Persistent FoxE3 Expression Blocks Cytoskeletal Remodeling and Organelle Degradation during Lens Fiber Differentiation

Abstract

purpose. The anterior hemisphere of the lens is covered by an epithelial monolayer that acts as the stem cell population for lens fiber progenitors. Foxe3, a forkhead transcription factor, is essential for proliferation and survival of the epithelial cells, and cessation of Foxe3 expression at the lens equator coincides with the cell cycle arrest that marks initiation of fiber differentiation. In this study, the consequences of persistent Foxe3 expression during fiber differentiation was investigated.

methods. The α-A-crystallin (Cryaa) promoter was used to drive transgenic expression of Foxe3 in murine differentiating lens fibers.

results. Transgenic mice have a dramatically disturbed lens histology and grave cataracts. Microarray transcript profiling showed an increase of mRNAs normally enriched in epithelial cells, consistent with an epithelialization of the transgenic fibers. Some aspects of fiber differentiation were unaffected, such as the expression of α- and β-crystallins and aquaporins, whereas cytoskeletal remodeling, cell adhesion, organelle degradation, and antimitotic signaling were compromised.

conclusions. Proper inactivation of FoxE3 expression at the lens equator is important for many aspects of fiber differentiation, and persistent expression leads to a partial epithelialization of fiber cells, with severe consequences for lens function.

During embryonic development, the lens of the eye is formed by invagination of the head surface ectoderm and detachment of a lens vesicle, initially consisting of epithelial cells. ¹ The posterior cells undergo differentiation to primary lens fibers that elongate and fill the lumen of the vesicle. Through this process, the basic organization of the lens is established: a sphere filled with long, narrow fiber cells and a lens epithelium that covers its anterior hemisphere. Growth occurs by recruitment of secondary lens fibers from the posterior cells of the epithelium. All internal lens cells are postmitotic, but proliferation is intense in the epithelium, particularly during development when the lens grows. Throughout life, cell division in the epithelium provides progenitors for renewal of lens fibers, a process essential for optical clarity and prevention of cataract. 

The transition from epithelial to fiber cell occurs at the lens equator and involves dramatic changes in cell shape, physiology, and content. ² Proliferation ceases, and large amounts of lens-specific proteins are synthesized (e.g., crystallins, aquaporins [MIP] and a unique type of intermediary filament proteins that form beaded filaments). 

The epithelial-to-fiber switch requires the homeodomain protein Prox1. ³ An important target for Prox1 is Cdkn1c, encoding the Cdk inhibitor p57, and expression of Cdk inhibitors in equatorial cells is required for cell cycle exit and fiber differentiation. ⁴ Maf, Sox1, and Sox2 are transcription factors important for driving parts of the fiber differentiation process, such as activation of crystallin genes. ^{5 6 7 8 9}  

Many growth factors and their cognate receptors are expressed in the lens, and several (e.g., Tgf-β and Wnt) have been shown to influence lens cell fate. ¹⁰ Fgf invokes the lens fiber differentiation program, including cell cycle withdrawal, elongation, and expression of differentiation markers. ¹¹ For example, does a secreted form of the Fgf receptor Fgfr3 block fiber differentiation, move the transition zone posteriorly, and delay the expression of markers such as Prox1, Cdkn1c, and Maf. ¹² The retina is an important source of Fgf, which creates an anteroposterior Fgf gradient. As a consequence, genetic manipulations that result in retinal ablation, such as overexpression of Bmp7 in the lens, result in posterior expansion of the lens epithelium, a process that could be rescued by ectopic Fgf3. ¹³  

The capsule that encloses the lens is a thick basement membrane that consists of extracellular matrix (ECM) proteins such as collagen IV, laminins, heparan-sulfate proteoglycans (HSPGs), and fibronectin. In addition to its structural and anchoring role, it regulates growth factor availability ¹⁴ ; for example, does the Fgf-binding ability of HSPGs increase the local concentration of Fgf around the lens epithelium. ¹⁵ The importance of integrin-mediated ECM anchorage for epithelial cell survival is illustrated by increased epithelial apoptosis and microphthalmia in mice lacking β1 integrin. ¹⁶  

