla (Figure 1F) compared with heterozygous Egfr^neo/wt littermates (Figure 1G). The papillary tip was flattened, and medullary collecting tubules were moderately dilated. Terminal deoxynucleotidyl transferase–mediated digoxigenin (DIG)-deoxyuridine nick-end labeling (TUNEL) staining revealed widespread apoptosis of papillary epithelia (Figure 1H) compared with heterozygotes (Figure 1I).

Figure 1.

Egfr mRNA is expressed in normal mouse fetal kidney and renal phenotype in Egfr^neo/neo mice. (A through C) E15.5 (A) and E18.5 (B and C) wild-type CD1 mouse kidney cryosections were probed with a 150-bp cRNA for mouse Egfr exon 3. Egfr mRNA is expressed predominantly in maturing collecting ducts of the renal medulla. (D) In one informative Egfr^neo/neo mouse surviving until postnatal day 3, severe runting was noted at birth. (E through G) Kidneys were hypoplastic (E); the papillary tip was flattened and medullary collecting tubules were moderately dilated (F) compared with wild-type or heterozygous Egfr^neo/wt littermates (G). (H and I) TUNEL staining of sections from P3 kidney revealed widespread apoptosis in papillary collecting duct cells of Egfr^neo/neo mice (brown color; H) compared with heterozygotes (I).

Because ubiquitous Egfr knockout produces generalized embryonic effects and perinatal death, we targeted Egfr inactivation to renal collecting ducts by generating CD1 mice with a Hoxb7_promCre/Egfp transgene, an Egfr knockout allele (Egfr^neo), and an Egfr allele in which exon 3 was flanked by loxP sites (Figure 2A). Selective expression of the Egfp-tagged Cre-recombinase transgene in collecting ducts is illustrated in Figure 2, B and C. Successful targeted inactivation of Egfr in renal collecting ducts was confirmed by in situ hybridization of E18.5 mouse kidney cryosections (Figure 2, D and E). Of 137 Egfr mice with Hoxb7-targeted inactivation of Egfr, all survived in the predicted Mendelian proportions.

Figure 2.

Hoxb7-targeted Egfr is inactivated in renal collecting ducts. (A) Mice with loxP sites introduced into the Egfr gene allow Cre-recombinase excision of exon 3, introducing a frameshift that truncates the Egfr transcript. Mice with a Hoxb7_promCre/Egfp transgene express the Cre-recombinase open reading frame immediately downstream of the Hoxb7 promoter; an internal ribosomal entry site (IRES) permits coexpression of enhanced green fluorescence protein cDNA (Egfp). (B and C) Egfr^CDKO mice exhibit EGFP throughout the trunk and tips of the UB in sections (B) and whole mounts (C) of E18.5 kidney. (D and E) Targeted inactivation of Egfr was confirmed by comparing mRNA expression in Egfr^flox/flox control kidney (D) with that in Egfr^CDKO mice (E). (F and G) A representative E14 Egfr^CDKO kidney (F) and Egfr^flox/flox littermate control (G) illustrate the branching UB marked by Hoxb7_promCre/Egfp. (H) UB branch tips (arrows) were counted with imaging software; branch tip number was equivalent in Egfr^CDKO embryos (n = 4) versus Egfr^flox/flox littermate controls (n = 5) lacking the Cre-recombinase transgene. Magnifications: ×10 in B; ×1.25 in C; ×4 in D and E; ×3.2 in F and G.

Effects of EGF on filopodia formation in mIMCD3 cells led Sakurai et al.⁵ to propose that EGFR ligands account for a large fraction of branching morphogens in fetal kidney; however, mIMCD cells (which lack RET expression) may not be representative of UB tips, and it has been difficult to confirm that EGF directly induces arborization of the UB. We examined the pattern of EGFP expression in E14 Egfr^flox/neo/Hoxb7_promCre/Egfp (Egfr^CDKO) mice and counted the number of UB branch tips. As illustrated in Figure 2, F and G, UB branching pattern and branch tip number were equivalent in embryos with targeted Egfr knockout (Figure 2F) and controls lacking the Cre-recombinase transgene (P = 0.19; Figure 2, G and H) and identical to heterozygous Egfr^neo/Hoxb7_promCre/Egfp (data not shown). This experiment does not rule out indirect effects of EGF via stromal cells or through release of other morphogens but makes it unlikely that EGFR is directly involved in UB branching.

