Deletion of Autophagy-related 5 (Atg5) and Pik3c3 Genes in the Lens Causes Cataract Independent of Programmed Organelle Degradation*

Background: The role of autophagy-dependent quality control in the lens remains unclear. Results: Deletion of Atg5 and Pik3c3/Vps34 in the lens does not affect programmed organelle degradation but causes cataract and a developmental defect, respectively. Conclusion: These genes are important for quality control and development of the lens. Significance: This study provides new insights into biology and age-related pathology of the lens. The lens of the eye is composed of fiber cells, which differentiate from epithelial cells and undergo programmed organelle degradation during terminal differentiation. Although autophagy, a major intracellular degradation system, is constitutively active in these cells, its physiological role has remained unclear. We have previously shown that Atg5-dependent macroautophagy is not necessary for lens organelle degradation, at least during the embryonic period. Here, we generated lens-specific Atg5 knock-out mice and showed that Atg5 is not required for lens organelle degradation at any period of life. However, deletion of Atg5 in the lens results in age-related cataract, which is accompanied by accumulation of polyubiquitinated and oxidized proteins, p62, and insoluble crystallins, suggesting a defect in intracellular quality control. We also produced lens-specific Pik3c3 knock-out mice to elucidate the possible involvement of Atg5-independent alternative autophagy, which is proposed to be dependent on Pik3c3 (also known as Vps34), in lens organelle degradation. Deletion of Pik3c3 in the lens does not affect lens organelle degradation, but it leads to congenital cataract and a defect in lens development after birth likely due to an impairment of the endocytic pathway. Taken together, these results suggest that clearance of lens organelles is independent of macroautophagy. These findings also clarify the physiological role of Atg5 and Pik3c3 in quality control and development of the lens, respectively.

The lens of the eye is composed of epithelial cells and fiber cells (1,2). During embryogenesis, primary fiber cells are differentiated from epithelial cells at the posterior surface of the lens vesicle, whereas secondary fiber cells are subsequently differentiated from newly proliferated epithelial cells at the edges of the anterior epithelium. In the process of terminal differentiation of these fiber cells, membrane-bound organelles such as nuclei, the endoplasmic reticulum (ER), 2 and mitochondria are degraded, forming the organelle-free zone (OFZ) (3)(4)(5). Organelle degradation begins during the embryonic period in primary fiber cells and continues throughout life in secondary fiber cells.
Although epithelial-to-fiber cell differentiation continues throughout life, the rate and speed decrease with age (6,7). In rodents, the entire process, from epithelial mitosis to final differentiation, requires only 1 week if initiated during the embryonic period, but it takes 9 months from 5 months of age (6). Because these differentiating fiber cells are metabolically active but quiescent like neurons (8), they require an efficient intracellular quality control system until terminally differentiated, especially the slowly differentiating fiber cells of the adult lens. However, how the intracellular quality in the lens cells is maintained in vivo has not been fully understood.
Macroautophagy (referred to as "autophagy" hereafter) is one of the major intracellular degradation pathways along with the ubiquitin-proteasome system (9). Cytoplasmic proteins and organelles are enclosed by the autophagosome and then delivered to the lysosome by autophagy. Genetic studies in yeast have identified a set of autophagy-related (ATG) genes that are essential for autophagy. Several ATG genes, including Atg5, are well conserved in higher eukaryotes (10). Reverse genetic techniques have revealed various physiological functions of autophagy in mammals such as adaptive responses to starvation, quality control of intracellular proteins and organelles, embryonic development, tumor suppression, and elimination of intracellular microbes (9).
The presence of autophagic vacuoles in differentiating fiber cells was reported using in vitro culture systems (11). We also showed that autophagy is constitutively active in the in vivo mouse lens (12,13). However, contrary to our expectation, the results of our study using conventional Atg5 knock-out mice showed that autophagy is not essential for organelle degradation at least in primary fiber cells (13). Nonetheless, as Atg5 knock-out mice die soon after birth (14), the importance of autophagy in organelle degradation in secondary fiber cells remains unclear. In addition, the role of autophagy in intracellular quality control of lens cells, particularly slowly differentiating fiber cells in the adult lens, has not been determined.