The epithelium-to-fiber transition involves a coordinated cell elongation and movement that results from extensive actin cytoskeleton remodeling, contraction, and adhesion. Presumptive fiber cells elongate, and their ends move posteriorly along the lens capsule and anteriorly along the epithelial-fiber interface until they reach the nucleus of the lens, where they detach from their substrata and form junctions with fibers from the opposite side in sutures. The importance of cytoskeletal dynamics in this process is illustrated by the lenticular expression of many genes that are necessary for fibroblast migration. The Rho family of small GTPases, which modulate actin polymerization, regulates lens cell movement, and inhibitors of Rho GTPase activity cause cataract. ^{17 18} When epithelial cells initiate differentiation, membrane-associated complexes assemble that contain focal adhesion kinase (FAK), paxillin, caldesmon, and other proteins that link the cytoskeleton to the extracellular environment ¹⁹ ; N-cadherin cell–cell junctions associate with the cytoskeleton and basal membrane complexes; actin is redistributed from stress fibers to cortical bundles ²⁰ ; and beaded filaments are assembled from filensin (Bfsp1) and phakinin (Bfsp2). ²¹  

During the final stages of the differentiation process, fiber cells lose all membrane-bound organelles. This process has been compared with apoptosis, ²² even though caspases, central players in the apoptotic process, have yet to be shown to play a role in lens fiber differentiation. ²³ The trigger of organelle breakdown is unknown, but evidence has been presented of a passive process, driven by gradients of oxygen or metabolites within the fiber cell mass, ²⁴ rather than a coordinated signaling event. Because of the degradation of organelles, homeostasis of the central fibers relies on metabolites provided by cortical fibers and epithelial cells and are tightly interconnected by a network of gap junctions, formed by connexins-46 and -50. ^{25 26} This metabolic coupling is essential for proper lens function, and loss of connexins results in nuclear cataracts. ²⁷  

The gene encoding the forkhead transcription factor Foxe3 is expressed in the lens epithelium and is turned off at the stage when fiber differentiation begins. ²⁸ In Foxe3-null mutants, the lens epithelium gradually disappears during embryonic development. The result is a small lens with severe cataract that fails to separate from the cornea. ^{28 29 30 31} Mechanisms behind the disappearance of the lens epithelium involve failure to proliferate, apoptosis, and premature onset of fiber differentiation. ²⁸ Without functional Foxe3, expression of Prox1 and Cdkn1c expands forward, into the primordial lens epithelium, which suggests that one way for Foxe3 to uphold proliferation and prevent differentiation in the epithelium is to exclude the expression of Cdk inhibitors. ²⁸ In the present study we investigated the consequences for lens fiber differentiation of persistent expression of Foxe3. 

Materials and Methods

Construction of Tg(Cryaa-Foxe3)

A 1364-bp AvrII fragment containing the entire Foxe3 coding sequence was cloned downstream of the −366 to +46 Cryaa promoter in pACP3 ^{32 33} and upstream of the SV40 small t intron and poly-A site. The transgenic construct was released from the plasmid backbone, purified, and used for pronuclear injection in C57Bl/6 x CBA F1 oocytes. Integration-positive animals were identified by Southern blot, hybridized with a Foxe3 probe, and the copy number was estimated by comparing the intensity of the band from the endogenous Foxe3 locus with that derived from the transgene. Two low-copy (lines 31 and 43) founders and one-high copy (line 48) founder were bred to establish the lines. To make the transgenes congenic with the Foxe3-null mutant, ³¹ we maintained the Tg lines were maintained by crossing them with BALB/c. All animals used in this study (except 1 Figs. 2A 2B , which show the founder of line 43 and an integration-negative littermate) have at least five generations on this background. Experimental procedures adhered to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Histology and Immunohistochemistry

Paraffin sections, 3 to 5 μm, of formaldehyde-fixed eyes were stained with mouse monoclonal antibodies against ATP synthase (A211351, Anti-Ox Phos Complex V subunit Beta; Invitrogen-Molecular Probes, Eugene, OR), connexin-50 (33-to 4300; Zymed, South San Francisco, CA), N-cadherin (33-to 3900; Zymed) using the ARK mouse-on-mouse kit (Dako, Glostrup, Denmark), antigen retrieval with Na-citrate buffer, amplification by TSA (Perkin Elmer, Wellesley, MA), and detection with HRP/DAB. For immunofluorescence, polyclonal antibodies against Prox1 (kindly provided by Stanislav Tomarev), γ-crystallins (kindly provided by Samual Zigler) and Foxe3 ³¹ were detected by biotinylated anti-rabbit antibodies and streptavidin-AlexaFluor (Invitrogen-Molecular Probes). Histology and transmission electron microscopy (TEM) were performed as previously described. ³¹  

Microarray

Total RNA was prepared from six pools of 12 lenses each, freshly dissected from 2-day-old pups (GenElute Mammalian total RNA kit; Sigma-Aldrich, Stockholm, Sweden). Three pools were from Tg(Cryaa-Foxe3), line 43 (a low-copy-number line), and three were from wild-type BALB/c. RNA integrity was verified by electrophoresis on a bioanalyzer (model 2100; Agilent, Palo Alto, CA). Five micrograms of each RNA preparation was labeled and hybridized to a mouse whole genome array (MOE430 2.0; Affymetrix, Santa Clara, CA). Hybridization and scanning of the arrays were performed at the Microarray Resource Centre (MARC; Lund, Sweden). 