The UB expresses other tyrosine kinase receptors such as MET, FGFR2, and RET. Unlike EGFR, however, these other receptors are expressed at the UB tip. Targeted inactivation of the murine Fgfr2 (or Fgfr1 genes) causes a 30% decrease in nephron number,^6,7 and Costantini's group⁷ demonstrated the important role for RET (GDNF receptor) in UB branching. Ishibe et al.⁸ used the Hoxb7 promoter to target inactivation of the Met gene in UB cells and noted a 35% reduction in final nephron number. Interestingly, Egfr mRNA and Egfr phosphorylation were increased in the mutant mice, and the Met(−/−) branching defect was significantly more severe when combined with a homozygous missense mutation in the Egfr gene.⁸ We also noted a 40 ± 2.4% decrease in E14 UB branch tip number in mice with untargeted Egfr^neo/neo knockout (n = 13) versus controls (n = 38; P < 0.001); however, the conventional global Egfr knockout affects many developing organs, and the mice are severely runted.³ Presumably, the effect of generalized Egfr knockout on UB branching is indirect, because it does not occur in Hoxb7-targeted knockout of Egfr in the collecting duct.

Elongation of collecting ducts in the renal papilla occurs in the perinatal period and requires sustained postnatal cell division in the proximal papilla.⁹ We tracked the morphogenesis of the renal medulla into adult life. At 30 to 33 weeks of age, the renal papilla remained hypoplastic (Figures 3, A and B), but aquaporin 2 (AQP2) expression was normal in residual papillary collecting ducts (Supplemental Figure 2). Kidney weight (0.21 ± 0.03 g) of Egfr^CDKO mice was 20% less than that of controls (lacking Cre-recombinase; P = 0.03; Figure 3C). Although the renal cortex was normal, medullary thickness was only 84% of controls and medulla-cortex ratio was 77% of control (P = 0.02; Figure 3, C and D). Thus, the mechanisms regulating collecting duct elongation involve EGFR and seem to be distinct from those that drive UB branching during embryonic development. Yu et al.¹⁰ recently reported that Sox2 promoter-targeted inactivation of Wnt7b causes profound hypoplasia of the renal medulla. Similar to our observations of Egfr mRNA, Wnt7b is expressed in epithelial cells of the UB trunk but is absent from its tips.

Figure 3.

Hypoplasia develops in the renal medulla in Egfr^CDKO mice. (A) At 30 to 33 weeks of age, wild-type Egfr^flox/flox controls lacking the Cre-recombinase transgene had well-developed renal papillae projecting into the renal calyx. (B) In contrast, Egfr^CDKO littermates had blunted renal papillae. (C and D) Morphologic characteristics of Egfr^CDKO (n = 4) and Egfr^flox/flox control (n = 4) kidneys (C) were derived from measurements at three points on six serial sagittal sections from the center of both kidneys (D).

To examine the impact of Hoxb7-targeted Egfr knockout on adult collecting duct function, we analyzed renal concentrating ability in mutant mice. At 7 to 11 weeks of age, Egfr^CDKO mice exhibited polyuria during water deprivation. During 12 hours, urine volume of mutant mice was 6.2 times greater than that of controls (8.05 ± 5.50 versus 1.53 ± 0.85 μl/g body wt), and the mutants lost 9% body weight (versus 7.8% body weight loss in controls; P = 0.02; Figure 4, A and B). In response to deprivation, control mice increased urine osmolarity from 2111 ± 368 to 2500 ± 169 mOsm/kg, whereas mutant mice had reduced baseline urine osmolality (1209 ± 422; P = 0.03 versus controls) and failed to increase water resorption after 12 hours of water deprivation (902 ± 227 mOsm/kg; P < 0.001; Figure 4C).

Figure 4.

Egfr^CDKO mice have urine-concentrating defect. At 7 to 11 weeks of age, mice were deprived of water for 12 hours. (A) Body weight loss (% of baseline) was greater in Egfr^CDKO mice (n = 4) than in Egfr^flox/flox controls (n = 4) lacking the Cre-recombinase transgene. (B) Urine output during water deprivation was greater in Egfr^CDKO mutants (n = 4) than Egfr^flox/flox controls. (C) Urine osmolarity (mosm/kg) was measured at baseline (0 to 2 hours) and at the end of the water deprivation period (10 to 12 hours); urine osmolarity was reduced in Egfr^CDKO mice (▴) at baseline (*P = 0.03) and at the end of the water deprivation period (**P < 0.001) compared with Egfr^flox/flox controls (■).

A longer water deprivation test was performed at 33 weeks of age. After 22 hours of water deprivation, mean plasma vasopressin level of the control mice rose from 567 ± 34 to 2543 ± 827 pg/ml and urine osmolarity rose to 3604 ± 697 mosm/kg. In contrast, urine osmolarity of Egfr^CDKO mice was only 1910 ± 302 mOsm/kg after 22 hours of water deprivation (P = 0.02 versus controls) despite the fact that plasma vasopressin (2082 ± 956 pg/ml) was not significantly different from controls (P = 0.07).

