ABSTRACT
To determine whether fibroblast growth factor (FGF) has a role in lens development, we have generated transgenic mice expressing a dominant-negative form of the murine FGF receptor-1 (FGFRDN) in the lens. Using the fibre cell- specific αA-crystallin promoter to express the FGFRDN, we have asked whether FGF is required for fibre cell differentiation. The transgenic mice display diminished differentiation of fibre cells as indicated by their reduced elongation. In addition, transgenic lenses have an unusual refractile anomaly that morphological and biochemical data show results from the apoptosis of fibre cells in the central region of the lens. These results show that lens fibre cells are dependent on FGF for their survival and differentiation, and demonstrate that growth factor deprivation in vivo can lead to apoptosis.
INTRODUCTION
Development of the vertebrate lens requires the differentiation of epithelial cells into fibre cells (Mann, 1928). Classical lens rotation experiments performed in the chick (Coulombre and Coulombre, 1963) and mouse (Yamamoto, 1976), showed that the local environment plays a key role in the differentiation of lens fibre cells and suggested the presence of an inducer of lens fibre-cell differentiation in the vitreous surrounding the posterior lens hemisphere. FGF-1 (acidic FGF) and FGF-2 (basic FGF) became candidates for fibre cell differentiation stimuli with the demonstration that a heparin binding factor present in the eye, or recombinant FGF-1 and FGF-2, could induce all aspects of fibre cell differentiation in vitro including cell elongation and synthesis of the fibre-cell-specific β-crystallin and γ-crystallin (for reveiw see McAvoy et al., 1991). Furthermore, immunodetection showed FGF-1 and FGF-2 in the eye in locations consistent with a role in fibre cell differentiation (de Iongh and McAvoy, 1992) and an artificially secreted form of FGF-1 caused a differentiative response in lens epithelial cells (Robinson et al., 1995). Arguably the most illuminating analysis implicating FGF as an inducer of lens polarization was the demonstration that proliferation, migration and differentiation of lens epithelial cells were optimally stimulated by concentrations of FGF that sequentially increased (McAvoy and Chamberlain, 1989). As these cellular processes represent the three main elements of lens development, the data imply that a decreasing posterior to anterior gradient of FGF in the eye might result in a polarized structure such as the lens.
The proliferative and differentiative responses to FGF in lens progenitor cells in vitro illustrate a general feature of growth factor action. Another example of such activity is the ability of the haemopoietic growth factors to stimulate both the proliferation and differentiation of progenitors along specific lineages (Metcalf, 1985). In addition, in vitro studies have shown that growth factors promote cell survival and prevent apoptosis (Williams et al., 1990). However, studies of growth factor deficient mice (Mann et al., 1993; Mansour et al., 1993; Cecchini et al., 1994; Lieschke et al., 1994; Feldman et al., 1995) have failed to show apoptosis-suppressing activity of growth factors in vivo.
Members of the FGF family of growth factors have been implicated in the development of a variety of tissues besides the lens. Four FGF receptors as well as their known ligands are expressed widely throughout embryonic development and two methods have been favoured to assess their function. First, mice with targeted deletions of either FGF or FGF receptor genes have demonstrated a role for FGF in early development and a variety of aspects of organogenesis (Mansour et al., 1993; Hebert et al., 1994; Yamaguchi et al., 1994; Feldman et al., 1995). Second, transgenic mice expressing truncated, dominant-negative forms of FGF receptors have revealed a role for FGFs in development of suprabasal keratinocytes (Werner et al., 1993) and in branching morphogenesis of the lung (Peters et al., 1994). The use of tissue-specific dominantnegative FGF receptors as probes for FGF function during development has the advantage of inhibiting the action of all FGF receptors (Ueno et al., 1992; Li et al., 1994) as well as avoiding early developmental lethals (Yamaguchi et al., 1994).
To determine whether an FGF family member might induce fibre cell differentiation in vivo, we have generated transgenic mice that express a dominant-negative form of the FGFR1 under the control of the lens-specific αA-crystallin promoter. We show here that these transgenic mice are characterized by lens fibre cells that are perturbed in differentiation and ultimately apoptose. These data indicate that in vivo FGF is required for fibre cell differentiation and survival.
MATERIALS AND METHODS
Generation of transgenics
A restriction fragment encoding the murine dominant-negative FGF receptor 1 (Bernard et al., 1991) (truncated downstream of the transmembrane domain) was cloned into the pαASPLONO polylinker. pαASPLONO is a version of pGEM4z (Promega) modified to include (1) NotI restriction sites at the polylinker boundaries, (2) the αA-crystallin promoter (the 409 bp BamHI to BglII fragment) (Chapelinsky et al., 1985) adjacent to the HindIII restriction site and (3) the SV40 splice and polyadenylation sequences between the KpnI and EcoRI sites. The resulting transcription unit is described in Fig. 1A. Transgenic FVB/N (Taketo et al., 1991) mice were generated and screened according to established methods (Hogan et al., 1986).
Structure and expression pattern of the α.FGFRDN transgene. (A) Structure of the α.FGFRDN transgene. The αA-crystallin promoter is placed upstream of the truncated murine FGFR1. The transcription start point is indicated by the small, right-facing arrow. Specific coding regions of the receptor are indicated; S, signal sequence; IG1-3, immunoglobulin-like domains; TM, transmembrane domain; ATG/TGA, start and stop codons respectively; AB, acidic box of unknown function. Splicing and polyadenylation (pA) signals are provided by sequences from the SV40 small t antigen gene. The transgene fragment is flanked by NotI restriction sites. (B) Northern blotting analysis of α.FGFRDN-7 and α.FGFRDN-22 transgene expression. mRNA was isolated from the tissues and transgenic lines indicated and hybridized with a probe to the murine FGFR1. Eye mRNA samples were prepared from dissected lenses and from the remaining eye tissue (eye- (lens)). The predominant endogenous FGFR1 mRNA appears at approximately 3.6 kb and the transgene mRNA at approximately 2 kb. The panels labelled GAPDH show the northern filters re-hybridized with a probe to the glyceraldehyde-3-phosphate dehydrogenase mRNA to give a measure of loading. (C-G) In situ localization of transgene and αA-crystallin mRNAs. Eye sections from homozygous E16.5 α.FGFRDN-22 transgenic mice hybridized with an antisense probe to SV40 sequences in the transgene mRNA, shown in dark-field (C) and bright-field (D) and an antisense probe to αA-crystallin on a section from the same lens (E). Control E16.5 FVB/N mouse eye sections hybridized with the same probe used for transgenic mouse sections in C in dark-field (F) and bright-field (G) give an indication of background. The transgene (C) and αA-crystallin (E) probe examples were exposed for 7 days, and the control for 11 days. Counterstain is hematoxylin, magnifications 150×.
