Presenilins are required for the formation of comma- and S-shaped bodies during nephrogenesis

Presenilins (PSEN1 and PSEN2) are polytopic transmembrane proteins that are essential to process the β-amyloid precursor protein (APP) at theγ-secretase site to generate Aβ40 and Aβ42, peptides that constitute major components of the β-amyloid plaques characteristic of Alzheimer's disease (AD) (De Strooper et al., 1998; Selkoe,1998). Autosomal dominant inheritance of mutations in PSEN1 and PSEN2 leads to familial Alzheimer's disease and these mutations are known to affect the γ-secretase activity and to foster the β-amyloid plaque pathology(Selkoe, 1998). Through similar mechanisms, presenilins are required for the proteolytic cleavage of Notch (De Strooper et al.,1999), molecules that are crucial in various cell-fate specification processes, and this activity is highly conserved(Levitan et al., 1996; Levitan and Greenwald, 1995; Struhl and Greenwald, 1999; Ye et al., 1999). In mammalian systems, the two presenilins play compensatory roles in the Notch signaling pathway, as revealed by gene knockout studies. Specifically, Psen1-null mice die pre- or perinatally, exhibiting a Notch1-associated somite segmentation defect(Shen et al., 1997; Wong et al., 1997). Although the Psen2 knockout does not have overt impairment, Psen1-and Psen2-double (PSEN) null embryos die at embryonic day 9.5 (E9.5)with complete absence of somite patterning, which resembles a complete Notch1 loss-of-function, or more precisely, Notch1- and Notch4-double knockout phenotypes(Conlon et al., 1995; Donoviel et al., 1999; Herreman et al., 1999; Krebs et al., 2000). The redundant activities of PSEN1 and PSEN2 may be the explanation for the lack of overt organogenesis defects in each of the single knockouts. The early lethal phenotype of the PSEN-null embryos prevents the evaluation of potential physiological roles of presenilins in mid-to-late embryogenesis.

Organogenesis is a highly regulated developmental program involving cell-fate decisions, pattern formation and differentiation. In mammals,organogenesis initiates after the body axes are established. Kidney organogenesis in mice begins at E11 and involves reciprocal inductive interactions between the ureteric bud epithelium and metanephric mesenchyme. The tips of the ureteric bud induce nephrogenic mesenchyme to form condensates followed by pretubular aggregates, which then undergo mesenchyme to epithelium transition (MET) to progress into polarized epithelia of renal vesicles and comma- and S-shaped bodies (Saxen,1987). Morphogenesis and patterning of the epithelial structures then leads to the formation of distal and proximal tubules and the glomerulus;the latter consists of podocytes, mesangial cells and endothelial cells(Saxen, 1987).

A number of ductal- and mesenchymal-derived molecules have been identified and loss-of-function studies have established their crucial roles in nephrogenesis(http://golgi.ana.ed.ac.uk/kidhome.html). Relevant to this study, Ret is expressed at the tips of the newly formed branches of the ureteric bud and is crucial for the inductive interactions through binding to its ligand GDNF, which resides in the condensing mesenchyme (Moore et al.,1996; Pachnis et al.,1993; Schuchardt et al.,1994). The paired box genes Pax2 and Pax8 are expressed in various mesenchymal derivatives and ureteric bud epithelium(Dressler et al., 1993; Plachov et al., 1990; Rothenpieler and Dressler,1993), and together they regulate and define the nephric lineage(Bouchard et al., 2002). Wnt4 has been documented as an essential signaling molecule from the metanephric mesenchyme for the mesenchymal to epithelial cell conversion(Kispert et al., 1998; Stark et al., 1994). WT1 is expressed in both uninduced mesenchyme and podocyte precursor cells of S-shaped bodies and mature glomeruli. It is required in the earliest phase of kidney formation (Armstrong et al., 1993; Kreidberg et al.,1993). The winged helix transcription factor BF2 is specifically expressed in the interstitial stromal mesenchyme and has been shown to modulate the transition of condensed mesenchyme into tubular epithelium (Hatini et al.,1996).