Quantitative-Polymerase Chain Reaction

Microarray data were verified by Q-PCR on cDNA prepared from the same RNA preparations as the probes (Affymetrix) and from RNA extracted from independent pools of lenses. Gene-specific primer pairs were either purchased (SuperArray, Inc., Frederick, MD), or designed in house (for primers used, see Supplementary Table S1). SyberGreen nucleic acid stain was used for product detection on a real-time PCR system (model 7900HT; Applied Biosystems, Inc., Stockholm, Sweden). Each cDNA preparation was analyzed in triplicate with the gene-specific primer pair and normalized against triplicates for the transcript 36B4. 

Results and Discussion

Transgenic Expression of Foxe3 in Lens Fiber Cells

To investigate the role of Foxe3 in the transition from lens epithelium to fiber, we generated transgenic mice in which Foxe3 is under the control of the Cryaa promoter. ³³ Cryaa is weakly expressed in the lens vesicle and epithelium from around E11, and is induced to high levels at the onset of fiber differentiation ³⁴ (Fig. 1C) . Transgenic, Cryaa-driven, Foxe3 expression is thus induced at the stage in lens cell differentiation when endogenous Foxe3 is silenced. Tg(Cryaa-Foxe3) is activated later than the endogenous Cryaa gene, and—consistent with results from other laboratories using the same promoter construct—we did not detect transgenic transcripts in the lens epithelium. To avoid nonphysiological concentrations of transgene-derived protein, we inserted an intron downstream of the Foxe3 coding sequence to induce nonsense-mediated decay, ³⁵ thereby reducing the amount of protein produced. In situ hybridizations of E14.5 wild-type and Tg(Cryaa-Foxe3) lenses with Foxe3 and Cryaa probes are shown in Figures 1A 1B 1C . 

Three integration positive lines, two low-copy (lines 31 and 43) and one high-copy (line 48), were selected for further study (Fig. 1D) . All three had Foxe3 immunopositive nuclei in interior lens cells. Based on immunofluorescence, the low-copy lines had a transgene-derived Foxe3 concentration in fiber cells that was of the same magnitude as the endogenous Foxe3 in epithelial cells (Fig. 2) . At the earliest stage at which division between epithelium and primordial fiber cells is completed, embryonic day (E)14.5, no major alterations in lens morphology or histology were apparent, but as lens fibers matured in the wild-type the transgenic lenses gradually deteriorated. At birth, mutant lenses were filled with vacuoles and cavities (Figs. 3A 3B) . When the pups opened their eyes, the transgenic animals were easily identified by a severe cataract (Figs. 2A 2B) . The characteristic long, narrow lens fibers of mature lenses were replaced by wide, highly irregular cells (Figs. 3C 3D) , frequently separated by gaps of extracellular space. 

At the transition between epithelium and fiber cells, the transgenic lenses have a narrow zone devoid of Foxe3 nuclear staining. Here, the endogenous Foxe3 protein has vanished, and the transgene-derived has not yet appeared (Figs. 2C 2D) , which permits normal initiation of fiber differentiation before ectopic Foxe3 interferes with the process. Hence, this transgenic model does not answer the question of whether uninterrupted presence of Foxe3 in the transition zone would block initiation of fiber differentiation. 

Mice with spontaneous or targeted inactivation of Foxe3 have two main lens defects. The first is failure of the lens vesicle to close and disconnect from the surface ectoderm, and the second is loss of the lens epithelium due to cell cycle arrest, premature differentiation, and apoptosis. ^{28 31} Transfer of the Tg(Cryaa-Foxe3) transgene to a Foxe3-null background ³¹ did not rescue any part of the knockout phenotype; internal lens cells expressed Foxe3, but the epithelium was neither maintained, nor restored (Figs. 2E 2F 2G 2H) . 

Transcript Profiling of Tg(Cryaa-Foxe3) Lens

To identify changes in the gene expression pattern induced by ectopic expression of Foxe3, we compared the transcriptomes of lenses from 2-day-old Tg(Cryaa-Foxe3; line 43, a low copy line) and wild-type (BALB/c) pups, by whole genome chip hybridization (Affymetrix). RNA for microarray analysis was prepared from three pools per genotype, and each pool consisted of lenses from six animals. The microarray raw data are available at GEO with accession GSE9711 (Gene Expression Omnibus, http://www.ncbi.nlm.nih.gov/projects/geo/ provided by NCBI, National Institutes of Health, Bethesda, MD). 