Structure and expression pattern of the α.FGFRDN transgene. (A) Structure of the α.FGFRDN transgene. The αA-crystallin promoter is placed upstream of the truncated murine FGFR1. The transcription start point is indicated by the small, right-facing arrow. Specific coding regions of the receptor are indicated; S, signal sequence; IG1-3, immunoglobulin-like domains; TM, transmembrane domain; ATG/TGA, start and stop codons respectively; AB, acidic box of unknown function. Splicing and polyadenylation (pA) signals are provided by sequences from the SV40 small t antigen gene. The transgene fragment is flanked by NotI restriction sites. (B) Northern blotting analysis of α.FGFRDN-7 and α.FGFRDN-22 transgene expression. mRNA was isolated from the tissues and transgenic lines indicated and hybridized with a probe to the murine FGFR1. Eye mRNA samples were prepared from dissected lenses and from the remaining eye tissue (eye- (lens)). The predominant endogenous FGFR1 mRNA appears at approximately 3.6 kb and the transgene mRNA at approximately 2 kb. The panels labelled GAPDH show the northern filters re-hybridized with a probe to the glyceraldehyde-3-phosphate dehydrogenase mRNA to give a measure of loading. (C-G) In situ localization of transgene and αA-crystallin mRNAs. Eye sections from homozygous E16.5 α.FGFRDN-22 transgenic mice hybridized with an antisense probe to SV40 sequences in the transgene mRNA, shown in dark-field (C) and bright-field (D) and an antisense probe to αA-crystallin on a section from the same lens (E). Control E16.5 FVB/N mouse eye sections hybridized with the same probe used for transgenic mouse sections in C in dark-field (F) and bright-field (G) give an indication of background. The transgene (C) and αA-crystallin (E) probe examples were exposed for 7 days, and the control for 11 days. Counterstain is hematoxylin, magnifications 150×.
Transcription analysis
Poly(A)+ RNA was purified from tissues of 3 to 4 week old animals with oligo(dT)-cellulose (Boehringer Mannheim), electrophoresed through 1% formaldehyde gels and processed for northern blotting using standard techniques (Gonda et al., 1982). Northern filters were hybridized with random primed DNA probes labelled with α-[32P]-dCTP. Probes were generated using cDNA fragment templates encoding the truncated FGFR1 and rat GAPDH (glyceraldehyde 3-phosphate dehydrogenase).
In situ hybridization
In situ hybridizations were performed according to Van Leen et al. (1987) with the exception that 33P-UTP-labelled riboprobes were used and that an RNase A washing step (20 μg/ml in 500 mM NaCl, 10 mM Tris (pH 8.0) at 37°C for 15 minutes) was employed to reduce non-specific probe adherence.
FGF receptor cross-linking
Lenses from 3-4 week old animals were solubilized on ice using a Dounce homogenizer and 0.5 ml of lysis buffer (0.5% Triton X-100, 150 mM NaCl, 1.5 mM MgCl2, 10% glycerol, 20 mM Hepes-NaOH (pH 7.5)) with protease inhibitors and centrifuged at 100,000 g at 4°C for 20 minutes. 125I-labelled FGF-2 (specific activity = 7.3×104 cts/minute/ng), was added to 5.6 mg of lysate to a final concentration of 44 ng/ml and rocked gently at 4°C for 1 hour. Control lysates were prepared from a CHO1-3 cell line that expresses a 2 Ig-like domain, soluble form of the murine FGFR1 (Roghani et al., 1994). The reaction was competed using a 377-fold excess of unlabelled FGF-2. Products were cross-linked (Roghani et al., 1994) and concentrated on heparin sepharose beads (Pharmacia). Samples were resolved on 7% SDS-reducing polyacrylamide gels and analyzed on a PhosphorImager (Molecular Dynamics) using a linear scale range.
FGF-2 mediated phosphorylation of ERK-1 and ERK-2 in whole lenses
Day of birth (DOB) wild-type and transgenic lenses were explanted and starved for 16 hours in DMEM containing 0.05% fetal calf serum. Sets of 4 whole lenses were transferred to DMEM with or without 50 ng/ml human FGF-2 and incubated at 37°C for 20 minutes. Using watchmakers forceps, lens capsules (and adhering epithelial cells) were removed from the underlying fibre cell mass and the separated components sonicated on ice in 1% Triton X-100, 150 mM NaCl, 20 mM Hepes (pH 7.5), 10% glycerol, 1 mM EDTA, 1 mM EGTA, 100 μM sodium fluoride, 10 μM tetrasodium pyrophosphate, 200 μM sodium orthovanadate and protease inhibitors. 170 μg lysate from fibre cells (one lens equivalent) and 5 μg lysate from capsules (one lens equivalent) were analyzed on 10% reducing SDS-polyacrylamide gels and transferred to nitrocellulose membranes (Micron Separations, Inc.). Positive control lysates (5 μg) were prepared from unstimulated and FGF stimulated mouse fibroblasts (gift of I. Lax). Immunoblotting of membranes was performed according to standard procedures (Harlow and Lane, 1988) using Renaissance chemiluminescence visualizion reagents (Dupont). Protein A-HRP (Kirkegaard and Perry Laboratories Inc.) was used at a 1:5000 dilution. Blots were stripped in 65 mM Tris (pH 8), 2% SDS and 100 mM β-mercaptoethanol at 60°C for 30 minutes, reblocked and then probed with anti-ERK2 antibodies (that cross-react with ERK1) (gift of I. Dikic and J. Schlessinger) at a 1:500 dilution.