The Notch pathway has been implicated in kidney development. Specifically, Notch1 and Notch2 and their ligands Delta-like1(Dll1) and Jag1 are expressed in the maturing nephron and glomerulus (Beckers et al.,1999; McCright et al.,2001; Weinmaster et al.,1991; Weinmaster et al.,1992). In the Xenopus pronephros, Notch signaling has been proposed to be involved in both the early determination of duct versus tubule fate as well as in controlling tubule patterning(McLaughlin et al., 2000). Furthermore, mice with a hypomorphic expression of Notch2 exhibit a kidney glomerulogenesis defect (McCright et al., 2001).

We reported earlier that restricted expression of a human PSEN1transgene in the developing brain and vertebral/spinal column driven by the human Thy-1 promoter could rescue the Psen1-null patterning defects and lethal phenotype (Qian et al.,1998). To bypass the early lethality of the PSEN null embryos, we bred the transgene onto the Psen1- and Psen2-double knockout background. We showed that expression of the human PSEN1 transgene supported the survival of the PSEN-null embryos to the perinatal stage. However, loss of presenilins resulted in profound nephrogenesis defects prior to development of comma- and S-shaped bodies. We provide evidence that presenilins play an important role, probably through the Notch signaling pathway, in the patterning of renal epithelial structures.

The generation and genotyping of Psen1- and Psen2-null and human PSEN1 transgenic mice have previously been described(Donoviel et al., 1999; Qian et al., 1998; Wong et al., 1997). The breeding strategy is shown in Fig. 1A. The mice analyzed in this study were on a mixed genetic background of three strains (129, C57 and SJL). The day when the vaginal plug was observed was considered to be embryonic day (E) 0.5. Embryos with at least one wild-type allele of Psen1 gene were used as littermate controls.

Embryos were dissected and fixed in 10% neutral buffered formalin (Sigma)for 24-48 hours and dehydrated through graded alcohols and stored in 70%ethanol at 4°C. Embryos were vacuum-embedded in paraffin wax, sectioned at 5 μm and stained with Hematoxylin and Eosin.

In situ hybridization was carried out using digoxigenin-labeled or[³⁵S]UTP-labeled antisense riboprobes. Samples were fixed in 10%neutral buffered formalin and embedded in paraffin wax. Sections were cut at 5μm for radioactive probes and at 9 μm for nonradioactive probes. The protocol used for radioactive in situ was essentially as published(Qiu et al., 1994). For non-radioactive in situ, a modified protocol of Wilkinson was employed(Wilkinson, 1992). In brief,sections were dewaxed and dehydrated, then fixed in 4% paraformaldehyde (PFA)for 30 minutes at room temperature. After washing with PBS, sections were treated with proteinase K (20 μg/ml) for 10 minutes, refixed with PFA and washed with PBS. Samples were acetylated for 20 minutes, washed and hybridized(in 50% formamide, 20 mM Tris-HCl (pH 8.0), 500 μg/ml tRNA,1×Denhardt's solution, 10% dextran sulfate, 300 mM NaCl, 10 mM NaPO₄ (pH 8.0), 5 mM EDTA) with 200 ng probe/slide overnight at 65°C in a humidified chamber. Sections were washed for 40 minutes with 0.5×SSC/20% formamide at 60°C followed by treatment with RNase for 30 minutes at 37°C, and washed again for 20 minutes with 0.5×SSC/20%formamide at 60°C. Sections were blocked with blocking solution (100 mM maleic acid, 150 mM NaCl, 1% Boehringer blocking agent) for 1 hour at room temperature and incubated with alkaline phosphatase-conjugated sheep polyclonal antidigoxigenin antibody (Roche) diluted 1:2000 in TBS buffer overnight at 4°C. After washing twice with TBS, alkaline phosphatase activity was detected in the presence of NBT/BCIP (Roche).

PSEN-null and the littermate control kidneys were dissected at various developmental stages as specified in the figure legends. Immunohistochemical staining was performed in frozen (laminin α1, gift from D. Abrahamson)or paraffin-fixed sections (all others). Sections were blocked with 5% goat serum, incubated with primary antibodies at 4°C overnight, washed in PBS,incubated with 1/1000 Alexa Fluor-488 or Alexa Fluor-594 conjugated secondary antibody (Molecular Probe) for 1 hour at room temperature, washed in PBS,mounted in glycerol/PBS. Digital images were obtained with a Zeiss microscope(Axioskop 2). The primary antibodies used were Laminin (Sigma, L9393),E-cadherin (Signal Transduction, C20820), WT1 (Santa Cruz, sc192), cleaved Notch1 (val1744) (Cell Signaling, 2421), cleaved caspase 3 (Asp175) (Cell Signaling, 9661). NCAM monoclonal antibody, developed by T. Jessell, and TROMA1 monoclonal antibody, developed by P. Brulet, were obtained from the Developmental Studies Hybridoma Bank.

