ABSTRACT
Members of the fibroblast growth factor (FGF) family are thought to initiate biological responses through the activation of cell surface receptors which must dimerize to transmit an intracellular signal. Mammalian lens epithelial cells respond to exogenous extracellular FGF, either in tissue culture or in transgenic mice, by initiating fiber cell differentiation. The role of FGF signalling in normal lens development was evaluated by lens-specific synthesis of a kinase-deficient FGF receptor type I (FGFR1) in transgenic mice. This truncated FGF receptor is thought to act as a dominant negative protein by heterodimerization with endogenous FGF receptors. The presence of transgenic mRNA in the lens was confirmed by in situ hybridization and by polymerase chain reaction amplification of reverse transcribed lens RNA (RT-PCR). The presence of transgenic protein was determined by Western blotting with antibodies to an extracellular domain of FGFR1. Three of four transgenic families expressing the truncated FGF receptor exhibited lens defects ranging from cataracts to severe microphthalmia. While the microphthalmic lenses displayed a normal pattern of differentiation-specific crystallin expression, the lens epithelial cells were reduced in number and the lens fiber cells displayed characteristics consistent with the induction of apoptosis. Our results support the view that FGF receptor signalling plays an essential role in normal lens biology.
INTRODUCTION
Elucidation of the function of fibroblast growth factors (FGFs) during development is complicated by the presence of at least nine different FGF genes (FGF-1-9); (for review, see Baird 1994), many of which display overlapping expression patterns. In addition to the multiple FGFs, there are at least four different FGF receptor (FGFR) genes (FGFR1-4) which give rise to numerous receptor isoforms via alternative splicing of mRNA (for review, see Johnson and Williams 1993). Over the past few years, several experimental approaches have been used to elucidate the developmental roles of different FGFs and their receptors in vivo. Homologous recombination to generate null alleles in murine embryonic stem cells has revealed essential roles of FGF-3 for tail and inner ear development (Mansour et al., 1993), FGF-4 for postimplantation development (Feldman et al., 1995), FGF-5 for regulation of hair growth (Hebert et al., 1994) and FGFR1 for axial organization and mesodermal patterning (Yamaguchi et al., 1994; Deng et al., 1994). While generating targeted mutations can reveal the earliest essential function of a single FGF or FGFR family member, the role of a specific FGF or FGFR in the development of a particular tissue may be obscured using this approach if a null mutation in that gene leads to early embryonic lethality. Furthermore, the loss of one member of a gene family may be compensated for by overlapping expression of another, related gene (Schneider et al., 1994).
The use of a dominant negative FGF receptor offers the ability to explore the role(s) of FGF signalling in a tissue and developmentally restricted manner. Full-length FGFRs consist of an extracellular portion including an acidic domain and two or three Immunoglobulin-like (Ig-like) domains, a single transmembrane domain, and an intracellular tyrosine kinase domain split by a short kinase insert (Johnson and Williams, 1993). Truncated FGFRs (lacking an intracellular tyrosine kinase domain) are capable of ligand binding and act as dominant negative proteins by disrupting FGF-induced signal transduction from multiple FGF receptor isoforms (Ueno et al., 1992). FGFRs, like other tyrosine kinase receptors, undergo dimerization in response to ligand binding (for review, see Heldin 1995). Both the second and third extracellular Ig-like domains are involved in ligand binding, and differential mRNA splicing in the third Ig-like domain is particularly important for generating ligand specificity (Werner et al., 1992). For example, the FGFR1 splice variant IIIb has high affinity for FGF-1 but low affinity for FGF-2 while the splice variant IIIc has high affinity for both FGF-1 and FGF-2 (see Johnson and Williams, 1993). Receptor dimerization is followed by intermolecular transphosphorylation of the receptor subunits which initiates intracellular signal transduction. In vivo, truncated receptors would be expected to disrupt FGF signalling by heterodimerization with endogenous, full-length receptors, or by sequestering ligand. In either case, ligand-receptor complexes will be inactive. Dominant negative receptor strategies have suggested that FGFs are required for Xenopus gastrulation (Amaya et al., 1991) as well as the normal differentiation and organization of both keratinocytes (Werner et al., 1993) and lung epithelium (Peters et al., 1994) in transgenic mice.
The ocular lens is a tissue in which FGF signalling is likely to play an important developmental role. A single undifferentiated layer of cuboidal epithelial cells covers the anterior surface of the normal lens while the remainder of the lens mass is made up of differentiated lens fiber cells. At the lens equator, epithelial cells withdraw from the cell cycle and differentiate into fiber cells. Fiber cells differ from lens epithelial cells in that they are amitotic, elongated and express fiber cell-specific proteins such as β- and γ-crystallins (McAvoy, 1978). In the ocular lens, FGF molecules have been shown to induce the differentiation of epithelial cells into fiber cells, both in vitro (McAvoy et al., 1991) and in transgenic mice (Robinson et al., 1995). Other studies suggest that expression of FGF by lens epithelial cells is important for lens cell survival (Renaud et al., 1994).