Histological analysis
Tissues for histological analysis included staged mouse embryos or whole eyes from postnatal animals. Tissue samples were prepared and stained either with hematoxylin and eosin (Fig. 4) or only hematoxylin (Fig. 5) using conventional methods (Culling et al., 1985). Figures in this paper were prepared digitally using a ProGres 3012 digital camera, a Nikon slide scanner and Adobe Photoshop and Quark Express production software.
FGF receptors and perturbed signal transduction in α.FGFRDN mice. (A) Cross-linking of dominant-negative FGF receptors with 125I-FGF2. Lysates were prepared from CHO1-3 cells (expressing the 2 immunoglobulin-like domain, soluble form of FGFR1) and from the lenses of transgenic and negative littermate mice. The crosslinked 75×103 Mr soluble receptor- FGF-2 complex from CHO1-3 cells and the ∼110×103 Mr dominant-negative receptor-FGF complex (R) from transgenic lenses was detected but both eliminated by unlabelled FGF-2 competition. (B) Anti-phosphotyrosine immunoblotting of lysates from FGF-stimulated whole lenses. Whole lenses from DOB wild-type and α.FGFRDN-22 mice were starved and stimulated with FGF-2. Lysates were prepared from capsule (with adherent lens epithelial cells) and fibre cells separately and analyzed by anti-phosphotyrosine western blotting. Unstimulated and FGF-2 stimulated mouse fibroblasts served as controls. This analysis revealed a doublet migrating at 41 and 44×103 Mr (asterisk) that appeared only upon FGF-2 stimulation. Equal levels of phosphotyrosine are observed in the 41/44×103 Mr doublet in FGF-2-stimulated lens epithelial cells. A reduction in the level of phosphotyrosine is observed in the 41/44×103 Mr doublet from fibre cells of transgenic lenses relative to wild type. Cross-reactivity to lens crystallins can be observed in lysates from fibre cells (CR). (C) Western blot detection of MAP kinase family members ERK1 and ERK2 in lens lysates. The western blot from Fig. 2B was stripped and re- probed with polyclonal antibodies directed against ERK2 (that cross-react with ERK1). This revealed species migrating at 41 (ERK2) and 44×103 Mr (ERK1) that correspond to the doublet observed in Fig. 2B. Levels of ERK1 and ERK2 are identical in all fibre cell lysates and are observed at a lower, but consistent level in all capsule cell lysates.
FGF receptors and perturbed signal transduction in α.FGFRDN mice. (A) Cross-linking of dominant-negative FGF receptors with 125I-FGF2. Lysates were prepared from CHO1-3 cells (expressing the 2 immunoglobulin-like domain, soluble form of FGFR1) and from the lenses of transgenic and negative littermate mice. The crosslinked 75×103 Mr soluble receptor- FGF-2 complex from CHO1-3 cells and the ∼110×103 Mr dominant-negative receptor-FGF complex (R) from transgenic lenses was detected but both eliminated by unlabelled FGF-2 competition. (B) Anti-phosphotyrosine immunoblotting of lysates from FGF-stimulated whole lenses. Whole lenses from DOB wild-type and α.FGFRDN-22 mice were starved and stimulated with FGF-2. Lysates were prepared from capsule (with adherent lens epithelial cells) and fibre cells separately and analyzed by anti-phosphotyrosine western blotting. Unstimulated and FGF-2 stimulated mouse fibroblasts served as controls. This analysis revealed a doublet migrating at 41 and 44×103 Mr (asterisk) that appeared only upon FGF-2 stimulation. Equal levels of phosphotyrosine are observed in the 41/44×103 Mr doublet in FGF-2-stimulated lens epithelial cells. A reduction in the level of phosphotyrosine is observed in the 41/44×103 Mr doublet from fibre cells of transgenic lenses relative to wild type. Cross-reactivity to lens crystallins can be observed in lysates from fibre cells (CR). (C) Western blot detection of MAP kinase family members ERK1 and ERK2 in lens lysates. The western blot from Fig. 2B was stripped and re- probed with polyclonal antibodies directed against ERK2 (that cross-react with ERK1). This revealed species migrating at 41 (ERK2) and 44×103 Mr (ERK1) that correspond to the doublet observed in Fig. 2B. Levels of ERK1 and ERK2 are identical in all fibre cell lysates and are observed at a lower, but consistent level in all capsule cell lysates.
Refractile anomaly and focal cataract in α.FGFRDN mice. Normal (A,D) α.FGFRDN-22 heterozygous (B,E) and α.FGFRDN-22 homozygous (C,F) lenses shown in anterior (A-C) and lateral (D-F) view. The boundary of the refractile anomaly present in the central region of transgenic lenses is indicated by the short arrow and the focal cataract present in some homozygous lenses is indicated by the long arrow. Images were obtained by photographing the lenses in transmission illumination under a low power microscope after they had been dissected free of surrounding tissue. All magnifications are 22×.
Refractile anomaly and focal cataract in α.FGFRDN mice. Normal (A,D) α.FGFRDN-22 heterozygous (B,E) and α.FGFRDN-22 homozygous (C,F) lenses shown in anterior (A-C) and lateral (D-F) view. The boundary of the refractile anomaly present in the central region of transgenic lenses is indicated by the short arrow and the focal cataract present in some homozygous lenses is indicated by the long arrow. Images were obtained by photographing the lenses in transmission illumination under a low power microscope after they had been dissected free of surrounding tissue. All magnifications are 22×.