Antibody to phosphorylated histone H3 (Upstate Biochemicals) was used to detect mitotic cells (Wei et al.,1999). TUNEL assay was performed as described(Gavrieli et al., 1992). All sections were stained with methyl green to identify nucleated cells. Four-hundred to 600 cells per section were counted and the proportion of mitotic or apoptotic cells was determined as a fraction of the total number of nucleated cells, four kidneys/stage/genotype, five sections/kidney were counted.

Embryonic day 12.5 kidneys were isolated and cultured in MEM/F12 containing 10% fetal bovine serum on Millicell culture plate insert (Millipore) at the medium/gas interface for 3 days. Kidneys were fixed in 2% paraformaldehyde for 10 minutes, then in 95% methanol for 15 minutes and washed in PBST(phosphate-buffered saline 0.1 M, pH 7.4, 0.1% Tween 20). They were stained whole-mount with anti-pan-cytokeratin (Sigma, C2562) and WT1 (Santa Cruz,sc192) antibodies at 4°C overnight, washed in PBS, incubated with 1/1000 Alexa Fluor-488 or Alexa Fluor-594 conjugated secondary antibody (Molecular Probe) for 2 hours at room temperature, washed in PBS, mounted in glycerol/PBS and viewed under a Zeiss confocal microscope (LSM510).

To generate mice carrying the human PSEN1 transgene on the endogenous mouse Psen1- and Psen2-double knockout(Psen1^–/–Psen2^–/–PSEN1,or presenilin rescue) background, we bred mice that were heterozygous for mouse Psen1 and positive for PSEN1(Psen1^+/–PSEN1, line 16-4)(Qian et al., 1998) with Psen2-null (Psen2^–/–)(Donoviel et al., 1999) to produce offspring that were compound heterozygous for endogenous presenilins and positive for PSEN1(Psen1^+/–Psen2^+/–PSEN1). These mice were then crossed twice with animals heterozygous for Psen1 and homozygous for Psen2(Psen1^+/–Psen2^–/–)to yield the desired mutant genotype(Psen1^–/–Psen2^–/–PSEN1)(Fig. 1A).

We reported earlier that expression of the PSEN1 transgene was able to rescue the somite patterning defect of the Psen1-null mice(Qian et al., 1998). Similarly, such transgene expression restored the somite structures that were completely absent in PSEN-null embryos(Fig. 1B)(Donoviel et al., 1999). Presenilin rescue mutants were indistinguishable when compared with their littermate controls prior to E12.5. At E13.5, the mutants were overtly similar to their littermates, although they were slightly smaller in size, and this difference became more dramatic over time(Fig. 1B). The mutants could be unambiguously identified because their eyes lacked pigmentation, a phenotype that is the subject of a separate study.

The presenilin rescue mutants died within 2 hours of birth. Examination of the organs revealed that, at P0, the kidneys were dramatically smaller than that of the littermate controls (Fig. 1C), suggesting a defect in kidney development. Western blot analysis showed that the PSEN1 protein could be readily detected in control kidneys (Fig. 1D, lane 1,2). However, as it was not detectable in Psen1^–/–Psen2^–/–PSEN1rescue mutant (Fig. 1D, lane 3), it suggests that the defect is due to the loss of presenilins and the mutant kidney is herein referred as PSEN-null.

Histological analysis at various stages of kidney development revealed that, at E12.5, the ureteric buds were readily identified and were surrounded by condensed mesenchyme in both the control and the PSEN-null kidneys(Fig. 2A,B, asterisks). This result suggests that the inductive interactions between the ureteric bud epithelium and the metanephric mesenchyme proceeded normally in the absence of presenilins. However, at E13.5, while the kidney development was further advanced in the control, as evidenced by the appearance of pretubular aggregates/renal vesicles (distinguished upon further analysis, see Fig. 4) and comma- and S-shaped bodies (Fig. 2C, thin arrow and arrowheads respectively), only pretubular aggregates/renal vesicles(Fig. 2D, thin arrow) could be found in the PSEN-null kidney but no comma- and S-shaped bodies were identified. This result indicates that presenilins play a critical role in the progression of pretubular aggregates/renal vesicles towards comma- and S-shaped bodies during nephrogenesis. Analysis of kidneys at later stages(E15.5 and P0) supports this view as mature glomeruli, which could be readily detected in the controls (Fig. 2E,G, thick arrows), were completely absent in the PSEN-null kidneys (Fig. 2F,H). The same kidney defect is present in another PSEN rescue line (17-3) (data not shown),and we therefore conclude that the phenotype is the result of the loss of presenilin expression rather than transgene integration.

