|Home | About | Journals | Submit | Contact Us | Français|
Notch signaling is essential for proper lens development, however the specific requirements of individual Notch receptors have not been investigated. Here we report the lens phenotypes of Notch2 conditionally mutant mice, which exhibited severe microphthalmia, reduced pupillary openings, disrupted fiber cell morphology, eventual loss of the anterior epithelium, fiber cell dysgenesis, denucleation defects, and cataracts. Notch2 mutants also had persistent lens stalks as early as E11.5, and aberrant DNA synthesis in the fiber cell compartment by E14.5. Gene expression analyses showed that upon loss of Notch2, there were elevated levels of the cell cycle regulators Cdkn1a (p21Cip1), Ccnd2 (CyclinD2), and Trp63 (p63) that negatively regulates Wnt signaling, plus down-regulation of Cdh1 (E-Cadherin). Removal of Notch2 also resulted in an increased proportion of fiber cells, as was found in Rbpj and Jag1 conditional mutant lenses. However, Notch2 is not required for AEL proliferation, suggesting that a different receptor regulates this process. We found that Notch2 normally blocks lens progenitor cell death. Overall, we conclude that Notch2-mediated signaling regulates lens morphogenesis, apoptosis, cell cycle withdrawal, and secondary fiber cell differentiation.
The ocular lens develops from invagination of the lens placode within the surface ectoderm, which transforms into a lens vesicle by E10.5 in mice (reviewed in Lovicu et al., 2011). Cells in the posterior vesicle then elongate and terminally differentiate to form lens fiber cells, while the anterior cells maintain a cuboidal epithelial cellular morphology and continue proliferating. These anterior lens progenitor cells constitute the lens growth zone, termed the anterior epithelial layer (AEL). Several growth factor and signaling pathways have been implicated in this process, including Fgf, Bmp, Wnt (Belecky-Adams et al., 2002; Boswell et al., 2008; Cain et al., 2008; Chen et al., 2006; Chen et al., 2008; Lovicu and McAvoy, 2005; McAvoy et al., 1999; McAvoy et al., 1991; Stump et al., 2003). The transcription factor, Foxe3 is initially expressed in all lens vesicle cells, but becomes restricted to AEL by E12.5, along with Cdh1 (E-Cadherin) (Blixt et al., 2000; Brownell et al., 2000; Medina-Martinez et al., 2005; Xu et al., 2002). Thereafter, lens growth and development requires precise control of proliferation and differentiation within the AEL, which contains the progenitor pool for secondary fiber cells that make up bulk of the lens. Differentiated fiber cells express Cdh2 (N-Cadherin), cMaf, Prox1, and Cryb, Cryg (β-, γ-Crystallin)(reviewed in Lovicu et al., 2011). In addition, Cdkn1c (p57Kip2) is expressed at the early stages of fiber differentiation and serves as a useful marker for the initiation of fiber differentiation (Lovicu and McAvoy, 1999).
Notch signaling is a highly conserved, cell-cell signaling pathway that regulates cell fate determination during development (reviewed in Fortini, 2009; Kopan and Ilagan, 2009). Key components of this pathway include receptors (Notch1-4) and ligands (Deltalike1,3,4, Jag1,2), which are transmembrane proteins with large extracellular domains. During canonical Notch signaling, ligand binding activates a particular receptor, which undergoes proteolytic cleavage, leading to the release of the C-terminal Notch intracellular domain (NICD). The NICD translocates to the nucleus, where it forms a transcriptional complex with the DNA binding protein Rbpj (RBPJκ), and its co-activator Maml (Mastermind), leading to the activation of target genes. Genes activated by Notch signaling include the Hes and Hey (Herp) family of transcription factors. The Notch pathway has a wide variety of functions in both developing and adult tissues. In the developing mouse eye, multiple Notch pathway genes are expressed, including Notch1, Notch2, Notch3, Jag1, Dll1, Rbpj and the effector Hes1 (Bao and Cepko, 1997; Bettenhausen et al., 1995; Ishibashi et al., 1995; Jia et al., 2007; Le et al., 2009; Rowan et al., 2008; Weinmaster et al., 1991). The roles of Notch signaling during lens development and fiber differentiation have only just begun to be understood. Previous studies of Rbpj in the lens identified distinct functions for canonical Notch signaling in eye morphogenesis, lens progenitor cell proliferation, transit through the transition zone, and fiber cell differentiation (Jia et al., 2007; Rowan et al., 2008). In addition, conditional deletion of either Rbpj (Jia et al., 2007; Rowan et al., 2008) or Jag1 (Le et al., 2009), suggested cell cycle regulatory genes like Ccnd1 (CyclinD1), Ccnd2 (CyclinD2), Cdkn1b (p27Kip1) and/or Cdkn1C (p57Kip2), may be regulated by Notch signaling during the decision of lens progenitor cells to divide further versus differentiate.