Recently, "alternative autophagy," which is independent of Atg5 and Atg7, was reported and suggested to have a potential role in removal of mitochondria in reticulocytes (15). This type of autophagy was shown to be dependent on some of the upstream Atg factors such as Ulk1, FIP200, Beclin 1, and Pik3c3 (the class III phosphatidylinositol 3-kinase (PtdIns3K), also known as Vps34) (15). These factors are mostly multifunctional; for example, Pik3c3 is important for endocytosis and multivesicular body formation as well as autophagy (16 -18). If alternative autophagy is involved in lens organelle degradation, we might have missed it in our previous study using Atg5 knockout mice (13).
The aim of this study was to generate lens-specific Atg5 and Pik3c3 knock-out mice to define their physiological role in the lens. We found that neither Atg5-dependent nor Pik3c3-dependent autophagy was essential for lens organelle degradation. However, Atg5-dependent autophagy was essential for intracellular quality control and suppression of age-related cataract. We also demonstrated that Pik3c3 was required for development of the lens after birth and suppression of congenital cataract.

EXPERIMENTAL PROCEDURES
Mice-Experimental procedures to produce Atg5 flox/flox (19), MLR10-Cre transgenic (20), and GFP-LC3 transgenic mice (12) have been described previously. Methods to produce Pik3c3 flox/flox mice will be presented elsewhere. 3 Briefly, two loxP sequences were introduced into introns 19 and 21 of the Pik3c3 gene to flank exons 20 and 21. Upon Cre-mediated recombination, the two exons encoding the kinase domain essential for phosphorylation of PtdIns are deleted. Wild-type C57BL/6 mice were obtained from Japan SLC, Inc. All mice were fed ad libitum. All animal experiments were approved by the Institutional Animal Care and Use Committee of Tokyo Medical and Dental University (no. 0130091C).
Immunoblotting-Lenses from both eyes dissected from neonatal and adult mice were homogenized in 0.1 and 1.0 ml, respectively, ice-cold 0.25 M sucrose buffer (50 mM Tris-HCl (pH 7.5), 1 mM EDTA, and Complete EDTA-free protease inhibitor (Roche Applied Science)). For cell fractionation, homogenates were treated with 1% Triton X-100 and centrifuged at 15,000 ϫ g for 15 min to separate the supernatant (Triton X-100-soluble fraction) and pellet fractions. The pellets were resuspended in 1% SDS in phosphate-buffered saline (PBS) (Triton X-100-insoluble fraction). Protein extracts were boiled in sample buffer and subjected to SDS-PAGE and immunoblotting. The amount of protein was quantified by densitometric measurements using ImageJ software.
Proteasomal Activity Assay-Lenses from both eyes dissected from mice at 8 months old were homogenized in 1 ml of ice-cold 0.25 M sucrose buffer containing 1% Triton X-100. The chymotryptic activity of the proteasome was measured by mixing the lysate with an assay buffer containing 100 M fluorogenic peptide substrate succinyl-Leu-Leu-Val-Tyr-7-amino-4methylcoumarin (Peptide Institute, Inc.) in 50 mM Tris-HCl (pH 8.0) and 1 mM DTT in the presence or absence of 50 M MG132. After incubation for 30 min at 37°C, hydrolysis of the synthetic peptides was measured at excitation and emission wavelengths of 355 and 460 nm, respectively, using an ARVO MX Plate Reader (PerkinElmer Life Sciences). MG132-sensitive activity was considered to be proteasome-specific.
Analysis of Oxidized Proteins-Carbonyl-oxidized proteins were detected using the OxyBlot protein oxidation detection kit (Millipore) according to the manufacturer's instructions. In brief, Triton X-100-soluble and -insoluble fractions were first denatured with 6% SDS and then treated with either 2,4-dinitrophenylhydrazine solution or with derivatization control solution (negative control) for 15 min. After neutralization, the samples were subjected to SDS-PAGE and immunoblotting using an antibody specific to the 2,4-dinitrophenol moiety.