Histological analysis. (A) E13.5 wild-type eye (200×). (B) E13.5 α.FGFRDN-22 homozygous eye showing epithelium separated (asterisk) from the pfc mass (200 ×). (C) Wild-type lens at the DOB showing sfc in contact with the lens epithelium. The pfc are apparent in the central lens and are surrounded by sfc (100 ×). (D) α.FGFRDN-22 homozygous lens at the DOB. Separation between lens epithelium and fibre cell mass equator to equator is evident (asterisks) and a central region of degeneration manifests as a pale area (100×). (E) Equatorial lens (400 ×) from α.FGFRDN-22 homozygous mouse at the DOB showing separation of epithelium and anterior surface of the fibre cell mass (arrows). The pale staining central region of degeneration and a fissure that has its origin in a suture line are also apparent. (F) Wild-type lens at 2 weeks after birth (50×). (G) α.FGFRDN-22 heterozygous lens at two weeks with mild refractile anomaly and absence of focal cataract. The section shows the pale central region of degeneration inside normally staining SFCs and separation of lens epithelium from the primary the fibre cell mass (asterisk) (50×). (H) α.FGFRDN-22 homozygous lens at three weeks with both refractile anomaly and focal cataract. The central region of degeneration is apparent inside normally staining SFCs. The distance from the capsule to the beginning of the degenerated region is less than in 2 week lens (G) (50×). (I) Section from wild-type lens at three weeks of age showing MIP26 antibody labelling at the fibre cell boundaries and regular stacking pattern extending centrally (400×). (J) Differential interference contrast (DIC) image of the section shown in I (400x). (K) Section fromα. FGFRDN-22 homozygous lens at three weeks of age showing fibre cell boundaries labelled with MIP26 antibody. The regular stacking pattern extends centrally only a short distance (400×). (L) DIC image of the section stained in K, showing that the section is intact despite the lack of MIP26 labelling (400×). (M) MIP26 labelling of fibre cell boundaries in another example of a 3 weekα.FGFRDN-22 homozygous lens showing a disorganized fibre cell stacking pattern and large cell size (400×). In all panels, the following abbreviations are used: c, capsule; ca, cornea; dr, degenerated region; e, lens epithelium; fc, focal cataract; pfc, primary fibre cells; r, retina; s; suture line fissure; sfc, secondary fibre cells; v, vitreous. All sections are 4μm in thickness, hematoxylin and eosin stained from paraffin embedded, Petrunkewitz’s fluid fixed tissue. Original magnifications are listed in brackets.
Histological analysis. (A) E13.5 wild-type eye (200×). (B) E13.5 α.FGFRDN-22 homozygous eye showing epithelium separated (asterisk) from the pfc mass (200 ×). (C) Wild-type lens at the DOB showing sfc in contact with the lens epithelium. The pfc are apparent in the central lens and are surrounded by sfc (100 ×). (D) α.FGFRDN-22 homozygous lens at the DOB. Separation between lens epithelium and fibre cell mass equator to equator is evident (asterisks) and a central region of degeneration manifests as a pale area (100×). (E) Equatorial lens (400 ×) from α.FGFRDN-22 homozygous mouse at the DOB showing separation of epithelium and anterior surface of the fibre cell mass (arrows). The pale staining central region of degeneration and a fissure that has its origin in a suture line are also apparent. (F) Wild-type lens at 2 weeks after birth (50×). (G) α.FGFRDN-22 heterozygous lens at two weeks with mild refractile anomaly and absence of focal cataract. The section shows the pale central region of degeneration inside normally staining SFCs and separation of lens epithelium from the primary the fibre cell mass (asterisk) (50×). (H) α.FGFRDN-22 homozygous lens at three weeks with both refractile anomaly and focal cataract. The central region of degeneration is apparent inside normally staining SFCs. The distance from the capsule to the beginning of the degenerated region is less than in 2 week lens (G) (50×). (I) Section from wild-type lens at three weeks of age showing MIP26 antibody labelling at the fibre cell boundaries and regular stacking pattern extending centrally (400×). (J) Differential interference contrast (DIC) image of the section shown in I (400x). (K) Section fromα. FGFRDN-22 homozygous lens at three weeks of age showing fibre cell boundaries labelled with MIP26 antibody. The regular stacking pattern extends centrally only a short distance (400×). (L) DIC image of the section stained in K, showing that the section is intact despite the lack of MIP26 labelling (400×). (M) MIP26 labelling of fibre cell boundaries in another example of a 3 weekα.FGFRDN-22 homozygous lens showing a disorganized fibre cell stacking pattern and large cell size (400×). In all panels, the following abbreviations are used: c, capsule; ca, cornea; dr, degenerated region; e, lens epithelium; fc, focal cataract; pfc, primary fibre cells; r, retina; s; suture line fissure; sfc, secondary fibre cells; v, vitreous. All sections are 4μm in thickness, hematoxylin and eosin stained from paraffin embedded, Petrunkewitz’s fluid fixed tissue. Original magnifications are listed in brackets.
Apoptosis in α.FGFRDN lenses. (A) Equatorial region of a 3 week, wild-type lens showing the characteristic distribution of nuclei through the epithelium (e), transitional zone (tz) and in the secondary fibre cells (sfc) (200×). (B) Equator of a 3 week, α.FGFRDN-22 homozygous lens showing normal tz and peripheral sfc. More mature sfc show densely staining chromatin fragments characteristic of apoptosis (large arrows). Apoptotic bodies are also present as prominent, hematoxylin-staining objects in the degenerated region (dr) of the central lens (arrowheads) (200×). (C) Bright-field micrograph of mature apoptotic bodies (arrows) in a section from an α.FGFRDN-22 homozygous lens. The section has been TUNEL labelled (630×). (D) Fluorescence illumination of the same tissue preparation shown in C. Labelling of most apoptotic bodies is apparent (arrows) (630×). (E) Apoptotic nucleus (arrow) in a sfc in 3 week α.FGFRDN-22 homozygous lens (400×). (F) Fluorescence illumination of the same tissue preparation shown in E shows intense TUNEL labelling of apoptotic nucleus (400×). Adjacent nuclei normal in appearence are not labelled. Magnifications are listed in brackets.