While these studies clearly demonstrated the importance of Notch signaling in the developing lens, it is unknown when and where specific receptors act, including whether they exhibit distinct functions. The first suggestion of a specific role was provided by a global gene profiling study comparing mRNA expression between human lens epithelial cells and cortical fiber cells, which found that Notch2 expression was significantly higher in the epithelium (Hawse et al., 2005). In vitro studies of postnatal rat lens epithelial differentiation demonstrated that Notch2 signaling is activated during FGF-dependent secondary fiber cell differentiation (Saravanamuthu et al., 2009). No corresponding activation of Notch1 was detected, but the unique requirements of Notch2 were not investigated further. In this study we used the Le-Cre driver to conditionally delete the Notch2 receptor during mouse lens development. Jag1 and Rbpj were previously shown to act during lens vesicle morphogenesis, AEL progenitor cell proliferation and fiber cell differentiation. Here, our results indicate that although the lens-specific loss of Notch2 phenocopies the absence of Rbpj or Jag1 during vesicle morphogenesis and fiber cell differentiation, Notch2 is not required for proliferation in the lens. Instead, we find that this receptor uniquely blocks lens cell apoptosis.
Notch2CKO/CKO mice were obtained from Tom Gridley and genotyped using published protocols (McCright et al., 2006). Throughout this paper the abbreviation CKO indicates a “Conditional Knockout Allele”. Le-Cre mice were obtained from Dr. Joram Piatigorsky and Ruth Ashery-Padan, and genotyped according to published protocols (Ashery-Padan et al., 2000). All mice were housed and cared for in accordance with the guidelines provided by the National Institutes of Health, Bethesda, MD and the Association for Research in Vision and Ophthalmology.
Embryonic and postnatal tissues were fixed in 4% paraformaldehyde/PBS for 1–2 hours at 4°C, processed through a sucrose/PBS series, cryoembedded and 10 micron sections generated. Primary antibodies used were anti-BrdU (Serotec clone BU1/75 1:100), anti-cPARP (Cell Signaling 1:500), anti-Ccnd2 (Santa Cruz 34B1-3 1:200), anti-Cdh1(Zymed ECCD-2 1:500), anti-Foxe3 (gift from Peter Carlsson 1:1000), anti-Cryb (gift from Richard Lang 1:8000), anti-Crya (gift from Eric Wawrousek 1:1000), anti-Cryg (Santa Cruz SC22415, 1:1000), anti-GFP (Molecular Probes 1:1000), anti-Hes1 (1:1000), anti-Jag1 (Santa Cruz 1:1000), anti-Cdkn1c (Abcam 1:200), anti-Prox1 (Covance 1:6000), anti-Cdh2 (BD Transduction Laboratories, 1:100), and anti-cMaf (Santa Cruz 1:200). Secondary antibodies were directly conjugated to Alexa Fluor 488, Alexa Fluor 594 (Invitrogen) or biotinylated (Jackson Immunologicals) and sequentially labeled with Alexa 488- or 594-Streptavidin (Invitrogen), followed by DAPI nuclear labeling. Microscopic imaging was performed on a Zeiss fluorescent microscope with a Zeiss camera and Apotome deconvolution device. For S-phase analyses, BrdU (Sigma) was injected intraperitoneally and animals sacrificed 1.5 hours later for tissue processing, following the method of (Mastick and Andrews, 2001). Standard histology on paraffin embedded sections was also performed. Images were processed using Axiovision (v7.0) and Adobe Photoshop (CS4) software and electronically adjusted for brightness, contrast and pseudocoloring.
Tissue sections, separated by at least 60 μm, were antibody-stained and counted using NIH ImageJ or Axiovision software. At least two independent sections from each animal, using 3 or more mice per genotype and age were quantified. Labeling indices for BrdU+, Foxe3-neg or cPARP+ cells were generated by dividing the number of antibody-positive cells by total DAPI-labeled nuclei, and compared against one other genotype using a two-tailed, unpaired Student T-Test, to determine p values.
Lens were removed from E14.5 mice and fixed in 2% glutaraldehyde and 2% paraformaldehyde in 0.1 M sodium phosphate buffer (PB), pH 7.4 for 12 h at room temperature (RT). After a wash with rinsing buffer (RB, 4% sucrose and 0.15 mM CaCl2 in PB), pH 7.4 at 4°C, tissues were postfixed in 1% OsO4 in 0.1 M PB, pH 7.4 for 1 h. After rinsing and dehydration, tissues were embedded in Epon 812 for 72 h at 60°C. One micron semi-sections were used for tissue orientation. Then 70–90 nm ultrasections were collected in 200 mesh grids and counterstained with 5% uranyl acetate and 0.3% lead citrate. Sections were viewed on a JEOL 1010EM at 60 KV and digital images were acquired at 2000–10,000X magnifications by AMT software (Advanced Microscopy Techniques, Corp.).