Of 45,101 probe sets, 23,914 did not produce significant signals, and 17,880 gave signals, indicating that the mRNA was present in all hybridizations: 1,402 only from transgenic lenses and 1,905 only in wild-type. The raw data were normalized by using GEPAS (Gene Expression Pattern Analysis Suite) Expresso with rma background correction (http://gepas.nbn.ac.za/cgi-bin/expresso/expresso/ provided in the public domain by the Principe Felipe Centro de Investigacion, Valencia, Spain), quantiles normalization, and median polish summary method with perfect match values only. ³⁶ Of the 21,187 probe sets representing genes expressed in at least one genotype, 627 had a statistically significant difference in expression level between transgenic animals and wild-type (ANOVA, P < 0.05) with ratios in the range of 1.5- to 24-fold (Supplementary Table S2). Microarray data were verified by quantitative real-time PCR mRNA for 16 genes with different tg-to-wt ratios. The two data sets showed excellent correlation (r = 0.95), with a tendency of the microarray to underestimate the concentration difference (Fig. 4 , Table 1 ). 

The most abundant transcripts in lens encode the dominating proteins of lens fibers—for example, α-, β-, and γ-crystallins, major intrinsic protein (Mip), and Cd24a. Most of these genes had tg-to-wt ratios close to unity, which shows that important aspects of fiber differentiation, such as activation of most crystallin genes, are undisturbed by persistent Foxe3 expression. One exception was γ-crystallins; mRNA for Cryga and -f, encoding γA- and γF-crystallins, were reduced three- and twofold, respectively, and moderate reductions in transcript levels were observed for another three Cryg genes. Immunofluorescence confirmed that γ-crystallin protein levels were reduced (Figs. 5C 5D) . Since γ-crystallins are the last of the major crystallins to accumulate, their selective reduction is consistent with the interpretation that transgenic fibers do not reach maturity. 

The changes in cellular architecture that occur during lens fiber differentiation require extensive alterations of the cytoskeleton, and this remodeling is also a cue for other aspects of the differentiation process. ^{37 38} Synthesis of the intermediary filament protein Bfsp1 is linked to fiber differentiation, and forms, together with Bfsp2, the beaded filaments—a cytoskeletal component specific for lens fibers. ³⁹ Consistent with a threefold reduction in Bfsp1 mRNA, Tg(Cryaa-Foxe3) lack the characteristic intracellular filaments seen with TEM in normal lenses (Figs. 3C 3D)and resemble Bfsp1 knockouts. ⁴⁰ The reduction in Bfsp1 and -2 expression in Tg(Cryaa-Foxe3) is balanced by an increase in vimentin, which makes the transgenic fiber cells more epithelium-like with respect to intermediate filament composition. The abnormal lens fiber cell morphology of Tg(Cryaa-Foxe3) is accompanied by altered expression levels of many additional genes encoding components of the cytoskeleton, or enzymes modifying it. Examples of upregulated transcripts are Pdlim1, an adaptor recruiting the Clik1 kinase to actin stress fibers in nonmuscle cells ^{41 42} ; caldesmon1, a component of the membrane complex that anchors fiber cells to the lens capsule ¹⁹ ; Tubb2c; and Angptl2—all upregulated two- to fourfold. Examples of downregulated cytoskeleton genes are: Ngef, a guanine exchange factor for Rho GTPases; spectrin β2, an actin-binding protein involved in TGF-β signaling ⁴³ ; Tmod1, tropomodulin, expressed in differentiated fibers where it forms apical- and basal-end contacts between fiber cells and epithelium/capsule, respectively ⁴⁴ ; fukutin; fascin; Pstpip2; Dync2h1; Pdlim7; and Mtpn—all with expression reduced 1.5 to 5-fold. 

Cell membranes undergo major composition changes during fiber differentiation. For example, nutrition and maintenance of internal fiber cells are dependent on cytoplasmic flux of metabolites through gap junctions, and any disturbance in this process leads to cataract. ^{27 45 46} Tg(Cryaa-Foxe3) has twofold reduction in Gja3 mRNA, encoding one of the two connexins present in lens fibers, connexin-46. The other connexin of lens fibers, connexin-50 (Gja8), is proteolytically processed during fiber maturation, which leaves the central part of wild-type lenses nonreactive to an antibody directed to its C terminus. ⁴⁷ Transgenic lenses had the anti-Gja8 staining concentrated to the subepithelial zone, and lacked the immunonegative nucleus of normal lenses (Fig. 5B) . 