The finding that only secreted forms of FGF-1 induce lens epithelial cell responses in transgenic mice suggests that cell surface FGF receptors play an essential role for FGF signalling in the lens (Robinson et al., 1995). The lens expresses at least three different FGF receptor genes: FGFR1, FGFR2 and FGFR3 (Orr-Utreger et al., 1991; Peters et al., 1993). A truncated FGFR1 has been shown to act as a dominant negative FGF receptor in Xenopus oocytes, inhibiting FGF-induced signal transduction from multiple different full-length FGFRs including FGFR1, FGFR2 and FGFR3 (Ueno et al., 1992). The ability of the αA-crystallin promoter to direct transgene expression specifically to the lens of transgenic mice (Overbeek et al., 1985; Robinson et al., 1995; Reneker et al., 1995) makes the lens particularly amenable to dominant negative receptor strategies in vivo. High level expression of a dominant negative FGFR1 in the lens could theoretically disrupt FGF signalling from all endogenous lens FGFRs. We have chosen to express a truncated FGFR1 in the lens of transgenic mice to assess the roles of FGFR signalling during lens development.
MATERIALS AND METHODS
DNA constructs
A cDNA clone encoding a truncated murine FGFR1 was generously provided by the laboratory of Dr L.T. Williams, San Francisco. This clone encodes an FGFR1 protein with two extracellular Ig-like domains and with the splice variant IIIc, which binds to both FGF-1 and FGF-2 with high affinity (Werner et al., 1992). The cDNA encoding the truncated FGFR1 contained a termination codon 26 amino acids into the cytoplasmic juxtamembrane domain, which results in an FGF receptor without an intracellular tyrosine kinase domain. To generate the truncated FGFR1 construct (αA-tR1), the 369 bp αA-crystallin promoter from CPV2 (Robinson et al., 1995) was ligated 5′ of the rabbit β-globin intron, replacing the bovine K10 promoter in the truncated FGFR1-expression vector described by Werner et al. (1993). This plasmid was designated CPV5/αA-tR1 (Fig. 1).
The transgenic construct (αA-tR1) consisted of a truncated FGFR1 cDNA driven by the lens-specific 369 bp αA crystallin promoter (αAp). The truncated FGFR1 cDNA encodes the extracellular (shaded portion), the transmembrane (solid bar) and a small portion of the cytoplasmic juxtamembrane domains. The coding region of the αA-tR1 was flanked by a rabbit β-globin intron and a polyadenylation signal derived from the human growth hormone gene (GHpA). The sequences used to make a transgene-specific riboprobe for in situ hybridization are indicated. Half-arrows indicate a sequence-specific amplimer pair (C5 and FR1) used for PCR analysis of αA-tR1.
The transgenic construct (αA-tR1) consisted of a truncated FGFR1 cDNA driven by the lens-specific 369 bp αA crystallin promoter (αAp). The truncated FGFR1 cDNA encodes the extracellular (shaded portion), the transmembrane (solid bar) and a small portion of the cytoplasmic juxtamembrane domains. The coding region of the αA-tR1 was flanked by a rabbit β-globin intron and a polyadenylation signal derived from the human growth hormone gene (GHpA). The sequences used to make a transgene-specific riboprobe for in situ hybridization are indicated. Half-arrows indicate a sequence-specific amplimer pair (C5 and FR1) used for PCR analysis of αA-tR1.
Transgenic mice
All plasmid purifications were performed using plasmid purification kits from QIAGEN (Hilden, Germany). The 2.5 kb αA-tR1 microinjection fragment was isolated by digestion of CPV5/αA-tR1 with KpnI, SstII and BglI (BglI cut only the vector) followed by gel electrophoresis and extraction using the QIAEX kit (QIAGEN). Linear DNA fragments were eluted in 10 mM Tris HCl (pH 7.4), 0.1 mM EDTA and microinjected into individual pronuclei of FVB/N embryos (Taketo et al., 1991) at a concentration of 2 ng/μl. Injected embryos were transferred into pseudopregnant ICR females. Potential transgenic mice were screened by isolating genomic DNA (Hogan et al., 1994) from tail biopsies and testing for transgenic sequences using the polymerase chain reaction (PCR) (Saiki et al., 1988).
PCR analyses
Primers used for PCR (see Fig. 1) were CCCAGAGGCTCCTGTCT-GACTCACT (termed C5), a sense primer for the 5′ untranslated region of the murine αA-crystallin transcript (present in the transgenic constructs), and TCCGAGGATGGGAGTGCATCTTGTT (termed FR1), an antisense primer that hybridizes to sequences in the murine FGFR1 cDNA in αA-tR1. Primers C5 and FR1 amplify an 872 bp band from genomic DNA and a 295 bp band from lens cDNA in transgenic animals expressing αA-tR1. Total ocular RNA from newborn transgenic and control mice was isolated using RNA STAT-60 (Tel-Test ‘B’, Inc., Friendswood, TX); (see Chomczynski and Sacchi, 1987). The RNA was then reverse transcribed and the resultant cDNA was amplified by PCR (see Robinson et al., 1995).