Apoptosis in α.FGFRDN lenses. (A) Equatorial region of a 3 week, wild-type lens showing the characteristic distribution of nuclei through the epithelium (e), transitional zone (tz) and in the secondary fibre cells (sfc) (200×). (B) Equator of a 3 week, α.FGFRDN-22 homozygous lens showing normal tz and peripheral sfc. More mature sfc show densely staining chromatin fragments characteristic of apoptosis (large arrows). Apoptotic bodies are also present as prominent, hematoxylin-staining objects in the degenerated region (dr) of the central lens (arrowheads) (200×). (C) Bright-field micrograph of mature apoptotic bodies (arrows) in a section from an α.FGFRDN-22 homozygous lens. The section has been TUNEL labelled (630×). (D) Fluorescence illumination of the same tissue preparation shown in C. Labelling of most apoptotic bodies is apparent (arrows) (630×). (E) Apoptotic nucleus (arrow) in a sfc in 3 week α.FGFRDN-22 homozygous lens (400×). (F) Fluorescence illumination of the same tissue preparation shown in E shows intense TUNEL labelling of apoptotic nucleus (400×). Adjacent nuclei normal in appearence are not labelled. Magnifications are listed in brackets.
Immunofluorescence
Axial sections (4 μm) of paraffin embedded, Petrunkewitz’s fluid fixed, staged embryos were immunofluorescently labelled according to conventional methods (Harlow and Lane, 1988). Polyclonal rabbit antisera for MIP26 (gift of J. Horwitz) was used at a dilution of 1:200.
Visualization of fragmented DNA in situ
We have used a modified version of the TUNEL (terminal deoxynu- cleotide transferase nick-end labelling) technique (Gavrieli et al., 1992) to identify cells undergoing apoptosis in sections of the α.FGFRDN mouse lens. The procedure was performed as described previously (Lang et al., 1994) with the exception that tissues were fixed in Petrunkewitz’s fluid. TUNEL labelling is visualized either in bright-field or fluorescence microscopy (with a rhodamine filter set) as a red reaction product.
RESULTS
Generation of α.FGFRDN transgenic mice
The transgene used in these experiments (α.FGFRDN) consists of the murine αA-crystallin promoter driving transcription of a truncated, dominant-negative form of the murine FGFR1 (Fig. 1A). Expression of the αA-crystallin gene in both the embryo and adult begins with transitional zone epithelial cells and includes mature fiber cells (Tréton et al., 1991). In the adult rat lens, the boundary of αA-crystallin expression is found at coordinate position 20 (Van Leen et al., 1987), a pattern of expression that is conserved in the mouse (Tréton et al., 1991). Thus, the expression boundary of αA-crystallin in the epithelium corresponds precisely to the boundary between proliferating and differentiating cells (McAvoy, 1978). As a consequence, the α.FGFRDN transgene construct is designed to ask whether FGF is required for fibre cell differentiation.
Potential founder transgenic mice were screened for the transgene by genomic blotting analysis of tail DNA. A probe to the mouse FGFR1 revealed bands corresponding to the transgene in three out of 22 animals (data not shown). Further breeding of the transgenic animals established two transgenic lines (one potential founder did not transmit the transgene) designated α.FGFRDN-7 and α.FGFRDN-22. These were maintained in matings of both heterozygous and homozygous animals (Table 1).
Transgene mRNA expression
Northern blotting of lens mRNA (Fig. 1B) demonstrated that the transgene was expressed at a 13- to 17-fold excess over the endogenous FGFR1 in heterozygous mice of both transgenic lines (Table 1). As anticipated, neither transgene showed any sign of expression in tissues other than the lens. In situ hybridization analysis with a transgene-specific SV40 probe (Fig. 1A) revealed an αA-crystallin-like expression pattern in lens sections from E16.5 α.FGFRDN-22 mice (Fig. 1C-E). The αAcrystallin pattern of expression excludes almost all epithelial cells (Van Leen et al., 1987). Those epithelial cells that do express αA-crystallin are closest to the transitional zone at the lens equator where fibre cell differentiation begins. The approximate border between non-expressing and expressing cells is marked on one side of the lens for both the transgene-specific probe (Fig. 1C, arrow) and for the αA-crystallin probe (Fig. 1E, arrow). Control lens sections from wild-type animals did not show significant hybridization (Fig. 1F,G).
Dominant-negative FGF receptor expression
The dominant-negative form of FGFR1 was detected by cross-linking of lens lysates to iodinated human FGF-2 (Roghani et al., 1994). The FGFRDN/125I-FGF-2 complex displayed the expected relative molecular mass of 110×?? 3 (Fig. 2,R) and the appearance of this species was eliminated by competition with unlabelled FGF-2. CHO1-3 cells expressing a soluble form of the FGFR1 with two immunoglobulin-like domains were used as positive controls and showed the expected 75×103Mr cross-linked complex. Endogenous FGF receptors were not readily detected by the cross-linking method but FGFR1 could be detected in whole lens lysates using an immunoprecipitation-western blotting analysis (R. L. C. and R. A. L., unpublished). These analyses confirmed the expression of endogenous FGFR1 in the lens (Orr-Urtreger et al., 1991) and demonstrated, through quantitative analysis of cross-linking gels, an excess of the dominantnegative receptor of at least 20-fold relative to endogenous receptors in heterozygous transgenic mice (Table 1). Homozygous transgenic animals may be expected to express the dominant-negative receptor at a 40-fold excess.