In a previous description of Egfr^neo/neo knockout mice, Threadgill et al.³ noted dilation of renal collecting ducts in mutant kidneys; we observed a similar phenomenon in Egfr^CDKO mice with Hoxb7-targeted disruption of Egfr.

In humans, nephrogenic diabetes insipidus may be caused by X-linked mutations of the vasopressin receptor¹¹ or recessive mutations of the AQP2 water channel¹²; however, water resorption is completely eliminated in these conditions, and profound polyuria can be lethal during periods without fluid intake. In Egfr^CDKO mice, the renal papilla is atrophic and collecting duct elongation is blunted during development, but there is clearly some residual ability to concentrate the urine. The Egfr^CDKO phenotype overlaps somewhat with that of mice bearing targeted mutations of the angiotensinogen (Agt−/−), angiotensin-converting enzyme (Ace−/−), or angiotensin receptor-1 (Agtr−/−) genes. These last two mutations disturb proximal tubular development, but the knockout mice also have medullary hypoplasia, atrophy of renal papillae, dilation of the renal pelvis, and impaired urine-concentrating ability.^13–19 Thus, EGFR may normally cooperate with both WNT7b and AT₁R signaling during maturation of the renal collecting ducts.²⁰

Concise Methods

Experimental Mice

All mice were maintained on CD-1 outbred stock. For timed pregnancies, males and females were put together in the late afternoon and separated the next morning (designated E0.5).

Generation of Homozygous Egfr^neo/neo Knockout Mice.

Egfr^neo/wt have been previously described.³ Briefly, 155 bp surrounding the splice acceptor site of exon 2 was replaced with a Neo (Neomycin) cassette, causing aberrant splicing and a null frameshift mutation.³ Hoxb7_prom/Gfp mice bearing a green fluorescence protein (GFP) transgene under the control of the murine Hoxb7 promoter,²¹ was provided by Dr. Frank Costantini (Medical Center of Columbia University, New York, NY). Heterozygous Egfr^neo/wt mice were mated with Hoxb7_promGfp-expressing mice to produce Egfr^neo/neo/Hoxb7_promGfp compounds.

Generation of Targeted Egfr^CDKO Mice.

Mice bearing a floxed Egfr exon 3 allele have been previously described.²² Hoxb7_promCre/Egfp transgenic mice were a gift from Dr. Carlton Bates (Ohio State University, Columbus, OH).⁶ Heterozygous Egfr^neo/wt mutant mice were crossed with Hoxb7_promCre/Egfp mice to introduce Cre recombinase throughout the fetal UB and renal collecting ducts. These mice were then bred to homozygous Egfr^flox/flox mice to produce offspring with collecting duct–targeted inactivation of both Egfr alleles (Egfr^CDKO).

Genotyping

For genotyping, genomic DNA was extracted from tail samples with the wizard DNA purification Kit (Promega, Madison, WI). The Egfr^neo null allele was identified using primers for the Neo cassette: 5′-GCCCTGCCTTTCCCACCATA-3′ and 5′-TTGCAGCACATCCCCCTTTC-3′. The wild-type Egfr allele was identified with the following primers: 5′-GCCCTGCCTTTCCCACCATA-3′ and 5′-ATCAACTTTGGGAGCCACAC-3′. The Hoxb7_promCre/Egfp allele was identified with the following primers: 5′-AGCGCGATCACATGGTCCTG-3′ and 5′-ACGATCCTGAGACTTCCACACT-3′. The Egfr^flox allele was identified with the following primers: 5′-ACACTAGCACTGACTGCTGG-3′ and 5′-GGCGAGATAAACCCAAAGCA-3′.

RNA In Situ Hybridization with DIG-Labeled cRNA Probes

The efficiency of Egfr deletion from the UB was determined by RNA in situ hybridization. The DIG-labeled riboprobe was generated by amplifying a 150-bp fragment of Egfr exon 3 with the following primers: 5′-ATGTCCTCATTGCCCTCAAC-3′ and 5′-GCATGGGCAGTTCCCTAA-3′. The resulting fragment was cloned into TA cloning vector kit, and riboprobes were generated with a DIG-dUTP cRNA probe synthesis kit (Roche, Mannheim, Germany).