In situ hybridization
CPV5/αA-tR1 was digested with EcoRI and HindIII to isolate a 600 bp fragment containing the human growth hormone sequences which constitute the 3′ untranslated portion of the transgenic construct. This fragment was subcloned between the EcoRI and HindIII sites of pBluescript KS− (Stratagene, La Jolla, CA). Sense and antisense ribo-probes were generated by in vitro transcription using T3 (Stratagene) and T7 (Pharmacia, Piscataway, NJ) polymerase, respectively. Tissues were processed and hybridization was carried out as described previously (Robinson et al., 1995). Hybridized slides were air dried and dipped in Kodak NTB-2 emulsion and exposed for 2 weeks at 4°C before development with Kodak D-19 developer. Slides were counterstained with Harris hematoxylin.
Immunohistochemistry
Postnatal eyes or fetal heads were removed and fixed in 10% phosphate-buffered formalin before processing, embedding and sectioning (see Robinson et al., 1995). For cell proliferation studies, pregnant mice 17.5 days postcopulation were injected intraperitoneally with 5-bromo-2′-deoxyuridine (BrdU); (Sigma, St. Louis, MO) using 100 μg/g of body weight, and with 5-fluoro-2′ deoxyuridine (Sigma) at 10 μg/g of body weight, dissolved in phosphate-buffered saline (PBS). One hour later pregnant females were killed, fetuses delivered by Cesarean section and fetal heads were fixed and processed as described (see Robinson et al., 1995). Transgenic fetuses were identified by PCR analysis of tail DNA. To determine the sites of BrdU incorporation, sections were rehydrated, exhausted of endogenous peroxidase activity and incubated in 5% normal mouse or rabbit serum (Robinson et al., 1995). Sections were then incubated overnight at 4°C in a 1:50 dilution of primary mouse anti-BrdU antibody (Dako, Carpinteria, CA): 5% normal mouse serum in PBS. Binding of the primary antibody was visualized using biotinylated secondary antibodies and substrate kits according to manufacturer’s specifications (Vector Laboratories, Burlingame, CA). Similar immunohistochemical approaches were used to visualize β- and γ-crystallin. In all cases, negative controls included incubations without primary antibody. Sections were lightly counter-stained with nuclear fast red (Sigma), or hematoxylin. Rabbit polyclonal antibodies to β- and γ-crystallin were provided by Dr J. Fielding Hejtmancik (National Eye Institute, Bethesda, MD) and Dr J. S. Zigler (National Eye Institute, Bethesda, MD), respectively. For histochemical analysis, sections were stained with hematoxylin and eosin.
Western blot analysis
Lenses were removed from transgenic and control eyes and immediately homogenized in ice-cold lysis buffer (20 mM Tris-HCl (pH 7.4), 150 mM NaCl, 10% glycerol, 1% Triton X-100, 4 mM EGTA, 0.5 mM sodium ortho-vanadate, 0.1 mM PMSF, 1 μg/ml pepstatin A, 10 μg/ml leupeptin). Homogenates were flash frozen at −70°C. Upon thawing, homogenates were sonicated using a microtip sonicator at the 40% power setting for two 10 second bursts on ice. Protein concentrations were determined by the method of Bradford (1976). 75 μg of each protein sample was separated by 15% SDS-polyacrylamide gel electrophoresis (PAGE). Proteins were electrophoretically transferred to nitrocellulose (0.45 μm pore size) at 80 V for 2 hours in transfer buffer containing 10 mM 3-(cyclohexylamino)-1-propane-sulfonic acid (CAPS) buffer (pH 11) and 10% methanol. The nitrocellulose membranes were incubated in 5% non-fat dry milk (dissolved in Tris-buffered saline (TBS)/0.05% Tween-20) for 1 hour at room temperature to block non-specific protein binding. For detection of the truncated FGFR1 in lens extracts, blots were incubated for 1 hour at room temperature with either the mouse monoclonal anti-FGFR1 antibody M17D10, diluted 1:100, or the polyclonal anti-FGFR1 antibody Rb(2-7), diluted 1:150, both of which recognize an extracellular domain of FGFR1. Endogenous FGFR1 was detected using the mouse monoclonal anti-FGFR1 antibody 5G11, diluted 1:2500. This antibody recognizes an intracellular domain of FGFR1 which is not encoded by αA-tR1. After incubation in primary antibody, blots were washed briefly three times in TBS/Tween 20, followed by two 7 minute washes in buffer B (0.05% NP40/ 0.125% sodium deoxycholate/ 0.05% SDS in TBS) and three additional washes in TBS/Tween 20. Blots were then incubated in TBS/Tween 20 with either goat anti-mouse horseradish peroxidase (HRP)-linked secondary antibody (1:30,000) or goat anti-rabbit HRP-linked secondary antibody (1:20,000) for 45 minutes at room temperature. Blots were then washed briefly three times in TBS/Tween 20, followed by five 5 minute washes in buffer B and three more brief washes in TBS. Antibody binding was detected using the enhanced chemiluminescent assay (ECL) method (Kirkegaard and Perry Labs). Polyclonal antibodies (Rb 2-7) raised to a synthetic peptide (CLDVVERSPHRPILQAGLPAN) common to both the 2- and 3-Ig-like forms of FGFR1 were the gift of Dr Robert Bjercke (Texas Biotechnology Corp.). The mouse monoclonal antibody M17D10 and the two Ig-like loop FGFR1 protein standard were the gifts of Dr Wallace McKeehan (Texas A&M, Houston).