Preparation of NaOH-soluble and -insoluble Fractions-The sequential preparation of water-soluble, NaOH-soluble, and NaOH-insoluble fractions was performed as described previously (24). In brief, lens homogenates were first centrifuged at 15,000 ϫ g for 15 min and separated into supernatant (watersoluble) and pellet fractions. The pellets were treated with 20 mM NaOH, centrifuged at 15,000 ϫ g for 15 min, and further separated into supernatant (NaOH-soluble) and pellet fractions. The resultant pellets were washed with 1 mM Na 2 CO 3 three times and resuspended in sample buffer containing 10 mM DTT (NaOH-insoluble fractions).
Immunohistochemistry-Lenses were fixed with 4% paraformaldehyde (PFA) in PBS (pH 7.4) overnight and treated with 15% sucrose in PBS for 4 h and then with 30% sucrose solution overnight. Lens samples were embedded in Super Cryoembedding Medium compound (SECTION-LAB, Japan) and stored at Ϫ80°C. Mid-sagittal lens slices were sectioned at a thickness of 7 m with a cryostat (CM3050 S, Leica) using adhesive film (Cryofilm type IIC9, SECTION-LAB) according to a previously described method (25). Cryosections were stained with hematoxylin and eosin and photographed using a microscope (BX51, Olympus) equipped with a digital camera (DP70, Olympus). For immunohistochemistry, after treatment with 0.1% Triton X-100 for 15 min and blocking in 5% bovine serum albumin in PBS for 30 min, lens sections were incubated with primary antibodies for 1 h, followed by further incubation for 1 h with secondary antibodies. For staining polyubiquitin and p62, lens sections were first prepared without fixation, followed by fixation with 4% PFA for 10 min. Sections were analyzed using a microscope (IX81, Olympus) equipped with a CCD camera (ORCA ER, Hamamatsu Photonics). Images were acquired using Meta-Morph version 7.0 (Molecular Devices Japan).
Electron Microscopy-The lens from adult mice was fixed with 2.5% glutaraldehyde and 4% PFA in 0.1 M cacodylate buffer (pH 7.4) at room temperature overnight. The lens from neonatal mice was fixed with 2.5% glutaraldehyde and then post-fixed with 1.5% OsO 4 in 0.1 M sodium phosphate buffer (pH 7.4) for 2 h. Tissues were dehydrated in a graded series of ethanol and embedded in epoxy resin. Ultra-thin sections were made using an ultramicrotome (Reichert). Sections were stained with uranyl acetate and lead citrate and observed under an H7100 electron microscope (Hitachi).
Quantitative RT-PCR-To prepare fiber cells, lenses from at least six mice were dissected by forceps into fiber cells and capsules that included adhering epithelial cells and vascular cells (the tunica vasculosa lentis). Total RNA was extracted from fiber cells using ISOGEN (Nippon Gene) and reversetranscribed using ReverTraAce (Toyobo). PCRs were performed in triplicate using SYBR Premix Ex Taq (Takara Bio) and were monitored by a Thermal Cycler Dice TP800 (Takara Bio). Expression level of Pik3c3 was normalized to that of Actb. Primers used are listed as follows: Pik3c3, 5Ј-TGTCAGATGA-GGAGGCTGTG-3Ј and 5Ј-CCAGGCACGACGTAACTTCT-3Ј; Actb, 5Ј-TCCCTGGAGAAGAGCTACGA-3Ј and 5Ј-AGC-ACTGTGTTGGCGTACAG-3Ј.
Lens Growth and Cataract Analysis-Dark field images of the dissected lens in PBS were acquired using a microscope (SZX12, Olympus) and photographed. Lens equatorial diameter was traced on digitized images using Adobe Photoshop CS5. The presence of cataract was detected by gross examination and defined as any opacity in the lens of either one or both eyes. Cataract was also confirmed using microscopy after dissecting the lens in most representative cases.
Statistical Analysis-All numerical data represent the mean Ϯ S.E. Statistical comparisons were made using the twotailed Student's t test. The incidence of cataract was statistically validated using the 2 test.