Tg(Cryaa-Foxe3) lenses had a twofold reduction in Capn3, encoding the Ca²⁺-activated protease calpain-3, which is important for crystallin maturation. ⁴⁸ Calcium homeostasis in the lens depends on the gap junction network, as illustrated by calpain-3 cleavage of γ-crystallins in the Gja3-null mutant. ⁴⁹ Hence, tight control of calpain-3 is essential for protein stability and solubility, and errors in it are major causes of cataract. Other genes of importance for protein stability and folding, and with altered expression in Tg(Cryaa-Foxe3) lenses, are members of the heat-shock protein family: Hsp110, Hspa9a, Hspe1, Hspa4l, Dnajb10, Dnajb6, and Dnajb4—all downregulated 1.5- to 2.5-fold. Of particular interest is Hspb1 (Hsp25 or -27; mRNA reduced fivefold), a heat-shock protein whose phosphorylation status changes as lens epithelial cells differentiate into fibers. ⁵⁰ Hspb1 has a major role in regulating actin fiber dynamics and is phosphorylated by TNF-α through activation of the p38 map kinase and activated transcriptionally by TGF-β. ^{51 52 53}  

The alterations in expression of genes encoding ECM components and their cell surface receptors in response to persistent Foxe3, are extensive. Examples include a shift from integrin α5 (normally found in lens fibers) to α6 (the lens epithelium isoform), downregulation of all ceacam cell adhesion proteins (1, 2, and 10), and upregulation of perlecan (Hspg2) and B2m. Genes involved in cell–cell adhesion include Lims1, Sorbs1 (both down 2-fold), and Shroom3 (up 1.5-fold). Defective cell-adhesion is obvious from TEM images. Figure 3Dshows the disruption of fiber cell organization and the many gaps of extracellular space between fiber cells in the transgene. As a result, cells were highly irregular in shape, both in cortical and central parts of the lens. N-cadherin immunostaining revealed how the densely stained pattern characteristic of epithelium and cortical fibers expanded to the entire lens in Tg(Cryaa-Foxe3) (Figs. 5E 5F) . These changes are all consistent with the epithelialization of the transgenic lens. 

Further support for epithelialization of the fiber cells in Tg(Cryaa-Foxe3) lenses was obtained when the microarray data were compared with expression profiles of cortical fibers versus epithelial cells (GEO dataset GSE7533). ⁵⁴ Genes preferentially expressed in the epithelium are generally upregulated in the Tg(Cryaa-Foxe3) lens—for example Ctgf, Bmpr1a, Sdpr, S100a1, Hhip, Fabp3, and Bmp2k. Likewise, fiber-specific genes tended to be downregulated in the transgenic lenses: Hrasls3, cyclin G1, spectrin β2, Bfsp1, Hmox1, Maf, Hspb1, Traf6, Fgf12, Stat5a, and Gadd45b, to name a few. Exceptions to this rule also exist: Gas2, Fzd1 (down in Tg(Cryaa-Foxe3)) and Tgfb3, Tnfrsf22, Tnfrsf23, Cald1 (up in Tg(Cryaa-Foxe3)). However, when the entire datasets of genes with more than 1.5-fold altered expression in both experiments are compared, there is a highly significant (Spearman rank-order correlation r _s = −0.52, P < 10⁻⁶) covariance between epithelium (versus fiber) and Tg(Cryaa-Foxe3) (versus wild-type). 

The homeodomain transcription factor Prox1 is essential for lens fiber differentiation and is transcriptionally activated in the equatorial zone, where transition from epithelial to primitive fiber cell takes place. ³ Inactivation of Foxe3 results in the spread of Prox1 into the primordial epithelial layer, cell cycle arrest, and premature differentiation. ²⁸ This suggests an antagonistic relationship between Foxe3 and Prox1, and expanding the expression range of the first may therefore reduce that of the latter. Indeed, the Prox1 microarray signal from transgenic lenses was reduced twofold, compared with wild-type, and immunofluorescence confirmed that the protein is significantly reduced in the transgenic fiber cells (Figs. 3E 3F) . In the lens epithelium, on the other hand, the low level of Prox1 was comparable in transgenic and wild-type lenses. We were unable to verify the reported shift in subcellular localization of Prox1 at the lens equator ⁵⁵ ; in our sections Prox1 staining appeared to be predominantly nuclear also in the epithelial cells. Other transcription factors involved in lens development that exhibit a differential expression in response to ectopic Foxe3 are Maf (down 2-fold), which activates the Crygf promoter, Mitf (down 1.5-fold), Bach2 (up 2-fold), and Klf6 (down 2-fold). 