TUNEL assay for apoptosis
Fetal heads were collected, processed and sectioned for routine histology as described (Robinson et al., 1995). Tissue sections were then incubated with 50 U of terminal deoxynucleotidyl transferase (Promega) and 1 nmol of biotin-16-dUTP (Boehringer Mannheim), as described by Gavrieli et al. (1992). Biotin incorporation was detected according to Fromm et al. (1994).
RESULTS
Transgenic mice
To assess the function of FGFRs during lens development, transgenic mice that express a cDNA encoding a truncated FGFR1, driven to the lens by the αA-crystallin promoter, were generated. Transgenic founders were identified by PCR using transgene-specific amplimers: C5 and FR1 (see Fig. 1). 92 injected embryos resulted in 31 newborn mice, of which four were transgenic for αA-tR1. These four transgenic mice were used to establish four independent families (OVE 497-500).
Three of the four transgenic lines established with the dominant negative receptor construct exhibited ocular abnormalities (see Fig. 2). Line OVE 497 consistently presented bilateral cataracts by 3 weeks of age (Fig. 2B), and lines 498 and 499 exhibited severe microphthalmia (Fig. 2C). No abnormal ocular phenotype was observed in any transgenic animal from line OVE 500.
External ocular appearance of animals expressing αA-tR1. (A) Non-transgenic FVB/N mouse exhibiting normal external ocular appearance. (B) A typical mouse from family OVE 497. The eye is nearly normal in size, but has an opaque lens. (C) A typical mouse from OVE 498. The eye is microphthalmic as a result of retarded lens development.
External ocular appearance of animals expressing αA-tR1. (A) Non-transgenic FVB/N mouse exhibiting normal external ocular appearance. (B) A typical mouse from family OVE 497. The eye is nearly normal in size, but has an opaque lens. (C) A typical mouse from OVE 498. The eye is microphthalmic as a result of retarded lens development.
Transgene expression
Transcription of αA-tR1 was confirmed by RT-PCR analysis of ocular RNA from newborn animals. PCR amplification of ocular cDNA with primers C5 and FR1 (see Fig. 1) distinguished the correctly spliced transgenic mRNA (295 bp) from amplification of unspliced RNA and/or contaminating DNA (872 bp); (Fig. 3A). All four transgenic families express trangenic mRNA (lanes 4-7). No transgene-specific bands were seen from non-transgenic DNA (lane 3) or cDNA (lane 9) or from transgenic RNA which had not been reverse transcribed (lane 8).
PCR analysis of genomic DNA and reverse transcribed ocular RNA from αA-tR1 mice. Lane 1, 1.0 kb ladder (Gibco/BRL); lanes 2 and 3 contain transgenic and non-transgenic (FVB/N) genomic DNA, respectively; lanes 4-7 contain reverse transcribed ocular RNA from transgenic mice for families OVE 497-500; lane 8 contains ocular RNA which was not treated with reverse transcriptase; lane 9 contains non-transgenic (FVB/N) reverse transcribed ocular RNA. All samples were amplified with primers C5 and FR1. Amplification yields an 872 bp band from transgenic DNA and a 295 bp band from correctly spliced, reverse transcribed, transgenic transcripts. The mobility of relevant DNA molecular weight standards (1 kb ladder, Gibco/BRL) are indicated by arrowheads.
PCR analysis of genomic DNA and reverse transcribed ocular RNA from αA-tR1 mice. Lane 1, 1.0 kb ladder (Gibco/BRL); lanes 2 and 3 contain transgenic and non-transgenic (FVB/N) genomic DNA, respectively; lanes 4-7 contain reverse transcribed ocular RNA from transgenic mice for families OVE 497-500; lane 8 contains ocular RNA which was not treated with reverse transcriptase; lane 9 contains non-transgenic (FVB/N) reverse transcribed ocular RNA. All samples were amplified with primers C5 and FR1. Amplification yields an 872 bp band from transgenic DNA and a 295 bp band from correctly spliced, reverse transcribed, transgenic transcripts. The mobility of relevant DNA molecular weight standards (1 kb ladder, Gibco/BRL) are indicated by arrowheads.