In situ hybridization was performed on frozen mouse embryonic sections. RNA probe labeling and in situ hybridization were performed essentially as described (Rosen and Beddington, 1993). Embryos were snap frozen, cryosectioned, post-fixed, washed with PBS, prehybridized and then hybridized overnight at 66°C. Post-hybridization steps were carried out as described (Rosen and Beddington, 1993). The template was a 554bp region in the UTR of Notch2 (NCBI NM_010928) similar to the one described for rat Notch2 in (Lindsell et al., 1996). The template was amplified such that the sequence for the T7 bacteriophage promoter was built into the reverse primer sequence (Forward: 5′-GGGCCCGGAAYYCTCCACCTGCATTGACCGCGTGGCC-3′, Reverse: 5′-GGGCCCGGAATTCGTCATCAATGTCGATCTGGCACACTGGTCC-3′).
Intact lenses from E14.5 embryos were isolated into M199 medium. Medium was aspirated slowly to resuspend them in 1 ml of HBSS (Ca++ and Mg++ free). HBSS was slowly aspirated off and 100ul of collagenase solution was added. The collagenase digestion was carried out at 37°C using 3.4 U collagenase H (Sigma) in HBSS with 10% FCS. After collagenase digestion, the lenses were homogenized and resuspended using 200ul pipette tip. Then 1ml of PBS was added and the cells were washed with PBS by spinning at 1400rpm at 4°C for 5 min. Adult lens cells were passed through 100um strainer before washing.
The cell pellet was resuspended with pulse vortex by adding 1 ml of freshly prepared Fixation/Permeabilization working solution (E Biosciences). They were incubated at 4°C for 18 hours in the dark, followed by washing with 2 ml 1X Permeabilization Buffer. After decanting the supernatant the cells were blocked with 5% FCS in 1X Permeabilization Buffer, (Ebiosciences) in approximately 100 μl volume, at 4°C for 15 minutes. This was followed by washing cells with 2 ml 1X Permeabilization Buffer. Then the antibodies were added at required working concentrations in 1X Permeabilization buffer and the cells were incubated in the antibody solution at 4°C for at least 30 minutes in the dark. This was followed by washing the cells with 2 ml 1X Permeabilization Buffer. Then they were resuspended in 300ul volume PBS/BSA (0.5%) buffer and analyzed on FACS Calibur (BD Biosciences) cytometer. For cell counting 50ul (50,000 beads) of Countbright beads (Invitrogen) were added just before acquisition in the flow cytometer.
Intact lenses were dissected from three Notch2CKO/CKO controls and three Le-Cre; Notch2CKO/CKO mutant mice at E19.5 and the two lenses of every animal were pooled together. Total RNA was isolated using Trizol reagent (Invitrogen). Microarray data was collected at Expression Analysis, Inc. (www.expressionanalysis.com; Durham, NC). Before target production, the quality and quantity of each RNA sample was assessed using a 2100 BioAnalyzer (Agilent). Target was prepared and hybridized according to the “GeneChipR 3′ IVT Express Kit User Manual” using the reagents provided in the GeneChipR 3′ IVT Express Kit (Affymetrix Part#901228). The microarrays were then stained with Streptavidin Phycoerythrin and the fluorescent signal was amplified using a biotinylated antibody solution. Fluorescent images were detected in a GeneChip® Scanner 3000 and expression data was extracted using the GeneChip Operating System v 1.1 (Affymetrix). All GeneChips were scaled to a median intensity setting of 500. An estimate of signal for each transcript was calculated using the Microarray Suite Algorithm version 5.0 (Affymetrix) using the Expression Console version 1.1 (Affymetrix). The values of individual probes belonging to one probe set were averaged and normalized using GeneSpring v11.0.2 (Agilent Technologies Inc., Palo Alto, CA, USA). The average fluorescence intensity of all annotated gene was calculated using Robust Multi-array Analysis (RMA) algorithm. In order to identify differentially expressed genes between the two conditions, a one-way Analysis of Variance (ANOVA) was performed. Only genes with an uncorrected p-value less than 0.01 were used, ending up with 1299 probes. From this set of genes, 44 probes (32 genes) had a greater than 3-fold change. (GEO Accession Number GSE31643).