Foxe3 is normally restricted to the actively proliferating lens epithelium and is turned off at the cell cycle exit that marks the onset of fiber differentiation. ²⁸ If Foxe3 is involved in sustaining growth, as the null phenotype suggests, activation of mitogenic and decreased expression of growth inhibitory genes would be expected consequences of ectopic Foxe3 in fiber cells. Expression of genes encoding growth-inhibitory proteins are indeed significantly reduced in the transgenic lenses—for example, cyclin G (down 2.5-fold); Gadd45b (down 5-fold); Gadd45g (down 3-fold); cullin 1 (down 2-fold); Ing2, a protein that negatively regulates cell proliferation by activating p53 ⁵⁶ (down 2.5-fold); Hrasls3 (down 2-fold); Ppp2r5e, a regulatory subunit of Pp2a that interacts with cyclin G via p53 ⁵⁷ (down 2-fold); Scap2 (down 4-fold); Rassf5 (down 2.5-fold); Lats1 (down 2-fold); Ches1 (down 1.5-fold); Uhrf2, involved in G₁/S transition by binding to Cdk2/cyclinE complex ⁵⁸ (down 2-fold); and Chfr, an ubiquitin ligase that controls G₂/M transition through regulating the activation of the Cdc2 kinase ⁵⁹ (down 2-fold). Mxd1, encoding a Max/Myc inhibitory factor with a role in cell cycle arrest and terminal differentiation, ⁶⁰ was downregulated 2-fold. Mitogens, such as connective tissue growth factor (Ctgf, up 2.5-fold), and intracellular cell cycle stimulatory proteins, like Tensin3 (up 1.5-fold), are expressed at a higher level in the mutant. 

Major shifts in the composition of signal-transducing proteins were induced by ectopic Foxe3. The cross-talk between signaling pathways and the importance of posttranslational modifications such as protein phosphorylation make it difficult to predict the functional significance of alterations in expression of individual genes. However, several components and targets of the TGFβ pathway had altered expression levels (Tgfb3 and Ctgf up; Smad1, Gadd45b, and BMP receptors down). Conditional deletion of Bmpr1a in the lens results in disorganized fiber cells with displaced nuclei, ⁶¹ much like in Tg(Cryaa-Foxe3). Fgf and Wnt signaling are important, both in the epithelium and during fiber differentiation. ¹⁰ Tg(Cryaa-Foxe3) have reduced expression of Fgf12, but increased expression of Fgfr2. The Wnt receptor Fzd1 and a secreted Frizzled-related protein, Frzb, are reduced, and so is cullin1, a growth repressor mentioned earlier that is the target for both the Wnt and TGFβ pathways. 

Components of the interconnected pathways of TNFα and apoptosis are shifted toward increased expression of antiapoptotic and decreased expression of proapoptotic genes. For example, mRNA for the apoptotic executor caspase-7 is reduced 5-fold, and the regulatory apoptosis facilitators Bcl2l13, Dapk1, Arf5, Bcl2l2, and Gas2 are all decreased between 1.5- and 2-fold. Zfand5, which sensitizes cells to TNFα induced apoptosis, ⁶² and Zfp36 (Ttp), which induces apoptosis synergistically with TNFα by binding to its mRNA 3′ UTR, ⁶³ are both reduced twofold. Other changes consistent with a reduction in TNF-related signaling affect kinases in the Map kinase cascade. Map3k4 (MTK1 or MEKK4), Map3k5 (ASK1), Map4k4 (HGK), and Map4k5 (GCKR) are activated in response to TNF-receptor signaling ^{64 65 66 67} and trigger apoptosis by activation of JNK (Mapk8) and p38 (Mapk14). Targets of this pathway are Jun, Atf2, and other proteins. ^{68 69} Gadd45α, -β, and -γ act through the JNK/p38-pathway, by binding and activating Map3k4 ⁷⁰ and are transcriptionally activated by Tgf-β and NF-κB stimulation. ^{71 72} Gadd45b and -g are highly expressed in the murine lens and are both downregulated in Tg(Cryaa-Foxe3), five- and threefold, respectively. The same is true of two of the genes encoding the kinases in the cascade: Map3k5 (down 5-fold) and Map4k4 (down 2-fold). The TNF receptor adaptor Traf6, (down 1.5-fold), activates Map3k5 ⁶⁴ in a complex with Pellino1 (down 2-fold). ⁷³ Map4k4 is both a mediator and an Atf2-dependent transcriptional target of TNFα signaling. ⁷⁴ Atf2 (down 1.5-fold) is a target for JNK/p38 phosphorylation and activates the Hmox1 promoter. ⁷⁵ The genes encoding the murine TRAIL decoy receptors, Tnfrsf22 and -23, are upregulated (threefold) in Tg(Cryaa-Foxe3). Human decoy receptors have been shown to inhibit apoptosis in many different systems. ^{76 77} Although murine and human decoy receptors are not orthologous, they all bind TRAIL-ligand and lack a cytoplasmic death domain, arguing for a functionally analogous role. ⁷⁸ Thus, the changes in expression levels of signaling components in Tg(Cryaa-Foxe3) predicted a decrease in TNFα signaling, and this was confirmed by the reduction in expression of targets of the JNK/p38-pathway lenses (e.g., the fivefold reduction of Hmox1 mRNA). 