To confirm the lens specificity of αA-tR1 transgene expression, in situ hybridization was performed on developing eyes at embryonic day 14.5 (E14.5) on the most severely affected αA-tR1 trangenic line, OVE 498 (Fig. 4). A transgene-specific 35S-riboprobe was produced from the human growth hormone sequences present in the 3′ untranslated region of the αA-tR1 construct. As expected, transgene expression was confined to the ocular lens. Consistent with other transgenic constructs (Robinson et al., 1995; Reneker et al., 1995), the αA-crystallin promoter directed transgene expression specifically to lens fiber cells (Fig. 4A,B). The αA-tR1 construct, however, appeared to turn on particularly late in fiber cell differentiation leading to an ‘apple core’ pattern of hybridization (Fig. 4A). No hybridization was detected in nontransgenic heads or in any tissue other than the lens in trangenic heads at E14.5. A sense 35S-riboprobe to the same human growth hormone sequences failed to hybridize to transgenic or non-trangenic tissue (data not shown).
In situ hybridizations. An antisense 35S-labeled riboprobe specific for the human growth hormone sequences present in the 3′ untranslated region of the αA-tR1 construct was hybridized to sections from E14.5 transgenic OVE 498 (A,B) and non-transgenic (C,D) heads. Dark-field (A,C) and bright-field (B,D) images of the eyes are shown. Transgene expression is confined to the lens fiber cells. Scale bar, 100 μm.
In situ hybridizations. An antisense 35S-labeled riboprobe specific for the human growth hormone sequences present in the 3′ untranslated region of the αA-tR1 construct was hybridized to sections from E14.5 transgenic OVE 498 (A,B) and non-transgenic (C,D) heads. Dark-field (A,C) and bright-field (B,D) images of the eyes are shown. Transgene expression is confined to the lens fiber cells. Scale bar, 100 μm.
Transgenic protein was visualized using Western blot analysis (Fig. 5). Total lens proteins from transgenic and control mice were separated using 15% SDS-PAGE and probed using FGFR1-specific antibodies: M17D10, Rb 2-7 or 5G11. Antibodies M17D10 and Rb 2-7 are specific to extracellular portions of FGFR1 present in endogenous and transgenic forms of FGFR1. Antibody 5G11 is a monoclonal antibody specific for the cytoplasmic portion of FGFR1 (absent in αA-tR1). In transgenic and non-transgenic lens proteins, both M17D10 and 5G11 detected a band of approximately 100×103Mr which co-migrated with a two Ig-like loop fulllength FGFR1 protein standard (Fig. 5). This 100×103Mr band most likely represents the endogenous FGFR1 protein in the lens homogenates. In lens extracts from the αA-tR1 mice, additional bands of 60 and 52×103Mr were detected with either of the FGFR1 extracellular domain antibodies, M17D10 (Fig. 5A) or Rb 2-7 (not shown), but not with the cytoplasmic domain antibody, 5G11 (Fig 5B). The 60 and 52×103Mr bands likely represent the transgenic truncated FGFR1 because they are only present in the αA-tR1 transgenic lens extracts and are recognized by two different antibodies to extracellular domains of FGFR1, but not by an antibody to a cytoplasmic domain of FGFR1. Preliminary Western blotting experiments suggest that the abundance of the 60 and 52×103Mr bands in lens extracts from different αA-tR1 lines is proportional to the severity of the transgenic phenotype (data not shown).
Expression of FGFR1 proteins in αA-tR1 mice. (A) Western blot of lens extracts from non-transgenic (lane 2); and αA-tR1 trangenic (lanes 4,5) mice using the antibody M17D10, specific for an extracellular domain of FGFR1. Lane 3 contains lens protein from transgenic mice made with a related FGFR1 construct which fails to produce detectable transgenic protein. A band of approximately 100×103 Mr which co-migrates with a two loop full-length FGFR1 protein standard (uppermost band, lane 1) was present in each lens extract, while a faster migrating doublet of approximately 60 and 52×103 Mr was present only in the extracts from αA-tR1 transgenic mice. (B) Western blot of non-transgenic (lane 1) and αA-tR1 transgenic (lanes 2-4) lens extracts using the antibody 5G11, specific for a cytoplasmic domain of FGFR1. The mobility of molecular weight standards is indicated by arrowheads at the left sides of the blots. The numbers above each lane indicate the transgenic family from which the corresponding lens extract was prepared.
Expression of FGFR1 proteins in αA-tR1 mice. (A) Western blot of lens extracts from non-transgenic (lane 2); and αA-tR1 trangenic (lanes 4,5) mice using the antibody M17D10, specific for an extracellular domain of FGFR1. Lane 3 contains lens protein from transgenic mice made with a related FGFR1 construct which fails to produce detectable transgenic protein. A band of approximately 100×103 Mr which co-migrates with a two loop full-length FGFR1 protein standard (uppermost band, lane 1) was present in each lens extract, while a faster migrating doublet of approximately 60 and 52×103 Mr was present only in the extracts from αA-tR1 transgenic mice. (B) Western blot of non-transgenic (lane 1) and αA-tR1 transgenic (lanes 2-4) lens extracts using the antibody 5G11, specific for a cytoplasmic domain of FGFR1. The mobility of molecular weight standards is indicated by arrowheads at the left sides of the blots. The numbers above each lane indicate the transgenic family from which the corresponding lens extract was prepared.