As kidney organogenesis involves reciprocal inductive interactions between the ureteric bud epithelium and the metanephric mesenchyme, we performed in situ hybridization of various ductal and mesenchymal markers to delineate the origin of the defect. Analysis of Gdnf and Bmp7 showed that both molecules were expressed in PSEN-null kidneys similar to their littermate controls (data not shown), suggesting that the mesenchymal stem cells were preserved. To investigate the inductive processes, we analyzed the expression of Pax2, Pax8 and Wnt4. At E12.5, comparable levels of Pax2 in mesenchymal condensates were observed in PSEN-null kidney and the control (Fig. 3A, compare a with b), but expressions of Pax8 and Wnt4 were undetectable regardless of the genotype at this stage (not shown). At E13.5,levels of Pax2 (Fig. 3A, part f), Wnt4 (Fig. 3B,part b) and Pax8 (Fig. 3C, part b) were clearly detectable in mesenchymal derivatives,including aggregates (arrows), in PSEN-null kidney. However, there was reduced expression of these genes which is probably attributed to the lack of differentiation to comma- and S-shaped bodies(Dressler et al., 1990; Plachov et al., 1990; Stark et al., 1994). Examination of these molecules at E15.5 of development yielded similar results(data not shown).

Ret marks the tips of newly formed ureteric bud branches and its signaling is essential for inductive interactions(Pachnis et al., 1993). At E13.5, Ret expression was comparable in the control and the mutant(Fig. 3D, compare a with b),indicating that the primary branching was not affected by the loss of presenilins at this stage. As impaired nephrogenesis was unambiguously documented at this stage, this result suggests that the ureteric bud signaling is not the primary cause for the defect. The reduced number of Ret-positive tips in the mutant at E15.5(Fig. 3D, compare c with d)would thus probably be the result of impaired secondary or tertiary branching morphogenesis at this and later stages.

To exclude a possible contribution of stromal mesenchyme to the defects, we further evaluated stromal mesenchyme marker Bf2 expression and its levels were not significantly different in the mutants and controls (data not shown). Therefore, the combined data suggest that loss of presenilins does not disrupt the early inductive interactions and that the nephrogenic phenotype in PSEN-null kidney is probably due to a defect intrinsic to the mesenchyme.

The presence of mesenchymal condensates and aggregates but absence of comma- and S-shaped bodies in PSEN-null kidney prompted us to examine the mesenchyme to epithelium transition (MET). The neuronal cell adhesion molecule(NCAM) is highly expressed only in mesenchymal derivatives, but not in the ureteric buds. NCAM-positive structures (green) were readily detected in both the control and PSEN-null kidney at E14.5(Fig. 4A, parts a,b) and E16.5(Fig. 4A, parts c,d). The majority of NCAM-positive structures were also positive for pan-laminin(Fig. 4A, red), which is found in the basement membranes of both ductal and renal epithelia(Cho et al., 1998). These data suggest that the mesenchymal derivatives were competent for epithelization.

The proper expression of laminin A chain has been shown to be crucial for epithelial polarization (Ekblom et al.,1990; Klein et al.,1988). Polarized, laminin α1-positive renal vesicles were readily identifiable in PSEN-null kidney(Fig. 4Bb, arrow). The epithelial nature of these mesenchymal derivatives was further confirmed by their positive staining for E-cadherin (present in both ductal and renal epithelia, green) and negative for cytokeratin 8 (ductal epithelium only, red)(Fig. 4C). These results establish the notion that presenilins are not required for MET.