What may the functional significance of a shift to pro-life genes be? One of the mechanisms that eliminate lens epithelial cells in Foxe3-null mutants is apoptosis, ²⁸ which supports the notion of Foxe3 promoting cell survival, but apoptosis is not a common process in the mature, healthy, normal lens. On the other hand, a subset of the apoptotic program is used in the organelle degradation that occurs during the final stages of lens fiber differentiation. ²² TNFα plays a role in the removal of nuclei in the chicken embryo lens, which occurs without other hallmarks of apoptosis, such as membrane breakdown. ⁷⁹ Reduced expression of proapoptotic genes may therefore result in inhibition of fiber maturation. Histology and immunohistochemistry of Tg(Cryaa-Foxe3) supports this; the organelle-free zone characteristic of the core of wild-type lenses (Figs. 6A 6C 6E 6G)is missing in the transgenic lenses. Instead, nuclei and mitochondria persist throughout the mutant lens (Figs. 6B 6D 6F 6H) . The signature of failed nuclear breakdown is also evident in the microarray data: expression of DNase IIβ—the lens-specific nuclease responsible for chromatin fragmentation during fiber differentiation ⁸⁰ —is reduced fivefold and urate oxidase, part of purine catabolism, ⁸¹ fourfold. 

To summarize, the combined impression from microarray, histology, and immunohistochemistry, is that ectopic Foxe3 expression during lens fiber differentiation alters the gene expression profile in a way that interferes with many aspects of fiber development. Cell shape, cytoskeleton, adhesion, ECM, signaling related to apoptosis and organelle degradation are examples of processes affected by Foxe3 (outlined in Fig. 7 ). On the other hand, some aspects of fiber differentiation appear to proceed normally. Unaltered expression of major lens proteins such as α- and β-crystallins and aquaporin (Mip) are testimony to this. 

² Present affiliation: Departments of Ophthalmology and Medical Genetics, University of Alberta, Edmonton, Alberta, Canada.

 Supported by Swedish Medical Research Council Grant K2006–31X-13460-07-3 and Swedish Cancer Foundation Grant 07-0641.

 Submitted for publication May 5, 2008; revised May 22, 2008; accepted August 11, 2008.

 Disclosure: H. Landgren, None; Å. Blixt, None; P. Carlsson, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Figure 1.

 

In situ hybridizations of E14.5 eye sections and Southern blot of genomic DNA from Tg(Cryaa-Foxe3), line 43, hybridized with a Foxe3 probe. (A) Section of E14.5 wild-type eye hybridized with a Foxe3 RNA probe showing expression confined to the lens epithelium. (B) Section of E14.5 Tg(Cryaa-Foxe3) eye hybridized with Foxe3 RNA probe showing strong expression from the transgene in the lens fibers. (C) Wild-type section hybridized with a Cryaa RNA probe showing the inverse of (A): high expression in the fiber cells, and low in the epithelium. (D) Southern blot of genomic DNA from Tg(Cryaa-Foxe3) and wild-type littermate. The two top bands correspond to fragments from the Foxe3 locus and the bottom band in the right lane is derived from the integrated Cryaa-Foxe3 construct.

Figure 2.