RT-qPCR analysis was performed using SABiosciences/Qiagen qPCR arrays using reagents supplied by the manufacturer. Total RNA was isolated from E19.5 lenses using RNeasy Micro kit (Qiagen). RT-qPCR was performed using qPCR arrays PAMM-043A (Wnt pathway), PAMM-020 (Cell cycle pathway) and PAMM-059 (Notch pathway) purchased from SA Biosciences, Qiagen according to manufacturer’s instructions. A 7900 HT real time PCR instrument (Applied Biosystems) and SDS2.3 software were used to collect run data and threshold cycle (Ct) values of individual reactions. Further data analysis was carried out using Microsoft Excel based worksheets or online data analysis software found at www.sabiosciences.com.
Previous studies reported that Notch receptors are expressed in the lens, ciliary body, RPE and retina of embryonic mice (Bao and Cepko, 1997; Weinmaster et al., 1991), and Notch2 in particular, is present in the newborn rat lens epithelium (Saravanamuthu et al., 2009). However, the spatiotemporal expression pattern of Notch2 in the developing rodent lens has not been systematically examined. To better understand the potential role of Notch2 during lens development, we examined mRNA expression at various stages of lens development by in situ hybridization. At embryonic day 11.5 (E11.5) expression of Notch2 is mostly localized within the retinal pigmented epithelium (RPE) (Fig. 1A). At E13.5 Notch2 expression is detected in the lens vesicle, predominantly in the anterior epithelial layer (AEL) (Fig. 1B). By E16.5, Notch2 mRNA is present in both the AEL and transition zone (TZ), around the lens equator (Fig. 1C), as well as in the RPE. At postnatal day 3 (P3), Notch2 lens expression remains confined to the AEL and transition zone (Fig. 1D). The corneal epithelium shows no detectable levels of Notch2 mRNA while non-neuronal eye cup derivatives, including the RPE, iris, and ciliary tissue, also express Notch2, consistent with previous reports (Bao and Cepko, 1997). In addition, we also observed Notch2 expression in skin hair follicles of the eyelids, as previously reported (Schouwey et al., 2007).
To remove Notch2 in the developing mouse lens, we used Le-Cre mediated recombination of a Notch2 floxed allele (Ashery-Padan et al., 2000; McCright et al., 2006) (Fig. 2A). Adult Le-Cre; Notch2CKO/CKO mice showed severe microphthalmia, accompanied by reduced or absent pupillary openings (Fig. 2B-E) compared to Notch2CKO/CKO controls (n = 3). Since the Le-Cre transgene has an IRES-GFP expression cassette, GFP fluorescence serves as an indicator of Cre expression. Fig. 2G shows specific GFP expression in E11.5 Le-Cre; Notch2CKO/CKO lens vesicles, compared to background fluorescence in littermate Notch2CKO/CKO controls (Fig. 2F).
Histological analysis of 10 weeks old adult Le-Cre; Notch2CKO/CKO eyes showed a severely deformed lens with extensive vacuoles, unlike the control lenses (Fig. 3A,B). Heterozygous Le-Cre; Notch2CKO/+ eyes had no obvious phenotype (Supplementary Fig. S1). Importantly, the adjacent cornea, which does not normally express Notch2 mRNA (Fig. 1D), was unaffected in Le-Cre; Notch2CKO/CKO (Fig. 3C,D). However, in addition to an abnormal AEL, Le-Cre; Notch2CKO/CKO mutants also displayed fused irises and defective pupillary eye openings (Fig. 3E,F). Defects in the transition zone of the Le-Cre; Notch2CKO/CKO lenses include abnormal distribution of nuclei, globular fiber morphology, inappropriate nucleated fiber cells along the posterior capsule defective fiber cell elongation (Fig. 3G–J).
To define better the onset of defects in the lens, due to loss of Notch2, we compared the morphology of Notch2CKO/CKO controls and Le-Cre; Notch2CKO/CKO mutants, during prenatal development. H&E staining of histological sections of E12.5 Le-Cre; Notch2CKO/CKO mutant embryonic eyes showed a reduced lens vesicle, with incomplete separation from the surface ectoderm (Fig. 4A,B). This defective separation resulted in a persistent lens stalk, as highlighted by colabeling for Foxe3 (red) and Cdh1 (E-Cadherin, green) (Fig 4C,D see insets). Immunofluorescence staining of the Notch effector, Hes1, demonstrated that this protein has been downregulated, compared to Notch2CKO/CKO controls (Fig. 4E,F). Primary fiber cell elongation appeared normal and elongating cells expressed characteristic proteins, such as Jag1 (Fig. 4G,H), Cryb (β-Crystallin) (Fig. 4I,J), Ccnd2 and cMaf (Fig 4 K,L), supporting the idea that Notch2 has little if any role in primary fiber genesis.