Detailed examination of NCAM-positive structures revealed that multiple types were identified, including laminin-low irregular aggregates(Fig. 4A, asterisks),laminin-positive organized renal vesicles with lumen(Fig. 4A, thin arrows), and elongated laminin-expressing tubules (Fig. 4A, part d, thick arrow), suggesting that the pretubular aggregates progressed into renal vesicles and tubules in PSEN-null mutant kidney. Consistent with this assessment, the E-cadherin-positive renal derivatives in PSEN-null mutant, similar to the littermate control, have formed immature tubular structures which were either in close proximity(Fig. 4C, part b, arrowhead) or appear to have connected with the duct (thick arrows in d and arrowhead in f, Fig. 4C). These characteristics suggest the formation of distal tubules. However, in contrast to the control in which advanced tubules with adjoining glomerular clefts could be recognized(Fig. 4C, part e), the mutant structures failed to undergo further differentiation and patterning.

To investigate a possible defect in proximal tubule development in the PSEN-null kidney, we performed staining with Lotus Tetragonolobus lectin(LTL), which is a specific marker for proximal tubules(Cho et al., 1998). Remarkably,although the control kidney showed strong staining with LTL(Fig. 5A, parts a,c), no positive staining could be detected with LTL in the mutant(Fig. 5A, part b,d). Immunostaining for WT1, which is highly expressed in the podocyte precursor cells of S-shaped bodies and mature glomeruli(Ryan et al., 1995), revealed a complete absence of WT1 expression in PSEN-null kidney as well(Fig. 5B). Thus, presenilins are absolutely required for the formation of proximal structures including proximal tubules and glomeruli.

To strengthen our in vivo findings, we next studied the kidney development in organ cultures. Control and PSEN-null kidneys were dissected from E12.5 embryos and cultured for 3 days. Whole-mount immunostaining with an anti-pan-cytokeratin antibody documented significant branching in both the control and PSEN-null kidneys after three days of culture(Fig. 5C), consistent with the notion that the presenilins do not exert an effect on primary branching morphogenesis. In agreement with our in vivo results, PSEN-null kidneys were devoid of WT1 (Fig. 5C, part b). Close up examination of the ureteric bud tips showed that, in PSEN-null culture, weak pan-cytokeratin positive renal tubules were formed and at least some appeared to have fused with the ureteric bud, which were strongly positive for pan-cytokeratin (Fig. 5C, part d). However, compared with the control(Fig. 5C, part c), the PSEN-null renal derivatives were truncated and were negative for WT1 expression (Fig. 5C, part d). Similar results were also obtained when the control kidney cultures were treated with a presenilin γ-secretase inhibitor (data not shown). These in vitro organ culture experiments thus lend support for an essential role of presenilins in the patterning and differentiation of renal epithelial derivatives.

Analysis of cells undergoing active division with an anti-phosphorylated histone H3 antibody showed that the cell proliferation profile was similar between the mutant kidney and the control at both E13.5 (data not shown) and E15.5 (Fig. 6A, parts a,b). Double staining with the anti-NCAM (Fig. 6A, parts c,d) and cytokeratin 8 antibodies (not shown) indicated that the proliferation rate of renal and ductal derivatives, respectively,were also similar. However, we cannot rule out the possibility that minor differences in cell proliferation in renal structures affected the tubule growth and patterning in presenilin mutant. This finding is not unexpected because much of the mesenchymal derivatives that required active cell division have already formed in the PSEN-null kidney.

Measurement of apoptosis using the TUNEL assay revealed that, at E13.5, the number of apoptotic cells was low in both the control and PSEN-null kidney,and no appreciable differences could be identified (data not shown). However,at E15.5, the number of apoptotic cells was significantly higher in the PSEN-null kidney when compared with the control(Fig. 6B, compare a with b). Quantitative analysis showed that only 0.93% of the cells was apoptotic in the control, while the percentage of apoptotic cells reached 6.34% in PSEN-null. The apoptotic cells were widespread with the majority localized to the peripherial mesenchyme (Fig. 6B, part b, arrowheads). The enhanced general apoptosis would contribute to the reduced kidney size in PSEN-null mutant.

Using co-immunostaining to detect the activated form of caspase 3 and NCAM(Fig. 6B, parts c,d), we were able to evaluate the apoptotic profile of the renal structures. Similar to the control, pretubular aggregates and renal vesicles in PSEN-null exhibited no caspase 3 activity. However, high caspase 3 staining was observed in NCAM-positive tubular structures in the PSEN-null kidney(Fig. 6Bd, arrow). Similar to the TUNEL staining, cortical mesenchyme also showed higher caspase 3 activation (Fig. 6B, part d,arrowheads). Overall, our results suggest that defective patterning of immature distal tubules in the absence of presenilins leads to the attenuation of nephrogenesis and enhanced apoptosis.