Atg5-deficient Lens Develops Age-related Cataract-To
investigate the physiological function of autophagy in the lens, we generated lens-specific Atg5-deficient mice. Mice bearing an Atg5 flox allele (19) were crossed with MLR10-Cre transgenic mice expressing Cre recombinase under the control of a modified ␣A-crystallin promoter (20). Expression of Atg5 (detected as the Atg12 to Atg5 conjugate) and the Atg5-dependent conversion of LC3-I to LC3-II (21) were completely inhibited in the lens of 15-month-old Atg5 flox/flox ;MLR10-Cre mice (Fig. 1A). To confirm autophagy inhibition in the lens, we crossed these mice with autophagy-indicator mice carrying a GFP-LC3 transgene, a specific marker of the autophagosome (12). In the control lens from 15-month-old Atg5 flox/ϩ ;MLR10-Cre;GFP-LC3 mice, a number of GFP-LC3 puncta were observed in differentiating secondary fiber cells at the superficial lens cortex (referred to hereafter as the "cortical region") outside the OFZ (Fig. 1B, panels a-c) but not in differentiated fiber cells in the OFZ (data not shown), suggesting that autophagy is constitutively active only during the period of fiber cell differentiation. By contrast, Atg5 flox/flox ;MLR10-Cre;GFP-LC3 mice demonstrated almost no GFP-LC3 puncta in lens fiber cells (Fig. 1B, panels d-f). We also detected endogenous LC3-positive puncta in secondary fiber cells of Atg5 flox/ϩ ;MLR10-Cre mice but not Atg5 flox/flox ;MLR10-Cre mice at 7.5 days after birth (Fig. 1C). These results suggest that autophagosome formation is suppressed in the Atg5-deficient lens.
Atg5 Is Not Required for Organelle Degradation in Fiber Cells-We next determined whether autophagy is dispensable for organelle degradation not only in primary fiber cells but also in secondary fiber cells in adult mice. In secondary fiber cells of Atg5 flox/ϩ ;MLR10-Cre mice at 15 months of age, the ER, mitochondria, and nuclei were present in the cortical region but not in the central OFZ (Fig. 1F). Although Atg5 flox/flox ;MLR10-Cre mice developed cataract, the lens OFZ was normally generated in these mice. Electron microscopic analysis also confirmed the generation of an OFZ (Fig. 1G). Thus, these data suggest that autophagy is not required for lens organelle degradation throughout life and that the cause of age-related cataract development in the Atg5-deficient lens is not due to a defect in programmed organelle degradation.
Lens Fiber Cells Are Disorganized in the Cortical Region in the Lens of Aged Lens-specific Atg5-deficient Mice-We next analyzed the phenotype of Atg5-deficient lens in more detail.
Hematoxylin and eosin staining of the lens revealed no obvious morphological difference between Atg5 flox/ϩ ;MLR10-Cre mice and Atg5 flox/flox ;MLR10-Cre mice at 4 months of age ( Fig. 2A). At 21 months, control lens of Atg5 flox/ϩ ;MLR10-Cre mice showed highly ordered structure of the fiber cells in the cortical region (Fig. 2B, panels a-d). By contrast, the fiber cells in this region were disorganized and swollen in Atg5 flox/flox ;MLR10-Cre mice (Fig. 2B, panels e-h). These changes were observed in almost all differentiating fiber cells in the cortical region (Fig.  2B, panel g) and some of the differentiated fiber cells that have lost organelles (Fig. 2B, panel h). Electron microscopic analysis of the lens of 21-month-old Atg5 flox/flox ;MLR10-Cre mice revealed accumulation of vacuoles of various sizes and high density deposits in the cytoplasm of differentiating fiber cells (Fig. 2C) but not in terminally differentiated cells in the central region of the lens (Fig. 1G). Taken together, these results suggest that age-related cataract development in Atg5-deficient lens is caused by disorganization and accumulation of abnormal materials in cortical fiber cells.