Histological and developmental analyses of αA-tR1 eyes
Histological analyses were carried out on the most severely affected αA-tR1 line, OVE 498. At E17.5, the overall size of the transgenic lens is smaller than that of non-transgenic littermate lenses (Fig. 6A,B). The central fiber cells of these lenses appear to be degenerating and dark pyknotic bodies resembling condensed nuclei are evident. The epithelial cells of the embryonic transgenic lens were slightly elongated, and in some instances multilayered (Fig. 6A inset). At postnatal day 14 (P14), transgenic lenses are dramatically smaller than non-transgenic lenses (Fig. 6C,D). The transgenic lenses consist of a large central vacuole surrounded on the anterior side by a layer of epithelial cells and to the lateral and posterior sides by short, swollen nucleated fiber cells (Fig. 6C). An immunohistochemical analysis of fiber cell-specific β-(Fig. 7) and γ-(not shown) crystallins was undertaken to examine the ability of αA-tR1 expression to alter early fiber cell differentiation. At E17.5, β- and γ-crystallins were absent from the epithelium and present in the fiber cells of both transgenic and control lenses (see Fig. 7). Although the distribution pattern of both differentiation-specific crystallins was similar to normal, immunostaining in the transgenic lenses appeared heterogeneous relative to the controls. Therefore, although β- and γ-crystallin distribution in the transgenic fiber cells was less uniform than in the normal lens, expression of the truncated FGFR1 in the fiber cells does not greatly influence β-or γ-crystallin expression and accumulation.
Ocular histology of αA-tR1. Histological sections from eyes either transgenic (OVE 498) for αA-tR1 (A,C) or non-transgenic (B,D) at E17.5 (A,B) or at P14 (C,D) are shown. The epithelial cells of the E17.5 transgenic lens appear elongated and multilayered compared to those of the non-transgenic lens (A,B insets). By P14, the degeneration of the fiber cells in the interior of the transgenic lens results in an acellular cavity within the lens and microphakia. Scale bar, 100 μm (A,B); 200 μm (C,D); 25 μm (insets).
Ocular histology of αA-tR1. Histological sections from eyes either transgenic (OVE 498) for αA-tR1 (A,C) or non-transgenic (B,D) at E17.5 (A,B) or at P14 (C,D) are shown. The epithelial cells of the E17.5 transgenic lens appear elongated and multilayered compared to those of the non-transgenic lens (A,B insets). By P14, the degeneration of the fiber cells in the interior of the transgenic lens results in an acellular cavity within the lens and microphakia. Scale bar, 100 μm (A,B); 200 μm (C,D); 25 μm (insets).
Immunostaining for β-crystallin. Immunohistology was done to look at the pattern of β-crystallin expression (indicated by brown staining) in lenses of αA-tR1 (OVE 498) transgenic (A) and non-transgenic (B) E17.5 eyes. Expression of αA-tR1 in lens fiber cells does not alter the cell type-specific pattern of β-crystallin expression in the transgenic lens. Expression of β-crystallin is restricted to the fiber cells in both transgenic and non-transgenic lenses. Scale bar, 100 μm.
Immunostaining for β-crystallin. Immunohistology was done to look at the pattern of β-crystallin expression (indicated by brown staining) in lenses of αA-tR1 (OVE 498) transgenic (A) and non-transgenic (B) E17.5 eyes. Expression of αA-tR1 in lens fiber cells does not alter the cell type-specific pattern of β-crystallin expression in the transgenic lens. Expression of β-crystallin is restricted to the fiber cells in both transgenic and non-transgenic lenses. Scale bar, 100 μm.
Reduced epithelial cell number and fiber cell apoptosis
To determine if expression of αA-tR1 affected lens cell proliferation, transgenic and control embryos were labeled by bro-modeoxyuridine (BrdU) incorporation in utero. BrdU incorporates only into cells which are in S-phase of the cell cycle during the labeling period and the proportion of cells in S-phase can be used as a measure of cell proliferation (Yoshiki et al., 1991). The majority of BrdU-labeled cells appeared slightly anterior to the lens equator, but positive cells could be detected throughout the lens epithelium in both transgenic and non-transgenic lenses (data not shown). The BrdU incorporation into lens epithelial cells was reduced in the αA-tR1 transgenic embryos (OVE 498) relative to that of non-transgenic littermate controls at E17.5, while BrdU incorporation in other cell types (such as the retina) appeared similar (data not shown). At E17.5 the reduction in BrdU incorporation in the transgenic lens reflects a reduced number of lens epithelial cells. On 55 mid-sagittal lens sections from OVE 498 embryos, representing four transgenic eyes, the number of epithelial cells per lens was 116±27 s.d. of which 14.2±5.2% s.d. were stained positively for BrdU incorporation. In the non-transgenic littermates, on 84 sections representing four eyes, the number of epithelial cells per lens was 213±15 s.d. of which 18.9±3.8% s.d. were stained positively for BrdU incorporation. Therefore, while the total number of lens epithelial cells is significantly reduced in the transgenic E17.5 lens, the fraction of epithelial cells in S-phase is similar to or slightly reduced compared to that of the control lens.