To confirm that the FGFRDN detected in the lens perturbed FGF signal transduction, we examined the FGF-stimulated phosphorylation of MAP kinase family members ERK1 and ERK2 (extracellular regulated kinases), components of the FGF signal transduction pathway (Cobb et al., 1991). Control and transgenic (α.FGFRDN-22 homozygous) lysates were analyzed by SDS-PAGE and western blotting using anti-phosphotyrosine antibodies (Fig. 2B). Typically, FGF-stimulated fibre and epithelial cells from wild-type lenses showed a large increase in phosphorylation of 44 and 41×103Mr species characteristic of ERK1 and ERK2 (Cobb et al., 1991). Lens epithelial cells from wild-type and transgenic mice showed a similar level of ERK1 and ERK2 phosphorylation as would be anticipated given their lack of transgene expression. By contrast, fibre cells from α.FGFRDN-22 transgenic mice showed reproducibly decreased ERK1 and ERK2 phosphorylation. As a control for the amount of ERK1 and ERK2 protein in lysates, filters were stripped and reprobed with antibodies to ERK2 that cross-react with ERK1 (Fig. 2C). This showed equal quantities of ERK1 and ERK2 in lens fibre lysates and a lower, but consistent level in lysates from the epithelial layers. Quantitation of fibre cell ERK1 and ERK2 band intensities indicated a 3- to 4-fold reduction in ERK1 and ERK2 phosphorylation. These data confirm a reduced level of ERK1 and ERK2 phosphorylation and indicate perturbed FGF signal transduction in the α.FGFRDN-22 lens fibre cells.
Transgenic lenses have an unusual refractile anomaly
Lenses dissected from the eyes of 4 week old transgenic mice revealed significant alterations in lens structure (Fig. 3). A refractile anomaly absent from wild-type lenses (Fig. 3A/D) was observed in the lenses of transgenic animals (Fig. 3B,C,E,F, short arrow) and was due to a central, approximately spherical region of the lens matrix that appeared distinct in density from the peripheral matrix.. The refractile anomaly was not readily observed in lenses from newborn transgenic mice but developed progressively thereafter. This pattern of development of the refractile anomaly correlated with the appearance of a degenerating central region of the lens that could be detected in histological analyses (Fig. 4E,G,H). In addition to the refractile anomaly, we observed focal cataracts (Fig. 3C,F, long arrow) in 29% of homozygous α.FGFRDN-22 and a small percentage of heterozygous α.FGFRDN-7 transgenic mice (Table 1). We presume that the more severe phenotype in homozygous compared with heterozygous transgenic animals (Fig. 3B/E, C/F; Table 1) is a response to doubling the expression level of the dominant-negative receptor.
Differentiation of lens fibre cells is perturbed
Histological examination of transgenic lenses indicates that neither primary nor secondary lens fibre cells elongate to the expected extent. Primary lens fibre cells are derived from precursor cells of the posterior hemisphere of the lens vesicle. These cells begin differentiating and elongate into lens fibre cells at 11.5 to 12 days of embryonic development (Kaufman, 1992). Secondary fibre cells are derived from lens epithelial cells that migrate to the lens equator, re-orient and elongate to fill the major volume of the lens (Kaufman, 1992). In wild- type lenses at E13.5, primary fibre cells have made contact with the anterior lens epithelium (Fig. 4A). In lenses from homozygous α.FGFRDN-22 mice of the same age, the primary fibre cells have not elongated fully. This results in a gap between the anterior end of lens fibre cells and the posterior surface of the lens epithelium (Fig. 4B, asterisk). This frequent, but not fully penetrant feature of transgenic lenses, is also apparent at the DOB in α.FGFRDN-22 homozygotes (Fig. 4D, asterisks) and is not observed in wild-type lenses (Fig. 4C). Given the variation existing between animals, we cannot rule out the possibility that fibre cellepithelial separation is a transient feature. When transgenic lenses are viewed at higher magnification, the anterior border of the fibre cell mass is clearly separated from the posterior surface of the lens epithelium (Fig. 4E, double-headed arrows). The exposed anterior boundary of peripheral fibre cells (Fig. 4E, arrowhead) suggests that, in the transgenic lenses, contact between fibre cells and the epithelium is never established. In some cases, no contact is established between the fibre cell mass and the posterior surface of the epithelium at any point between the transitional zone and the central epithelium (Fig. 4D).
To confirm our morphological assessment of reduced fibre cell elongation, we quantitated fibre cell length in the transgenic lenses at the DOB. This was done on the largest axial lens section of a continuous series (to ensure a section through the lens center) using a microscope eyepiece reticle. Although the DOB equatorial diameter was not altered (wild type, 9.27+0.40 (n=3); α.FGFRDN-22 homozygotes, 9.38+0.59 (n=5) (arbitrary units)), the distance from posterior pole to the anterior surface of the fibre cell mass was significantly reduced in the transgenic lenses (wild type, 7.83+0.15 (n=3); α.FGFRDN-22 homozygotes, 6.94+0.28 (n=5) (arbitrary units)). This is not a direct measure of individual fibre cell length because they curve through the fibre cell mass, but does reflect the degree of elongation. At all ages, the epithelial layer of the transgenic lens appears normal in structure except for its separation from the fibre cell mass (Fig. 4B,D,E,G,H). Lens sutures were abnormally prominent in some lenses and probably reflected reduced ability of fibre cell termini to converge during suture formation (Fig. 4E,H). The diminished fibre cell elongation observed in the transgenic lenses indicates that differentiation of these cells has been perturbed (Coulombre and Coulombre, 1963; Piatigorsky et al., 1972; McAvoy, 1978).