Polyubiquitinated Proteins and p62 Accumulate in Atg5-deficient Lens-As autophagy is important for intracellular quality control, autophagy defects lead to accumulation of abnormal proteins and organelles, which are often ubiquitinated (9). Although 4-month-old Atg5 flox/flox ;MLR10-Cre mice were free of cataracts, immunoblot analysis revealed a slight accumulation of polyubiquitinated proteins (Fig. 3A). p62, which is a selective substrate of autophagy and is known to aggregate under autophagy-deficient conditions (26,27), also accumulated mainly in the Triton X-100-insoluble fraction in the lens (Fig. 3A). There was massive accumulation of both polyubiq-   (9), autophagy is important for intracellular quality control in the lens. We next determined where these polyubiquitinated proteins and p62 accumulated. In control lens of Atg5 flox/ϩ ;MLR10-Cre mice, punctate signals positive for ubiquitin and/or p62 were observed in the outermost region of the OFZ at 4 and 22 months (Fig. 3B) and at 0.5 days of age (data not shown). This seems to be related to organelle and protein degradation as part of the normal developmental process, and it is consistent with previous findings that the ubiquitin-proteasome system is active during the process (28). However, in Atg5 flox/flox ;MLR10-Cre mice, aggregates containing ubiquitin and p62 were also observed in the cortical region of the lens outside the OFZ (Fig.  3B), where autophagy constitutively occurs in control mice (Fig. 1B). They were already detected in transparent lens of 4-month-old mice and became more massive in cataractous lens at 22 months. These additional aggregates were not generated in the control lens of Atg5 flox/ϩ ;MLR10-Cre mice (Fig. 3B). No ubiquitin or p62 aggregates were observed in deeper regions of the OFZ of the lens of either Atg5 flox/ϩ ;MLR10-Cre or Atg5 flox/flox ;MLR10-Cre mice (Fig. 3B). These results suggest that constitutive autophagy in slowly differentiating cortical fiber cells is important for prevention of accumulation of abnormal aggregates.
There was no difference in proteasomal chymotryptic activity in the lens between control and Atg5 flox/flox ;MLR10-Cre mice at 8 months of age (Fig. 3C), as reported previously in neuron-specific Atg7-deficient mice (29). The protein level of p53, a typical substrate of the proteasome (30), in the lens was also comparable between control and Atg5 flox/flox ; MLR10-Cre mice (Fig. 3D). As polyubiquitinated proteins and p62 already massively accumulated in the lens of  Atg5 flox/flox ;MLR10-Cre mice at this age, these results suggest that the accumulation of ubiquitin aggregates in Atg5deficient lens was not caused by a defect in the ubiquitinproteasome system.

Insoluble Oxidized Proteins and Crystallins Accumulate in Cataractous
Atg5-deficient Lens-Because age-related cataract is frequently associated with oxidative stress and/or a decrease in crystalline solubility (31), we next investigated whether these Lens homogenates were mixed with fluorogenic peptide substrate succinyl-Leu-Leu-Val-Tyr-7-amino-4-methylcoumarin in the presence or absence of MG132. MG132-sensitive activities were quantified (mean Ϯ S.E.). n.s., not significant, unpaired Student's t test; n ϭ 6. D, immunoblot analysis of p53, a proteasome substrate, in the lens of 8-month-old control and Atg5 flox/flox ;MLR10-Cre mice. Lens homogenates were analyzed by immunoblotting using anti-p53, anti-ubiquitin, and anti-p62 antibodies. ␤-Actin was used as a loading control. Relative protein levels of p53 to ␤-actin were quantified by densitometric analysis (mean Ϯ S.E.). n.s., not significant, unpaired Student's t test; n ϭ 4. E, immunoblot analysis of oxidized (carbonylated) proteins in the lens. Triton X-100-soluble and -insoluble fractions were treated with or without 2,4-dinitrophenylhydrazine to derivatize carbonyl groups in oxidized proteins. These samples were subjected to immunoblot analysis using anti-2,4-dinitrophenol (DNP) antibody. F, immunoblot analysis of crystallins using anti-␥-and anti-␤-crystallin antibodies. Total lens homogenates (T) were first