The condensed nuclei present in the degenerating lens fiber cells of the microphthalmic αA-tR1 families morphologically resembled nuclei in cells undergoing apoptosis (see Arends et al., 1990). To determine if αA-tR1 expression in the lens induces apoptosis, the TUNEL assay (Gavrieli et al., 1992) was performed on OVE 498 embryos at E14.5 (not shown) and E17.5 (Fig. 8). The TUNEL assay detects nuclei that have undergone significant DNA breakage (typical of apoptosis). Lenses from OVE 498 are smaller than normal and are under-going central lens degeneration by E14.5. Many of the condensed nuclei within the center of the transgenic lenses stained positive with this assay at both embryonic stages, rep-resenting approximately 12±3% s.d. of total fiber cell nuclei. In contrast, at E14.5, no TUNEL-positive fiber cells were detected in lens sections from transgenic families OVE 497 and 499 and non-transgenics which appear histologically normal at this stage (data not shown). By P1 lens sections from OVE 497 remain normal in appearance and free of TUNEL-positive fiber cell nuclei. Histological abnormalities are evident at birth in OVE 499 eyes and approximately 3.5±0.8% of fiber cell nuclei from lens sections from this family are TUNEL positive at P2 and P21. Therefore, the onset of TUNEL-positive nuclear staining appears to correlate with the onset of histologically evident lens degeneration. These results are in agreement with those reported from experiments describing lens-specific expression of a three Ig-like loop form of truncated FGFR1 (Chow et al., 1995), and suggest that expression of a dominant negative FGFR1 in lens fiber cells results in cell lethality via a programmed cell death mechanism.
TUNEL assay in αA-tR1 lenses. E17.5 eyes from αA-tR1 (OVE 498) and non-transgenic mice were labeled with the TUNEL assay to detect nuclear DNA fragmentation. Positive TUNEL staining (brown) is indicative of cells undergoing apoptosis. Many of the condensed fiber cell nuclei of the transgenic lens (A,B) were TUNEL positive (arrows in B), while none of the fiber cell nuclei in the normal lens (C,D) were positive by this assay. Scale bar 100 μm (A,C); 25 μm (B,D).
TUNEL assay in αA-tR1 lenses. E17.5 eyes from αA-tR1 (OVE 498) and non-transgenic mice were labeled with the TUNEL assay to detect nuclear DNA fragmentation. Positive TUNEL staining (brown) is indicative of cells undergoing apoptosis. Many of the condensed fiber cell nuclei of the transgenic lens (A,B) were TUNEL positive (arrows in B), while none of the fiber cell nuclei in the normal lens (C,D) were positive by this assay. Scale bar 100 μm (A,C); 25 μm (B,D).
DISCUSSION
Extracellular FGF is capable of inducing fiber cell differentiation in lens epithelial cells (McAvoy et al., 1991; Robinson et al., 1995). FGF has also been implicated in lens cell proliferation (McAvoy et al., 1991) and survival (Renaud et al., 1994). To begin to assess the role(s) of FGFRs in the lens, a truncated FGF receptor, which has been shown to act as a dominant negative protein (Ueno et al., 1992), was expressed in transgenic mice using the lens-specific αA-crystallin promoter.
Severe lens abnormalities were consistently present in three of four transgenic lines expressing αA-tR1. These lens abnormalities correlated with the appearance of two proteins that exhibit mobilities consistent with truncated forms of the two Ig-like loop FGFR1 as detected by Western blots of lens extracts using extracellular-, but not intracellular-, specific antibodies to FGFR1 (Fig. 5). The levels of these truncated FGFR1 proteins exceeded that of endogenous FGFR1 protein in all αA-tR1 families except OVE 500 (data not shown), which may explain the lack of phenotype in the OVE 500 mice. Taken together, these findings, strongly suggest that the αA-tR1 mice are making transgenic protein and that this protein is directly responsible for the lens abnormalities observed. The presence of two bands for the truncated protein may represent differences in glycosylation or perhaps some type of degradative process. It is interesting that no higher molecular weight bands indicative of the three Ig-like loop forms of FGFR1 were evident in Western blots of transgenic or control lens extracts. This suggests that the major endogenous form of postnatal lens FGFR1 contains only two Ig-like loops.