At E14.5 Le-Cre; Notch2CKO/CKO lenses were distinctly microphthalmic (Fig 5A,G). Both the extent and thickness of the AEL were reduced and nuclei of the peripheral lens fibers failed to form a pronounced “bow”, suggesting a possible defect in elongation of the secondary fiber cells (Fig. 5A,B,G,H). Immunostaining for the Notch ligand, Jag1, was detected along the surface membranes of newly formed secondary fibers of both Notch2CKO/CKO control and Le-Cre; Notch2CKO/CKO mutant lenses; however, mutant secondary fiber cells were small and poorly elongated (Fig. 5C,I). Expression of the Notch effector Hes1 was detected in only a few, isolated cells in the AEL of Le-Cre; Notch2CKO/CKO mutant lenses (arrows in Fig. 5D,J). The anterior lens epithelium (AEL) marker Foxe3 was detected in most AEL cells, although occasional Foxe3-negative cells were observed (Fig. 5E,K). Ccnd2, which is normally upregulated in the transition zone (Fromm and Overbeek, 1996), was appropriately expressed in the corresponding cells of the Le-Cre; Notch2CKO/CKO mutant (Fig. 5E,K). Cdh1 expression was normal in the AEL of the Le-Cre; Notch2CKO/CKO mutant and properly downregulated in the elongating fibers (Fig. 5F,L). Moreover, Cdh1 expression in the AEL illustrated a reduced AEL thickness, as well as the persistence of a lens stalk (arrow in L). The transcription factor Prox1, like Jag1 and Ccnd2, is normally upregulated in the transition zone and differentiating fibers (Fromm and Overbeek, 1996; Le et al., 2009; Wigle et al., 1999), was also expressed appropriately in nuclei of both the primary and secondary fiber cells (Fig 5F,L). The loss of Notch2 did not affect the onset of the differentiation markers assayed, within the posterior fiber cell compartment. However, subsequent lens fiber cell morphology, particularly elongation, was abnormal.
To determine whether conditional deletion of Notch2 reduces AEL proliferation, as has been reported for Rbpj and Jag1 mutant lenses (Jia et al., 2007; Le et al., 2009; Rowan et al., 2008), we BrdU pulse-labeled embryonic lenses at E11.5 and E14.5. Interestingly, there was no significant difference in the percentage of BrdU+ S-phase cells in Le-Cre;Notch2CKO/CKO and Notch2CKO/CKO lenses, at either age (Figs. 6A,B,E), indicating that Notch2 is not required for AEL proliferation. In addition, a number of S-phase nuclei were apparent in the fiber cell compartment of Notch2 mutant lenses, suggesting a defect in cell cycle withdrawal during differentiation (arrow in Fig. 6B). Since BrdU+ cells are occasionally found in the posterior compartment of the normal prenatal mouse lens, we carefully quantified the number of BrdU+ fiber cells (that were Cdh1 negative) in multiple sections of Notch2CKO/CKO and Le-Cre;Notch2CKO/CKO lenses. The results showed a significantly larger proportion of BrdU+ fiber cells in the absence of Notch2 (Fig. 6F), suggesting that Notch2 normally promotes cell cycle withdrawal during secondary fiber cell formation.
To investigate the cell cycle status of the differentiating secondary fibers further, we used flow cytometry to identify cells in the early stages of secondary fiber cell formation. Such cells can be recognized by the progressive loss of Cdh1, which indicates the onset of fiber differentiation (Saravanamuthu et al., 2009; Xu et al., 2002). Fluorescent Activated Cell Sorting (FACS) of E14.5 lens cell preparations was used to enrich Cdh1+ lens epithelial cells and identified about 55 – 70% (n=3 experiments) of the cells as the Cdh1+ population (compare Figs. 7A and 7B, data not shown). The increased sensitivity of this FACS assay distinguished two sub-populations of Cdh1+ cells, one showing high expression, the other with low expression (Fig. 7C). The high Cdh1+ cells had 15-fold higher fluorescence intensity than the low Cdh1+ cells, and were represented by two distinct peaks of fluorescence intensity. To confirm that the Cdh1+ cells were derived from the lens, two populations of cells were also tested for the lens specific marker, Crya (α-Crystallin). The results demonstrated that all the Cdh1+ cells from the embryonic lens preparation were all also positive for lens marker Crya, thereby confirming that they derived from the lens, and not contaminating tissues (data not shown). Forward scatter parameter in the FACS assay distinguishes cells based on their size, and to lesser extent cell shape. Forward scatter of the high expresser Cdh1+ population indicated a larger, more rounded cell profile typical of epithelial cells. In contrast, the forward scatter of the low expressers indicated a smaller size and likely flattened cell shape, consistent with the cell elongation that occurs in the early stages of fiber cell differentiation (Fig 7D,E). Thus, we consider the low expresser Cdh1+ population to represent cells that have begun to lose Cdh1 expression and elongate at the lens equator. Examining this low Cdh1+ population for BrdU labeling revealed that Le-Cre; Notch2CKO/CKO mutant lenses harbored a significantly higher percentage of BrdU+ cells (Fig. 7F). This supports the view that Notch2 deletion leads to aberrant cell cycle withdrawal during differentiation.