Another feature of transgenic lenses is a central region that, by contrast with wild-type lenses (Fig. 4C (newborn), F (week 2)), does not stain normally (Fig. 4D,E,G,H). This region is discernible in some transgenic (Fig. 4D) but not wild-type (Fig. 4C) lenses at the DOB and develops progressively in the postnatal period to the point where it occupies most of the lens volume at week 2 (Fig. 4G) and week 3 (Fig. 4H). The abnormally staining central lens region is surrounded by secondary lens fibres that are normal in both appearance and eosin staining intensity (Fig. 4D,E,G,H). It was apparent that the boundary between the central region and normal secondary fibre cells (Fig. 4G,H, arrowheads) was the source of the refractile anomaly observed in unfixed, dissected lenses (Fig. 3, large arrows). Focal cataracts observed in unfixed dissected lenses (Fig. 3C,F) were apparent in histological sections as a central, eosinophilic area (Fig. 4H).
To investigate the origin of the abnormal central lens region, we performed immunofluorescence analysis with antibodies to the major fibre cell membrane protein MIP26 (Shiels et al., 1991). This allowed us to examine the organization of the fibre cells and showed that, unlike normal lenses where the stacking pattern of fibre cells is regular and continues some distance from the lens capsule (Fig. 4I,J), fibre cell membranes in 2 or 3 week transgenic lenses label only in a peripheral region (Fig. 4K) despite the existence of lens tissue throughout the section (Fig. 4L). Furthermore, in some examples, the pattern of fibre cell boundaries becomes disorganized adjacent to the region of no staining, and individual cells appear to occupy a larger volume than usual (Fig. 4M). The boundary between MIP26- positive and -negative regions corresponds to the boundary of the refractile anomaly observed in unfixed, dissected transgenic lenses (Fig. 3). Histological and immunofluorescence analysis suggested that both primary and secondary fibre cells in the central region of the lens had degenerated and that this degeneration developed progressively beginning approximately at the DOB.
Lens fibre cells ultimately apoptose
We investigated the possibility that fibre cells in the central region of the lens might be dying apoptotically. One feature typical of apoptosis is the condensation of chromatin and its fragmentation into so-called apoptotic bodies (Wyllie et al., 1980). These are recognized histologically as objects that stain dark blue with hematoxylin and are of characteristic, subnuclear size. Apoptotic bodies were not observed in the lenses of normal mice (Fig. 5A) but were present in histological sections of both heterozygous and homozygous α.FGFRDN-7 and α.FGFRDN-22 lenses (Fig. 5B,C). The morphologically typical chromatin-containing apoptotic bodies were found at the boundary between normal and degenerated secondary fibre cells adjacent to the nuclei of the lens bow (Fig. 5B, arrowheads) and were numerous in the lightly staining central region (Fig. 5B, small arrows). Apoptotic bodies and nuclei with the morphology of an early stage of apoptosis were also distributed sparsely in other areas (data not shown). Cell death in lens fibre cells was observed in both heterozygous and homozygous transgenic animals regardless of whether epithelial-fibre cell contact was established.
The TUNEL technique employs terminal deoxy-transferase and biotinylated dUTP to label the DNA fragments that are produced during apoptosis (Wyllie et al., 1980; Gavrieli et al., 1992). We used a modification of this technique (Lang et al., 1994) to confirm in histological sections that fibre cells were undergoing apoptosis. Many of the apoptotic bodies that could be found in the transgenic lenses (Fig. 5C, arrows) labelled with the TUNEL method (Fig. 5D, arrows). In addition, while normal nuclei in secondary fibre cells of the transgenic lens remained unlabelled, some nuclei of abnormal morphology labelled strongly with the TUNEL method (Fig. 5E,F) and probably represented an earlier stage in the apoptotic process. Elimination of the dUTP-biotin from the procedure eliminated the labelling of apoptotic bodies and no TUNEL labelling was observed in wild-type lenses. The nuclear degeneration that is a characteristic of normal fibre cells (Sanwal et al., 1986) is quite distinct in morphology from the apoptotic changes observed in the α.FGFRDN lenses and argues that apoptosis is a feature of the transgenic lens only.
DISCUSSION
In the experiments presented, we have addressed directly the question of whether FGF is required for lens fibre cell differentiation, one component of the establishment of lens polarity. A role for FGF in establishing a polarized lens had been suggested by results from previous in vitro (McAvoy et al., 1991) and in vivo experimentation (Robinson et al., 1995). Our observations were based on in vivo analysis of lens development in mice expressing a dominant-negative FGFR1 specifically within the differentiating cells of the lens. These experiments suggest that FGF stimulation is essential for normal differentiation of lens fibre cells and for fibre cells to escape apoptosis.
Our contention that the FGFRDN of transgene origin perturbs FGF signal transduction is supported by the large excess of the dominant negative FGF receptor that was detected in the transgenic lens (Ueno et al., 1992; Li et al., 1994) and, most important, by diminished FGF-mediated MAP kinase phosphorylation. The unique nature of the phenotype also argues for a specific perturbation of FGF signaling. The combination of diminished fibre cell elongation and a refractile anomaly caused by progressive fibre cell apoptosis has not been observed in a number of experiments in which cytokines (Egwuagu et al., 1994), toxins (Breitman et al., 1989) and growth factors (Desci et al., 1994; Robinson et al., 1995) have been expressed from the αA-crystallin promoter.
Diminished differentiation of lens fibre cells
That lens fibre cell differentiation is diminished in the α.FGFRDN transgenic mice is obvious from both morphological and quantitative assessment of fibre cell elongation. Lens fibre elongation is a primary differentiative response (Coulombre and Coulombre, 1963; Piatigorsky et al., 1972; McAvoy, 1978) and has been used as a quantitative measure of differentiation in vitro (Beebe et al., 1980). The histological analysis that we present shows that in homozygous α.FGFRDN transgenic mice, fibre cell length is noticeably diminished; a defect that is manifest most obviously as a lack of contact between the lens epithelium and the anterior surface of the fibre cell mass.