Because an impaired Notch signal transduction is considered the leading mechanism for the somite patterning defect of PSEN-null mice, we assessed the expression of Notch pathway molecules in E14.5 PSEN-null kidneys. First, we examined cell types exhibiting the presenilin γ-secretase activity by immunostaining with an antibody that recognizes a PSEN-cleaved and activated form of Notch1, the Notch intracellular domain (NICD)(Fig. 7A). Co-staining with an anti-cytokeratin 8 antibody (CK8) marked the ductal derivatives. NICD was not expressed above background in ureteric bud epithelium in both the control and PSEN-null samples (Fig. 7A). In the control kidney, NICD immunoreactivity could be detected in pretubular aggregates/renal vesicles (Fig. 7Aa), and comma- and S-shaped bodies (data not shown). PSEN-null kidney was devoid of NICD although pretubular aggregates were present(Fig. 7A, part b). Notch activation was correlated with expression of its downstream target Hesr1 (Hey1 – Mouse Genome Informatics)(Kokubo et al., 1999) in the control (Fig. 7B, part a). By contrast, Hesr1 expression cannot be detected in PSEN-null mutant(Fig. 7Bb), consistent with an obligatory role of presenilins in Notch processing and signaling. The residual signal probably results from staining of blood vessels that also express Hesr1. These results support the idea that a presenilin-mediated nephrogenic signal is derived from the mesenchyme, and that presenilin-dependent Notch activation and signaling is the mechanism in operation.

We next investigated Notch ligands Jag1 and Dll1expression by in situ hybridization. At E14.5, Jag1 was not expressed in condensed mesenchyme (Fig. 7C, part a,b, asterisks), but its expression was turned on in pretubular aggregates in both the PSEN-null kidney and the control(Fig. 7C, part a,b, arrows) and in comma-shaped bodies of the control kidney(Fig. 7C, part a, arrowhead). Interestingly, at E16.5, although Jag1 was strongly expressed in the control kidney (Fig. 7C, part c, arrowhead), its expression was almost undetectable in the mutant (Fig. 7C, part d). The loss of Jag1 is not caused by the absence of mesenchymal derivatives as NCAM-positive pretubular aggregates and renal vesicles could be readily identified in PSEN-null at this stage (Fig. 4A, part d). Therefore, this result suggests that Notch signaling is required for the maintenance of Jag1 expression during kidney organogenesis.

In situ hybridization analysis of another Notch ligand, Dll1,showed that, in contrast to Jag1, Dll1 was not expressed in condensed mesenchyme or pretubular aggregates. Dll1-positive staining could be seen in comma- and S-shaped bodies in the control kidney(Fig. 7D, part a, arrowheads),but no Dll1 expression could be detected in the PSEN-null at any stages of nephrogenesis (Fig. 7D, part b). This negative staining could be attributed by the lack of Dll induction caused by defective Notch signaling or by the absence of comma- and S-shaped bodies in PSEN-null mutant kidney. It is interesting to note that Dll1 expression seems to be restricted to a specific region of comma- and S-shaped bodies (Fig. 7D,part a, arrowheads), although the fate of these cells is not clear. Similar to NICD and Hesr1, both Notch ligands were not expressed in the ureteric bud, again supporting a cell autonomous mechanism of presenilin-dependent Notch signaling within the mesenchyme.