Lenses deficient in Rbpj or Jag1 showed an excess of differentiated fiber cells, as seen by a larger than normal percentage of Foxe3-negative cells at E11, E12.5, and E14.5 (Le et al., 2009; Rowan et al., 2008). To examine whether Le-Cre; Notch2CKO/CKO mutant lenses exhibit the same phenotype, we quantified the percentage of Foxe3-negative cells at E11.5 and E14.5 (Fig. 6G). There was no significant difference in this population at E11.5, but at E14.5 Notch2 lens mutants had a significant increase in the proportion of fiber cells. Paradoxically, there was no corresponding increase in cell proliferation during this period, suggesting that the increase in fiber cells might be balanced by a loss of AEL cells through apoptosis in the absence of Notch2. To test this possibility, the percentage of cPARP+ (caspase 3 cleavage product of polyADP-ribose polymerase) AEL cells was quantified (Figs 6C,D,H). The results showed that 1.1% of AEL cells were apoptotic in Le-Cre; Notch2CKO/CKO lenses, compared to zero in Notch2CKO/CKO controls. Thus, Le-Cre; Notch2CKO/CKO mutant lenses have a significantly higher rate of apoptosis in the AEL.
At older ages the AEL of Le-Cre; Notch2CKO/CKO mutants was even further reduced in size, as revealed by Cryb (Fig. 8A,F) Foxe3 (Fig. 8B,G) or Cdh1 (Fig. 8C,H,E,J) expression. Moreover, few cells were positively stained for the early differentiation marker Cdkn1c (Fig. 8E,J)(Zhang et al., 1998), indicating a defective transition zone in the Le-Cre; Notch2CKO/CKO mutant lenses. However, Cryb immunofluorescence revealed a loss of parallel arrangement of fibers and defective fiber elongation in the mutant (Fig. 8F). Late expression of Jag1 and Cdh2, were abnormal in Le-Cre; Notch2CKO/CKO mutant lenses, with only a few positively stained cells observable (Fig. 8B,G and 8D,I). As further evidence of fiber cell disorder, staining for Prox1 revealed fiber disorders in the Notch2 mutants (Fig. 8C,H).
To identify specific genes that may be affected by Notch2 deletion, we compared gene expression profiles across three biological replicates of E19.5 Notch2CKO/CKO (control) and Le-Cre; Notch2CKO/CKO (mutant) lenses, using Affymetrix mouse genome 430 2.0 arrays. Forty-four transcripts were identified as being elevated or reduced by at least 3-fold (as compared to controls) with p-values<0.001. A heat map distribution of fold change patterns of the top differentially expressed genes is shown in Fig. 9. A complete list of identified transcripts corresponding to these genes and their fold changes is provided in Supplemental Table 1. Genes that were significantly upregulated in the Le-Cre; Notch2CKO/CKO mutant lenses included Fgf12, Fgf15, which is noteworthy because multiple Fgf factors induce lens fiber cell differentiation; Cdkn1a (p21Cip1), which is involved in cell cycle control (Xiong et al., 1993); and two genes involved in thyroid hormone signaling, the thyroid binding protein Crym (μ-Crystallin) (Finckh et al., 1998), and the thyroid hormone responsive spot 14, Thrsp (Zhu et al., 2005; Zhu et al., 2001). Downregulated genes included the epithelial cell junctional protein, Cdh1, Olfml3 (olfactomedin-like3), which has been implicated in Bmp regulation (Inomata et al., 2008), Dnase2b, which is involved in fiber cell denucleation (Nishimoto et al., 2003; Torriglia et al., 1995), and the Pdgfra receptor, which regulates AEL cell proliferation (Kok et al., 2002).