Interestingly, fibre cells in the lenses of α.FGFRDN transgenic mice are perturbed but not completely blocked in differentiation. One explanation is that the transgene may not provide sufficient dominant-negative FGF receptors to completely block FGF signal transduction. This possibility is supported by the more severe phenotype of homozygous transgenic mice. Additionally, the differentiation of fibre cells may depend on multiple stimuli, of which FGF is just one. Insulinlike growth factor type I appears to be important for fibre cell differentiation in the chick (Beebe et al., 1987) and can combine with FGF to enhance rodent fibre cell differentiation in vitro (Chamberlain et al., 1991).
Apoptosis of lens fibre cells
Both morphological and TUNEL analyses indicate that the fate of centrally located fibre cells in the α.FGFRDN lens is apoptosis. Morphological observations argue that the border between living and dead fibre cells forms the refractile anomaly that is a feature of the transgenic lens and that collapse of the internal structure of the lens is likely to result in smaller transgenic lenses at 4 weeks of age. These observations raise the issue of why apoptosis should be a consequence of over-expression of a dominant-negative FGF receptor.
One possible cause of apoptosis in fibre cells of the transgenic lens is their physical separation from the epithelium. This might result in apoptosis due to deprivation of survival factors known to support lens epithelial cells in vitro (Ishizaki et al., 1993) or due to lack of contact with extracellular matrix (Ruoslhati and Reed, 1994). However, neither of these explanations is likely given that, in some α.FGFRDN transgenic mice, fibre cell apoptosis is observed in the absence of fibre cell-epithelial separation.
The most obvious suggestion for fibre cell death is that these cells are dependent upon FGF for survival. This is consistent with evidence from in vitro experiments demonstrating that many normal cells are growth factor dependent. Examples include primary haemopoietic cells that are dependent on colony stimulating factors (Metcalf, 1985), and cells of the CNS that are dependent upon neurotrophins (Raff, 1992). Combined with the observation that fibre cells are, in part, dependent upon FGF for their differentiation, we can now argue that lens fibre cells require FGF for both differentiation and survival. This is a feature of many cell lineages, perhaps best exemplified by the action of haemopoietic growth factors such as granulocyte-macrophage colony stimulating factor (GM-CSF) in eliciting differentiation of progenitor cells and survival of mature myeloid cells (Metcalf, 1985). In many systems, including lens epithelial cells in vitro (McAvoy and Chamberlain, 1989), growth factors also stimulate proliferation. It remains for further experimentation to determine whether FGF is required for the proliferation of cells of the lens lineage in vivo.
The vitreous is a source of FGF (McAvoy et al., 1991) and presumably that required for fibre cell differentiation. If this is the case, to stimulate fibre cell survival, FGF would be required to diffuse through the lens to the central region. Thus, the border between living and dead fibre cells seen in transgenic lenses may represent the new threshold concentration for FGFstimulated survival that is imposed by the dominant-negative FGF receptor (which will have the effect of desensitizing fibre cells to the factor). Appearance of fibre cell apoptosis postnatally may reflect the increasing distance that vitreal FGF would be required to diffuse as the lens increases in size. Our observation that the central region of dead fibre cells increases in size postnatally would support this notion. In considering this question, we cannot discount the possibility that the large excess of transgene-derived FGF receptors may modify the distribution of FGF throughout the lens and have an impact on fibre cell survival as a consequence. That the dominantnegative action of truncated FGF receptors may in part result from competition for ligand is suggested by experiments in which a block to signal transduction imposed by a truncated FGF receptor can be partially relieved by increasing the amount of ligand (Li et al., 1994).
It is widely held that one of the functions of growth factors is the suppression of apoptosis (Wyllie et al., 1980; Williams et al., 1990; Raff, 1992). This general statement is the result of numerous in vitro experiments showing that growth factor deprivation results in programmed cell death. Recently, mutant mice perturbed in growth factor action have been identified (Mann et al., 1993; Cecchini et al., 1994) or generated purposefully by deletion of growth factor genes (Mann et al., 1993; Mansour et al., 1993; Hebert et al., 1994; Lieschke et al., 1994; Feldman et al., 1995), receptor genes (Yamaguchi et al., 1994) or through the use of dominant-negative receptors (Werner et al., 1993; Peters et al., 1994). These are all in vivo systems in which apoptosis of target cell populations might be expected. In some cases, analysis is difficult due to a lethal phenotype or a diffuse target population. These may be some of the reasons why the suppressive effect of growth factors on apoptosis has not been confirmed in vivo. Our observation of apoptosis in lens fibre cells overexpressing a dominantnegative FGF receptor argues that FGF has significant apoptosis suppressive activity and raises the question of whether this will prove to be a general feature of growth factor action in vivo.
ACKNOWLEDGEMENTS
We are particularly grateful to the following colleagues for gifts of reagents: J. Horvitz for antisera to MIP26, H. Reid and O. Bernard for the murine cDNA to FGFR1, D. Ron for the cDNA to GAPDH, J. Piatigorsky for the promoter region and cDNA clones to αA-crystallin, B. Margolis for anti-phosphotyrosine antibodies, I. Dikic and J. Schlessinger for anti-MAP kinase antibodies. In addition, we would like to thank M. Robinson and P. Overbeek for sharing experimental results prior to publication and a number of local colleagues for technical advice. These include Taly Spivak-Kroizman, Ed Skolnik, Moosa Mohamadi, Alka Mansukhani, and Irit Lax. We are indebted to Jonathon Weider for his help in producing the figures associated with this work. This work was supported in part by funds from: the Searle Scholars Program of The Chicago Community Trust (grant number 94-F-118 awarded to R. A. L.), NIH RO1 #EY10559 (R. A. L.), NIH RO1 #CA42229 (D. A. M.) and NIH RO1 #CA34282 (D. B. R.). R. L. C. and G. D. R. were supported in part by the Cell and Molecular Biology Graduate Training Program funded by the NIH (grant number GM07238).