separated into water-soluble (S1) and -insoluble fractions. The latter was treated with NaOH and separated into NaOH-soluble (S2) and -insoluble (P) fractions. ␤-Actin and pan-cadherin were used as loading controls. Data are representatives of two (A, C, D, E, and F) or three (B) independent experiments. changes were found in the Atg5-deficient lens. Carbonylated proteins extensively accumulated in the insoluble fraction in the cataractous lens of 22-month-old but not in the transparent lens of 4-month-old Atg5 flox/flox ;MLR10-Cre mice (Fig. 3E); this suggests that oxidative stress is present in the aged Atg5-deficient lens. In general, crystallins are water-and NaOH-soluble in the normal lens but become insoluble under cataractous conditions (24). Indeed, ␥and ␤-crystallins, which are abundant proteins specifically expressed in lens fiber cells (32), were solubilized in water or NaOH in 4-month-old Atg5 flox/flox ;MLR10-Cre mice. However, in the lens of 22-month-old Atg5 flox/flox ; MLR10-Cre mice, the amount of water-soluble crystallins decreased, whereas that of NaOH-insoluble crystallins increased (Fig. 3F). There was no difference in the distribution of ␥and ␤-crystallins in the lens between Atg5 flox/ϩ ;MLR10-Cre and Atg5 flox/flox ;MLR10-Cre mice at 22 months of age (data not shown). These data indicate that cataract in Atg5-deficient lens is accompanied by accumulation of insoluble oxidized proteins and crystallins.
Pik3c3-dependent Autophagy Is Not Required for Lens Organelle Degradation-Following the recent report that alternative autophagy is dependent on Pik3c3 (15), we generated lens-spe-cific Pik3c3-deficient mice to examine the possible involvement of Atg5-independent alternative autophagy in lens organelle degradation. Pik3c3 was constitutively expressed in lens fiber cells in wild-type embryos at 15.5 and 17.5 days postcoitum (dpc), and neonates at 0.5 and 7.5 days old (Fig. 4A). Whereas endogenous LC3 puncta were detected in the embryonic lens (at 16.0 dpc) of Pik3c3 flox/ϩ ;MLR10-Cre mice, they were absent in the lens of Pik3c3 flox/flox ;MLR10-Cre embryos (Fig. 4B), suggesting that autophagy was inhibited at this time point. This is in line with previous reports using MLR10-Cre transgenic mice (33,34), in which expression of Cre initiated around 10.5 dpc (20), and therefore Pik3c3 flox/flox ;MLR10-Cre mice can be used for functional analysis in both primary and secondary fiber cells as reported previously (33)(34)(35)(36). Immunoblot analysis of the lens confirmed the absence of Pik3c3 proteins and accumulation of polyubiquitinated proteins and p62 in the 0.5-dayold Pik3c3 flox/flox ;MLR10-Cre neonates (Fig. 4C). LC3-I and LC3-II rather accumulated in Pik3c3-deleted lens, which is consistent with previous findings that Pik3c3 and Atg14, a subunit of the class III PtdIns3K complex, are not essential for LC3 conversion (37)(38)(39). Taken together, these data suggest that Pik3c3 was successfully deleted in the lens of Pik3c3 flox/flox ;MLR10-Cre embryos before initiation of organelle degradation in the lens. Next, we investigated the requirement of Pik3c3 in lens organelle degradation during the embryonic period. Because organelles in primary fiber cells are degraded between 17.5 dpc and birth (13,40), we analyzed this time period. At 17.5 dpc, the lens from both Pik3c3 flox/ϩ ;MLR10-Cre and Pik3c3 flox/flox ; MLR10-Cre embryos contained an abundance of ER, mitochondria, and nuclei in primary fiber cells located in the central region (Fig. 4D). By contrast, the organelles in primary fiber cells were completely degraded, and the OFZ was formed in the lens of 0.5-day-old Pik3c3 flox/ϩ ;MLR10-Cre neonates, and there were no significant differences in the lens of Pik3c3 flox/flox ; MLR10-Cre neonates (Fig. 4E). The absence of organelles was further confirmed by electron microscopy (Fig. 4F). These results suggest that Pik3c3 is not required for lens organelle degradation during the embryonic period in primary fiber cells and that neither conventional nor alternative macroautophagy is significantly involved in this process.