As an independent approach to identifying gene expression differences between Le-Cre; Notch2CKO/CKO mutant lenses and Notch2CKO/CKO control lenses, we used real time qPCR arrays focused on genes associated with proliferation and differentiation. These included genes in the Notch and Wnt signaling pathways, and cell cycle regulatory genes. Expression of 252 genes involved in one or more of these three pathways were analyzed. The qPCR outcomes for genes with at least 1.5-fold differential expression between control and Notch2 conditional mutant lenses (p value <0.05) are listed in Table 1. The most strongly upregulated gene identified by this method was the transformation related protein Trp63. Other upregulated genes included cell cycle regulatory genes such as Ccnd2, and three cyclin-dependent kinase inhibitors, Cdkn1a (p21Cip1), Cdkn1c (p57Kip2), and Cdkn2a (p16Ink4). Multiple Wnt pathway genes were also upregulated, including Wnt16 and Wnt8b; Wnt inhibitors Dkk1, Sfrp4; the Wnt receptor Fzd8; a Wnt pathway antagonist, Pparg (PPAR-gamma) and Wnt target genes Tcf7, Tle1, Tle2, and Wisp1. A number of genes were also downregulated, including Wnt4, Wnt7b, and the secreted frizzled-related protein, Sfrp1, which inhibits Wnt signaling, as well as Cdh1. These changes in gene expression provide candidates that may play a role in specific features of the Le-Cre; Notch2CKO/CKO mutant phenotype, such as loss of AEL, defective cell cycle withdrawal, and aberrant fiber cell differentiation.
Our findings show that Notch2 function is essential for prenatal lens development and differentiation. Conditional deletion of Notch2 causes a persistent lens stalk, followed by a progressive loss of the AEL and transition zone, disruption of secondary fiber cell differentiation, and apoptosis. Although in the absence of Notch2, fiber cell markers had normal spatiotemporal expression patterns, we found that secondary lens fibers failed to elongate, and properly withdraw from the cell cycle.
At the outset, we predicted that Notch2 lens mutants would have reduced AEL cell proliferation, like Jag1 and Rbpj conditional mutants. This was not the case, although Notch2 embryonic mutant lenses are smaller than controls. Instead, we found aberrant S-phase fiber cells. Our FACS analysis confirmed that Notch2 mutant lenses have a significant increase in a low Cdh1-expressing population, which are initiating fiber cell differentiation, yet also inappropriately contain more BrdU+ cells. This could occur if Notch2 mutant cells undergo longer cell cycles, or are unable to fully exit mitosis. We favor the former idea, since mutant fiber cells activate the expression of multiple postmitotic and terminal differentiation markers at the appropriate time and place.
Proper lens vesicle closure and separation from the overlying ectoderm depends on the Foxe3 transcription factor (Blixt et al., 2000; Brownell et al., 2000; Medina-Martinez et al., 2005; Medina-Martinez and Jamrich, 2007). Foxe3 is expressed in all lens vesicle cells until E12, when it becomes restricted to the AEL. Foxe3 mutant lenses are microphthalmic with persistent lens stalks, like Notch2 conditional mutants. Thus, Notch signaling and Foxe3 act similarly in the lens, further supported by a significant loss of Foxe3+ AEL cells in E14.5 Notch2 mutants. Foxe3 mutant lenses have reduced Dnase2b expression, which regulates fiber cell denucleation (Counis et al., 1998; Torriglia et al., 1995), and Pdgfra, which promotes AEL proliferation (Kok et al., 2002). Interestingly both of these genes were downregulated in the Le-Cre; Notch2CKO/CKO mutant lenses. Although this might suggest that Foxe3 is regulated by a Jag1-Notch2-Rbpj signal, this issue is unresolved, since Foxe3 initiates normally in the lens of each Notch pathway mutant. Alternatively, Foxe3 lens expression requires activation of Smad8 (Yoshimoto et al., 2005) which could theoretically interact with the Notch2 intracellular domain (N2-ICD), for noncanonical Notch functions, as was reported for N1-ICD and Smad3 (Blokzijl et al., 2003). In addition, we found that Le-Cre; Notch2CKO/CKO lenses have depressed levels of Olfml-3, a protein that promotes Bmp signaling in Xenopus (Inomata et al., 2008). Reduced levels of Olfml-3 would be expected to reduce Bmp signaling, Smad8 activation, and Bmp-dependent Foxe3 expression.
However, none of these gene expression changes can fully account for the small lens phenotype of Notch2 conditional mutants. Instead, we found Notch2-specific defects that might (separately or together) cause microphthalmic lenses. The persistent lens stalks of Notch2 mutants could trap progenitor cells, causing a progressive loss of proliferation, but with no significant loss of proliferation in Notch2 lens mutants, we do not favor this possibility. Instead, there is increased apoptosis in the absence of Notch2. Putative downstream genes that link Notch2 activity to the regulation of cell death were not identified; making this is a good starting point for future studies.