 

Phenotype and Foxe3 protein distribution in Tg(Cryaa-Foxe3). (A, B) Eye of adult Tg(Cryaa-Foxe3), line 43 founder. (B) Cataract that was not present in the integration-negative littermate (A). (C, D) Immunofluorescence with anti-Foxe3 serum (red) and DAPI-stained nuclei (blue) of wild-type (C) and Tg(Cryaa-Foxe3) line 31 (D) E14.5 eye sections. Insets: close-up of the lens equator and the zone of Foxe3-negative nuclei between positive epithelial and fiber cells. (E–H) Tg(Cryaa-Foxe3) did not rescue the Foxe3-null phenotype. Foxe3 immunofluorescence of E14.5 eyes from wild-type (E), Tg(Cryaa-Foxe3) (F), Foxe3^−/− (G), and Tg(Cryaa-Foxe3); Foxe3^−/− (H).

Figure 3.

 

Histology, TEM, and Prox1 localization in Tg(Cryaa-Foxe3). (A, B) Hematoxylin and eosin stained sections of eyes from 2-day-old wild-type (A) and Tg(Cryaa-Foxe3) (B) mice. (C, D) TEM of lenses from 3-week-old wild-type (C) and Tg(Cryaa-Foxe3) (D) mice showing the normal, homogenous pattern of lens fibers, filled with beaded filaments, replaced by highly irregular and heterogeneous cells. (E, F) Prox1 immunofluorescence (red) of E14.5 eyes showing weak staining in the lens epithelium, which was equal in wild-type (E) and Tg (F), and strong staining in the bow region, which persisted through the early stages of fiber differentiation in wild-type, but vanished in Tg(Cryaa-Foxe3).

Figure 4.

 

Correlation between ratios of values for mRNA concentration from Tg(Cryaa-Foxe3) and wild-type, obtained from microarray (ordinate) and Q-PCR (abscissa). Correlation coefficient, r = 0.95. The genes tested and the mRNA concentration values are found in Table 1 . For primer sequences, refer to Supplementary Table S2.

Table 1.

 

 

Fold change [Tg(Cryaa-Foxe3) vs. wt] of mRNA Levels Determined by Microarray and Q-PCR for 16 Genes

Table 1.

 

Fold change [Tg(Cryaa-Foxe3) vs. wt] of mRNA Levels Determined by Microarray and Q-PCR for 16 Genes

Gene	x-Fold Change
		Array	Q-PCR
Bfsp1	−3.3	−13.5
Calm3	1.0	1.1
Casp7	−4.2	−3.8
Ctgf	2.3	1.6
Dlad	−4.8	−5.4
Gadd45b	−4.6	−5.5
Jun	1.3	2.0
Map2k1	1.2	1.0
Map3k5	−4.2	−6.7
Psmc1	1.4	−1.5
Stat1	−1.7	−1.8
Tnfrsf19	1.1	1.5
Tnfrsf22	2.0	2.8
Tnfrsf23	2.5	2.2
Uox	−3.7	−6.5
Wnt5a	1.2	−1.1

 A graphic presentation of these results is shown in Figure 4 . For primer sequences, see Supplementary Table S2.

Figure 5.

 

Immunohistochemistry and immunofluorescence illustrating differentiation defects in P6 Tg(Cryaa-Foxe3) lenses. (A, B) Connexin-50 (C-terminal part) immunostaining was concentrated in the subepithelial zone of the fiber compartment in Tg(Cryaa-Foxe3). The central, nonimmunoreactive zone, where the C-terminal part of connexin-50 had been cleaved off, was absent in the transgenic lens. (C, D) Weaker staining in the interior of the Tg(Cryaa-Foxe3) lens with a pan-γ-crystallin antibody. (E, F) Persistence of the epithelial/cortical fiber pattern of N-cadherin throughout the Tg(Cryaa-Foxe3) lens.

Figure 6.

 

Lack of an organelle-free zone in Tg(Cryaa-Foxe3). Sections of eyes from 2-day-old wild-type (A, C, E, G) and Tg(Cryaa-Foxe3) (B, D, F, H) mice showing the organelle-free zone in the center of the wild-type lens (A, C), but persistent nuclei in Tg(Cryaa-Foxe3) (B, D). Immunohistochemistry with anti-ATP-synthase (E–H) showing absence of mitochondria in the center of the wild-type lens (E, G), but not in lenses from Tg(Cryaa-Foxe3) (F, H).

Figure 7.

 

Simplified summary of the interrelationships between some of the genes with altered expression in Tg(Cryaa-Foxe3) versus wild-type. Up arrowheads, increased, and down arrowheads, decreased, mRNA levels. Genes involved in signal transduction are contained within the dashed line, and boxes delineate genes that belong to the same pathway.

 

Supplementary Materials

Supplementary Table S1 - (.xls) 

Supplementary Table S2 - (.xls) 

The authors thank Zbynek Kozmik (Institute of Molecular Genetics, Prague, Czech Republic) for the Cryaa construct.