Histological examination at various stages of kidney development showed that presenilin deficiency does not affect the reciprocal inductive interactions between the ureteric bud epithelium and the metanephric mesenchyme. At E12.5, the mutant kidney was indistinguishable from that of littermate controls, as evidenced by the formation of condensed mesenchyme surrounding the ureteric buds. However, the difference became apparent at E13.5. While the control kidney undergoes further development to form pretubular aggregates, renal vesicles and comma- and S-shaped bodies, only pretubular aggregates and renal vesicles can be found in the PSEN-null kidney. In situ hybridization analyses of kidney development markers revealed that Gdnf and Bmp7 were normally expressed in PSEN-null kidney(data not shown), demonstrating that the uninduced mesenchyme was not affected by the loss of presenilins. The positive staining of Pax2, Pax8 and Wnt4 in the mutant established that the metanephric mesenchyme was competent in receiving signals from the ureteric bud to form condensates and aggregates. Last, the fact that Ret, which is a marker for the newly formed ureteric bud tips, was expressed similarly in the control and mutant at E13.5 suggests that presenilins do not exert a direct effect on primary branching morphogenesis. The reduced levels of mesenchymal markers Pax2,Pax8 and Wnt4 in the mutant at E13.5 and E15.5 are likely caused by the absence of comma- and S-shaped bodies. The combined effect led to impaired secondary and tertiary branching, as evidenced by the decreased number of Ret-positive tips in the mutant at E15.5. This notion is supported by kidney organ culture experiments in which loss of presenilins only affected branching in a minor fashion while formation of glomeruli were completely blocked. Limited secondary and tertiary branching and increased apoptotic cell death at later stages may in turn be the result of the failed progression through nephrogenesis in the PSEN-null kidney. These results,combined with our finding that presenilin γ-secretase activity is restricted to the mesenchymal derivatives, support a role for presenilins in the differentiation from pretubular aggregates and renal vesicles towards comma- and S-shaped bodies within the mesenchyme.

Progression from pretubular aggregates to comma- and S-shaped bodies requires the conversion and patterning of mesenchymal cells into highly organized and polarized epithelia. Analysis of cell adhesion molecules and extracellular matrix proteins, NCAM and E-cadherin, revealed that renal vesicles and tubular derivatives were partially formed in the PSEN-null kidney. Importantly, laminin stained these renal structures suggesting that they developed polarized cells with basement membranes. Thus, presenilins are not essential in the mesenchyme to epithelium transition. However, despite the formation and apparent fusion of some E-cadherin positive distal renal tubule structures to the bud epithelium, mature proximal tubules and glomeruli were completely absent in PSEN-null kidney. This conclusion was supported by the negative binding to LTL and lack of expression of WT1, respectively. From these data, we conclude that presenilins play an indispensable role in the patterning and differentiation of renal vesicles leading to proximal tubules and glomeruli.

In contrast to inductive interactions and MET, our current understanding about nephron patterning is limited. There are candidate regulatory mechanisms in tubule patterning by molecules such as the cadherins, which are differentially regulated in the various renal epithelial compartments(Cho et al., 1998). However,gene knockout studies have not supported a clear function of these molecules in nephron patterning (Dahl et al.,2002; Dressler,2002; Mah et al.,2000). The presenilins represent one of the first examples of molecules that play an essential role in patterning and differentiation of renal vesicle derivatives.

Presenilins are indispensable for the processing and signaling of Notch. This activity is highly conserved and likely accounts for the role of presenilins during somite patterning(Donoviel et al., 1999; Takahashi et al., 2000). Our analysis showed that the presenilin γ-secretase activated form of Notch1(NICD), Notch downstream target Hesr1 and ligands Dll1 and Jag1 were expressed in mesenchymal derivatives and that their expressions were critically impaired in PSEN-null kidneys. Thus, the Notch pathway, particularly signaling through Notch1, may be the underlying mediator for presenilin activity in kidney development. It is noteworthy that mice expressing a hypomorphic allele of Notch2 exhibit kidney glomerulogenesis defects (McCright et al.,2001). As all four mammalian Notch proteins are presenilin substrates in vitro (Saxena et al.,2001), a defective Notch2-mediated pathway may also account for the PSEN-null kidney phenotype. However, the presence of glomeruli in Notch2 mutant kidneys, albeit structurally abnormal, indicates that Notch2 functions downstream of tubule patterning (McCright et al., 2001).

Although an essential role for the presenilins in Notch signaling has been well established, how this regulation affects Notch ligand expression has been the subject of controversy. Some support a lateral inhibition model in which loss of Notch signaling resulting from presenilin deficiency leads to ectopic overexpression of the ligand (Donoviel et al., 1999; Handler et al.,2000); Others suggest that presenilin-mediated Notch signaling is necessary for proper induction of its ligand(Takahashi et al., 2000). The differences may be caused in part by variations of the systems of study; as ligand expression seems to be subjected to strict temporal-, spatial- and cell type-specific regulation.