The inappropriate S-phase fiber cells might arise from elevated expression of the transformation related protein, Trp63, in Le-Cre; Notch2CKO/CKO mutants. This protein has context-specific functions, since it has been linked with prolonged proliferation and aberrant differentiation, can either block or promote apoptosis, can act downstream of Wnt9b signaling, or negatively regulate other components of the Wnt pathway (Drewelus et al., 2010; Ferretti et al., 2011; Talos et al., 2010). In addition, our gene expression data indicated that the cell cycle inhibitor, Cdkn1a (p21Cip1), was upregulated in Le-Cre; Notch2CKO/CKO lenses (Fig. 9, Table1). When Cdkn1a is overexpressed, it can promote Fgf-dependent cell cycle progression, by favoring the assembly of active Cdk4/CyclinD complexes (Bansal et al., 2005). The derepression of Ccnd2 and Fgf isoforms in the Le-Cre; Notch2CKO/CKO lenses might further enhance this process, particularly in the absence of counterbalancing expression of cell cycle inhibitors (Fig 9, Table1). As a whole, our findings support the idea that Notch2 normally promotes cell cycle withdrawal.
Conditional deletion of Notch2 also had a marked effect on secondary fiber cell elongation and morphology. Instead of forming a regular array of closely adherent, elongated cells, secondary fiber cells of the Le-Cre; Notch2CKO/CKO mutant were randomly arranged and globular. Similar, abnormal fiber morphology occurs in lenses deficient in Itgb1 (β1-Integrin) (Simirskii et al., 2007), an adhesion protein normally present on basolateral surfaces of lens fiber cells (Menko and Philip, 1995; Walker and Menko, 2009), raising the possibility that Notch2 may play a role in Itgb1-dependent cell-cell adhesion in lens fiber cells. Interestingly, a non-canonical effect of the Notch intracellular domain has been shown to promote activation of Itgb1 by binding to, and activating Rras, thus reversing the Hras mediated suppression of integrin affinity (Karsan, 2008). Both Rras and Hras are expressed in lens fiber cells and overexpression of active Hras in the lens disrupts lens development (Reneker et al., 2004). Together, the aberrant fiber cell morphology and defective cell cycle withdrawal of secondary fiber cells in Notch2 deficient lenses demonstrate that Notch2 is essential for proper secondary lens fiber cell differentiation, as previous in vitro experiments have already demonstrated (Saravanamuthu et al., 2009).
When the Notch2 lens phenotypes are integrated with those Jag1 and Rbpj conditional mutants (Jia et al., 2007; Le et al., 2009; Rowan et al., 2008), a somewhat complex picture of Notch signaling emerges for the developing lens. First, during mouse lens induction and vesicle formation, Notch signaling has only a minor role, during vesicle separation from the overlying ectoderm. All other aspects of early lens development: specification, vesicle growth, morphogenesis and primary fibergenesis, occur independent of this pathway. However, a Jag1-Notch2-Rbpj mediated signal is necessary for proper secondary fiber formation. Loss of any gene results in Hes1 downregulation, progressive dysgenesis of the AEL and transition zone, and a failure of fiber cell denucleation.
Interestingly, Jag1 lens mutants phenocopy those for Rbpj, particularly a loss of AEL cell proliferation. Although we found the Notch2 receptor is not required for this process, it does uniquely block apoptosis. Furthermore, Rbpj and Jag1 are also required for some aspect of secondary fiber cell differentiation, but the nature of this abnormality was not apparent in those mutants, since proliferation and differentiation simultaneously arrested. By contrast, Le-Cre; Notch2CKO/CKO mutant lenses retain AEL cell proliferation, which allowed defective fiber cells to accumulate in the posterior compartment, making their inability to fully exit mitosis obvious. The distinct phenotypes among these three genes strongly suggest additional ligands, receptors and downstream effectors are involved in midgestational and postnatal lens development. In particular, the Dll1 ligand, Notch1 and Notch3 receptors are each expressed during rodent lens development (Bao and Cepko, 1997; Weinmaster et al., 1991; Le and Brown, unpublished data). Therefore, Notch signal transduction appears to segregate its regulation of particular cell processes (e.g. proliferation versus apoptosis) among distinct ligand and receptor combinations. Alternatively, as already mentioned, there may be additional, noncanonical functions for some or all of these genes during lens formation. It will be very interesting in the future to unravel these questions.
A 10 week old adult Le-Cre; Notch2CKO/+ eye H&E section images show normal histology of the A) anterior eye B) cornea C) lens bow region and D) posterior lens. Scale bar in A = 20 μM, in B–D = 100 μM.
The authors thank Tom Gridley for Notch2CKO/CKO mice; Ruth Ashery-Padan and Joram Piatigorsky for Le-Cre transgenic mice; Richard Lang for Cryb antibody; Peter Carlsson for Foxe3 antibody; Eric Wawrousek for Crya antibody, Barbara Norman for technical expertise; and Richard Lang and Tiffany Cook for valuable discussion. This work was supported by the National Eye Institute Intramural Research Program Z01-EY000238 to PSZ and NIH R01 grant EY18097 to NLB.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.