The distribution of β- and γ-crystallins in lenses expressing αA-tR1 displayed a relatively normal pattern, indicating that lens cells expressing the transgene are able to express biochemical markers of fiber cell differentiation. Furthermore, in embryonic development, epithelial cells continue to differentiate into fiber cells even in the most severely affected transgenic families. As development progresses, however, the transition from epithelial cell to fully differentiated fiber cells becomes impaired. A fully mature normal fiber cell is elongated and anucleate. In postnatal development, the fiber cells from microphthalmic families expressing αA-tR1 fail to fully elongate and remain nucleated despite expressing differentiation-specific β- and γ-crystallins. This inhibition of fiber cell elongation contributes to the appearance of the acellular cyst within the lens. At present, it is unclear why the epithelial cell number in the OVE 498 lenses is reduced relative to that of control lenses. The nuclei in the transgenic fiber cells become condensed and in addition undergo significant DNA degradation as indicated by positive staining with the TUNEL assay. Nuclear condensation and DNA degradation are two characteristics consistent with the induction of apoptosis (Arends, 1990). Not surprisingly, the most severely affected transgenic family (OVE 498) undergoes apoptosis at an earlier stage and to a greater extent than the other less severely affected families. It is likely that the early onset of fiber cell apoptosis contributes to the severe microphthalmia in this family. Cell culture studies have estimated that the level of the truncated FGFR1 receptor may need to be 10-to 75-fold higher than that of the endogenous FGFR1 receptor to block most of the endogenous receptor signalling (Ueno et al., 1992). The less severely affected families (OVE 497 and 499) may require a longer period of time for the levels of truncated FGFR1 expression to reach a level sufficient to inhibit endogenous FGFR1 activity. Overall, our results imply that FGFRs may play a critical role in fiber cell survival.
Interpretation of these results is complicated by several issues. In order to achieve a dominant negative effect from the expression of a truncated FGF receptor, the truncated receptor typically must be expressed at a level much higher than that of the endogenous FGF receptors (Ueno et al., 1992). The possibility exists that a slightly higher level of expression of the truncated FGF receptor may have toxic effects on lens cells unrelated to FGF signalling. Although our Western blots indicate that the truncated receptor proteins are substantially overexpressed relative to the wild-type receptors, the functional inhibition of endogenous receptor activity has not been quantitated. Also, while the αA-crystallin promoter efficiently directs high-level expression of transgenes to lens fiber cells, this promoter may be significantly less active in lens epithelial cells (see Fig. 4). Therefore levels of truncated FGF receptor expression sufficient to disrupt FGF-induced signal transduction are likely to be restricted to cells already committed to fiber differentiation. As a result, FGF signalling in lens epithelial cells, and in the early stages of fiber cell differentiation, is unlikely to be significantly inhibited in the transgenic mice. With respect to the importance of FGF-signalling in the initiation of fiber cell differentiation in vivo, our results remain inconclusive. However, if the truncated FGF receptor acts as a dominant negative protein in transgenic fiber cells as it does in Xenopus oocytes, then FGF must function as a survival factor for lens fiber cells.
We have expressed a two Ig-like loop FGFR1 which has been shown to have higher affinity for FGF than the three Iglike loop form of FGFR1 (Shi et al., 1993). Since FGF-1 and FGF-2 have been detected in the ocular media (Schulz et al., 1993), we used the IIIc splice variant of FGFR1 which binds to both FGF-1 and FGF-2 with high affinity (Werner et al., 1992). One mechanism by which the truncated FGFR1 transgene may disrupt FGF signalling is by binding to and sequestering both FGFs, thereby reducing the amount of FGF available in the ocular media. In the transgenic mice, the total number of truncated FGF receptors would be expected to increase with increasing numbers of fiber cells. Therefore, while fiber differentiation may initially occur appropriately in these mice, the number of FGF molecules available to induce lens differentiation or proliferation may become limiting with time. This might explain the postnatal inhibition of secondary fiber cell differentiation in the microphthalmic αA-tR1-expressing lens.
Elucidating the role of FGF in the initiation of fiber cell differentiation in vivo awaits the availability of a promoter which can direct high level expression of the truncated FGF receptor specifically to the epithelial cells. In spite of this caveat, our results suggest that FGF receptor function is required after the onset of fiber cell differentiation to complete the differentiation process and to maintain fiber cell viability.
ACKNOWLEDGEMENTS
We would like to thank Dr Lewis T. Williams (UC, San Francisco) for providing the truncated murine FGFR1 cDNA, Dr Sabine Werner (Max-Plank Inst., Martinsried, Germany) for helpful suggestions, Gabriele Schuster for microinjection of the αA-tR1 construct, Long Vien for assistance with animal husbandry, Dr Wallace L. McKeehan (IBT, Texas A&M, Houston, TX) for the FGFR1 antibody M17D10, Dr Robert Friesel (American Red Cross, Rockville, MD) for FGFR1 antibody 5G11, Dr Robert Bjercke (Texas Biotechnology Corp.) for the FGFR1 antibody Rb 2-7, Dr J. S. Zigler (National Eye Institute, Bethesda, MD) for antibodies to γ-crystallin, Dr J. Fielding Hejtmancik (National Eye Institute, Bethesda, MD) for antibodies to β-crystallin, K. Luo and Mark Bailey for technical assistance and Dr Frank J. Lovicu for insightful discussion, graphical assistance and critical review of the manuscript. We would also like to thank Robert L. Chow and Dr Richard A. Lang (Skirball Institute, NYU Medical Center) for sharing results prior to publication. These studies were supported by NIH grants EY-10448 (P. A. O.), HL-45990 and HL-48491 (J. A. T.) and an NIH NRSA fellowship EY-06511 (M. L. R.).