Lens-specific Pik3c3-deficient Mice Develop Congenital Cataract and Microphthalmia-Although Pik3c3 flox/flox ;MLR10-Cre mice showed normal organelle degradation in the lens during the embryonic period, they developed congenital cataract and microphthalmia (Fig. 5A). Pik3c3 flox/flox ;MLR10-Cre neonates at 0.5 days old showed bilateral lens opacity. The size of the lens in the Pik3c3 flox/flox ;MLR10-Cre mouse was smaller than that of the Pik3c3 flox/ϩ ;MLR10-Cre mouse at 0.5 and 7.5 days and 2 months, suggesting that post-neonatal lens development was defective. In addition, the size of the entire eyeball was smaller in Pik3c3-deficient mice, compared with control mice; this was identifiable at birth and more pronounced at 2 months of age (Fig. 5B).
Disruption of Pik3c3 Impairs Differentiation of Secondary Lens Fiber Cells with Accumulation of Vacuoles-At 17.5 dpc, secondary fiber cells in the outer cortical region, where differentiation is initiated, were disorganized and not elongated in Pik3c3 flox/flox ;MLR10-Cre embryos (Fig. 5C, panel e). These abnormalities were not detected in primary fiber cells (Fig. 5C, panel f). At 0.5 and 7.5 days after birth, these abnormal secondary fiber cells were still not elongated and had formed cellular aggregates (Fig. 5D). At 2 months, fiber cells were fragmented and markedly aggregated, and the entire lens structure eventually deteriorated. Other ocular abnormalities, including retinal folding and immature iris development, were also detected in these eyes (Fig. 5D). Expression of ␥and ␤-crystallins, which is a hallmark of early fiber cell differentiation (2), was reduced in the aggregated fiber cells in Pik3c3-deficient lens at 0.5 days of age (Fig. 5E), even though expression of these crystallins in whole lens was not significantly affected (Fig. 5F). These results suggest that defective secondary fiber differentiation is the major cause of the developmental defects in Pik3c3-deficient lens.
Electron microscopic analysis of the lens of 0.5-day-old neonates also showed disorganized alignment of swollen secondary fiber cells in the cortical region of the lens of Pik3c3 flox/flox ; MLR10-Cre but not Pik3c3 flox/ϩ ;MLR10-Cre mice (Fig. 6A). A large number of vacuoles were observed in these disorganized secondary fiber cells (Fig. 6A), but not in the primary fiber cells in the central region of the lens (Fig. 4F). Many Lamp-1-positive structures accumulated in the disorganized fiber cells of Pik3c3 flox/flox ;MLR10-Cre mice (Fig. 6B), suggesting that the vacuoles are derived from the late endosome or lysosome. Taken together, these results suggest that Pik3c3 has an essential role in a pathway other than autophagy, probably the endocytic pathway, for development of the lens likely through differentiation of secondary fibers.

DISCUSSION
In this study, we demonstrate that programmed organelle degradation in the lens requires neither Atg5 nor Pik3c3; this suggests that conventional macroautophagy and so-called "alternative" macroautophagy (15) are not primarily involved in this process. These results extend the findings of our previous study demonstrating a nonessential role of Atg5-dependent autophagy in organelle degradation in primary fiber cells (13). Although we could not analyze the role of Pik3c3 in secondary fiber cells due to the developmental defect, our data suggest that a yet uncharacterized degradation system is involved in lens organelle degradation. Several potential mechanisms have been reported, which include the ubiquitin-proteasome system (28,41,42), 15-lipoxygenase (43), and DNase II-like acid DNase (DLAD) (44). Among them, DLAD, a lysosomal DNase (45,46), is required for degradation of nuclear DNA in the lens (44). Thus, it is likely that the lysosome is involved in degradation of nuclear DNA in a manner different from any known types of autophagy. Further analysis would be required to reveal the novel function of the lysosome.