In kidney development, detailed examination of expression patterns of Jag1 and Dll1 yielded some interesting findings.

(1) Jag1 is not expressed in condensed mesenchyme but its expression can be detected in pretubular aggregates and renal vesicles. This distinct Jag1 staining pattern was similar in the control and the PSEN-null kidneys at E14.5, suggesting that transition from condensed mesenchyme to pretubular aggregates does not depend on presenilins. Interestingly, although pretubular aggregates and renal vesicles still exist in PSEN-null mutant at E16.5, Jag1 expression is lost. These data suggest that sustained Notch activation is necessary for the maintenance of Jag1 expression during nephrogenesis.

(2) In contrast to Jag1, Dll1 message cannot be detected in pretubular aggregates. In the control kidney, Dll1 expression seems to be restricted to a specific region of the comma- and S-shaped bodies at E14.5. Although the origin of these cells is not known, this spatiotemporally restricted expression pattern may imply functional significance. Indeed,during somite patterning, caudally restricted expression of Dll1 has been shown to be important for prefiguring somite identity, and proper Dll1 induction requires presenilin-dependent Notch signaling(Takahashi et al., 2000). Equally possible, however, the absence of Dll1 expression could also be due to the physical lack of comma- and S-shaped bodies in the PSEN-null kidney.

Besides Notch and APP, numerous other proteins have been reported as substrates of presenilin-dependent proteolysis(Lammich et al., 2002; Marambaud et al., 2002; Ni et al., 2001). Although the physiological significance is not clear, impaired processing of these molecules cannot be excluded as a potential mechanism for the kidney defects. Through processing-independent mechanisms, PSEN1 has been shown to interact with β-catenin and to facilitate its turnover(Kang et al., 2002; Xia et al., 2002). Therefore,it is possible that deregulation of the β-catenin pathway in the PSEN-null may contribute to the kidney phenotype. The Wnt pathway has been implicated in the inductive phase of kidney development(Dressler, 2002), but is essentially unaffected in the PSEN-null. In addition, we have created mice with a deletion of exon 10 of the endogenous Psen1 essential forβ-catenin interaction. Homozygous exon 10-deleted mice, when crossed onto the Psen2-null background, are viable and do not exhibit kidney defects (H.Z., unpublished). These observations thus argue against a potentialβ-catenin involvement in the PSEN-null kidney.

In addition to the kidney patterning defect, Psen1- and Psen2-double null embryos display other abnormalities, most noticeably defects in heart looping and vascular remodeling(Donoviel et al., 1999; Herreman et al., 1999). These defects were apparently corrected by the human PSEN1 transgene presumably because of its early expression in mesoderm and in relevant cell derivatives. Indeed, the human Thy-1 promoter has been shown to be active in endothelial cells (Gordon et al.,1987), and expression of Hesr1, although absent in developing nephrons of the PSEN-null kidney, can be detected in blood vessels. The analysis of mechanisms leading to these phenotypes is currently ongoing.

In summary, using our novel presenilin `rescue' system, we have identified a novel function of the presenilins in nephrogenesis. Specifically,presenilins are indispensable for the patterning of renal epithelial structures to form mature proximal tubules and glomeruli. Loss of presenilins is associated with failed progression from renal vesicles to comma- and S-shaped bodies. On the molecular level, we reveal an obligatory role of presenilins in the activation of Notch signaling and maintenance of the Notch ligand Jag1 expression in the mesenchymal derivatives during kidney development.

The probes and antibody were mainly donated by R. Johnson (Dll1, Jag1 and Hesr1), Y. Furuta (Bpm7 and Bf2), B. Hogan (Gdnf), F. Constantini (Ret) and D. Abrahamson (anti-laminin α-1 antibody). M. Shearman and H. Lewis provided us with the presenilin γ-secretase inhibitor. We are grateful to R. Kopan and H.-T. Cheng for their advice on NICD immunostaining, and for sharing their results prior to publication. We thank T. Zaidi, X. Chen and Z. Yu for expert technical support, and J. Xu for image assistance. Valuable inputs from J. Miner, R. Johnson, J. Rosen, R. Davis, S. Tsai, M. Justice and X. Wu are greatly appreciated. This work was supported by grants from NIH(NS40039 and AG20670) and Alzheimer's Association (IIRG-00-221). H.Z. is a New Scholar of the Ellison Medical Foundation.