All solutions were prepared in 0.1% DEPC water. E18.5 fetal mouse kidneys were fixed at 4°C overnight in 4% paraformaldehyde, transferred to 30% sucrose/PBS, and snap-frozen in OCT (Sakura Finetek USA Inc., Torrance, CA). Cryosections were prepared on superfrost plus slides, treated with proteinase K (20 μg/ml) in 1× PBS, and acetylated for 15 minutes in a solution containing 175 μl of acetic anhydride and 70 ml of 0.1 M triethanolamine. A total of 100 μl of prewarmed hybridization solution (65°C) was then added to the slides for 5 minutes at 65°C.

Before in situ hybridization, 1 μl of DIG-labeled cRNA probe (1 μg/μl) mixed with 100 μl of hybridization solution was denatured at 85°C for 3 minutes and applied to the slides at 65°C overnight. Posthybridization washing included 1 × SSC/50% formamide for 30 minutes at 65°C, TNE + RNase A (20 μg/ml) for 30 minutes at 37°C, 2 × SSC for 20 minutes, 0.2 × SSC for 20 minutes at 65°C, MABT for 5 minutes ×2 at room temperature, and blocking with 20% goat serum/MABT for 1 hour and DIG-AP (1:500) in 2% goat serum/MABT at 4°C overnight in a humidified chamber. The slides were then rewashed with 1 × MABT for 5 minutes ×3 and 1 × NTMT (pH 9.5) for 10 minutes at room temperature. For color development, 2 to 3 ml of prewarmed BM purple (37°C; Roche) was applied for 1 to 72 hours (in the dark), and the slides were rinsed in 1 × NTMT (pH 9.5) and fixed in 4% PFA/PBS for 10 minutes.

TUNEL Assay

Papillary cell apoptosis was detected with the In Situ Cell Death Detection Kit, POD (Roche). Paraffin sections were dewaxed, treated with proteinase K (20 μg/ml) for 10 minutes, incubated with 50 μl of TUNEL reaction mixture at 37°C for 2 hours in the dark, and rinsed three times. The slides were incubated with 50 μl of converter-POD for 30 minutes at 37°C and then 50 μl of diaminobenzidene tetrahydrochloride substrate for 10 minutes at room temperature, rinsed three times, and counterstained with hematoxylin.

Immunohistochemical Staining

Sections (5 μm) of mouse kidneys (30 to 33 weeks) were deparaffinized and rehydrated. Antigen retrieval was performed by microwave boiling of tissues in citrate-based buffer (pH 6). After cooling, the slides were incubated in 3% hydrogen peroxide for 20 minutes at room temperature to block endogenous peroxidase. Sections were then washed in 1 × PBS for 5 minutes three times, and nonspecific protein binding was blocked by incubation in 1% horse serum for 1 hour. Slides were then incubated in primary rabbit anti-mouse AQP2 (H-40) antibody (1:100; Santa Cruz Biotechnology, Santa Cruz, CA) for 1 hour at room temperature followed by three washes with 1× PBS for 5 minutes. Bound primary antibody was detected with universal biotinylated secondary antibody in blocking buffer (1% horse serum) for 1 hour at room temperature; slides were then incubated with avidin and biotinylated horseradish peroxidase according to the manufacturer's instructions, washed in PBS, and developed in diaminobenzidene tetrahydrochloride. As a negative control, normal rabbit serum was substituted for primary AQP2 antibody.

Quantification of UB Branching

The number of UB tips was determined in E14 mouse kidneys using Image J software (http://rsb.info.nih.gov/ij/) as described previously.²³

Water Deprivation Test

Two pairs of female and two pairs of male littermates (control versus mutants) from three different litters were studied at 7 to 11 weeks age. Body weight and urine volume were measured at baseline (0 to 2 hours) and after 12 hours of water deprivation. Urine samples were obtained by bladder massage after each 2-hour interval.²⁴ Urine osmolality was measured by freezing-point depression (microosmometer; Biochemistry Laboratory, Montreal Children's Hospital).

The same four pairs of mice were restudied at 30 to 33 weeks during a more prolonged period (22 hours) of water deprivation. At the end of the study period, plasma was obtained by cardiac puncture and assayed for vasopressin by RIA as described previously.²⁵

Relative Thickness of Renal Cortex versus Medulla

To assess the thickness of renal cortex and medulla, we prepared paraffin-embedded 7-μm sagittal sections of mouse kidney stained with hematoxylin and eosin. Mean cortical thickness was calculated by measurement (SPOT 3.5.9 image analysis software) in three separate places, from outer tissue rim to the edge of the outer medulla, in six serial sections from the center of the kidney.²⁶ Mean medullary thickness was calculated by measuring the distance from the edge of the outer medulla to the papillary tip in multiple sections as already described.

Statistical Analysis

Mean values for each parameter were compared using paired or unpaired t tests. Multiple groups were compared by ANOVA.

Disclosures

None.