Our findings instead revealed a critical role of autophagy in quality control of intracellular proteins and organelles, as observed in other tissues such as the liver, nervous system, heart, and glomerulus (9,47). This function in the lens is important for prevention of age-related cataract. We detected abnormal accumulation of aggregates containing ubiquitin and p62 in differentiating fiber cells in the cortical region of the lens from aged lens-specific Atg5-deficient mice. There was no abnormal accumulation of these aggregates in the center of the OFZ, where autophagy cannot occur even in wild-type cells because of the lack of lysosomes. Thus, the importance of autophagy is cell type-or age-specific. It was reported that fiber cell differentiation requires 2, 4, and 9 months when initiated at 1, 2, and 5 months of age, respectively (6). This time is almost constant between 5 and 10 months of age (6). It would be reasonable to hypothesize that autophagic quality control is particularly important when fiber cell differentiation is very slow. Once fiber cells are terminally differentiated, autophagy may no longer be important. This could explain why cataract develops in Atg5-deficient lens between 5 and 10 months of age with a relatively intact central region.
An important question is whether impairment of autophagy is associated with cataracts in humans. Recently, mutations in the FYCO1 gene were identified in patients with autosomalrecessive congenital cataracts (48). It was reported that FYVE and coiled-coil domain containing 1 (FYCO1) protein interact with PtdIns(3)P, LC3, and Rab7 and are involved in positioning of autophagosomes (49). FYCO1 and several ATG genes are indeed expressed in the human lens (50). Furthermore, muta-tions in the gene encoding ectopic P-granules autophagy protein 5 homolog (EPG5), a higher eukaryote-specific gene required for lysosomal degradation of autophagosome (51), were recently identified in patients with recessively inherited congenital Vici syndrome, which demonstrates cataracts in addition to other multiple abnormalities such as callosal agenesis, immunodeficiency, and cardiomyopathy (52). Thus, it would be important to determine whether the congenital cata-  ract seen in patients with these conditions is indeed caused by a defect in autophagy or in another function of FYCO1 (53) and EPG5 (54) and whether autophagic activity is affected in other types of human cataract. We also revealed that Pik3c3 is essential for lens development after birth. The abnormal development of the entire eyeball, including the retina and iris in Pik3c3 flox/flox ;MLR10-Cre mice, may be a secondary result of impaired lens development, as similar phenotypes were reported in other mouse models with lens-specific deletion of fibroblast growth factor receptor (33), PKC (34), ␤1-integrin (35), ␤-catenin (36), and N/E-cadherin (55). Alternatively, it may be due to an off-target effect of MLR10-Cre expression in non-lens tissues in the eye. However, this is unlikely because MLR10-Cre is highly specific to the lens within the eye; only a few cells in the cornea demonstrate ectopic expression (20). As development of the lens in Atg5 flox/flox ;MLR10-Cre mice is normal, the impaired development of Pik3c3-deficient lens is likely to be due to a defect in an autophagy-independent pathway. As reported previously in other Pik3c3-deficient cells (17,56,57), we observed many enlarged vacuoles in the lens, which were likely positive for Lamp-1, suggesting a defect in the late endosome or lysosome. Another possibility is that the lens phenotype is due to a defect in alternative autophagy (15). However, the lack of specific markers for alternative autophagy hampered further investigation.
Our data suggest that fiber cell differentiation is defective in Pik3c3-deficient secondary but not primary lens fiber cells. A similar pattern is observed in the zebrafish lens opaque (lop) mutant, in which a gene encoding PtdIns synthase is mutated, and lens development is compromised in the secondary but not primary fiber cells (58). Thus, the contribution of PtdIns metabolism during differentiation might be different in these two fiber cell types. Mutations in the gene encoding oculocerebrorenal syndrome of Lowe 1 (OCRL1), a phosphoinositide 5-phosphatase, are responsible for Lowe syndrome, characterized by congenital cataracts, mental retardation, and renal Fanconi syndrome (59). Therefore, further studies on the role of Pik3c3 in PtdIns metabolism of the lens will provide a more general insight into the pathogenesis of congenital cataracts.