Forkhead Box Protein P1 Is Dispensable for Retina but Essential for Lens Development

Abstract

Purpose: Forkhead box protein P1 (Foxp1) is a transcriptional repressor expressed in many tissues. We identified Foxp1 as a highly expressed gene in retinal progenitor cells and investigated its roles during eye development.

Methods: Mouse eyes with Foxp1 gain- or loss-of-function were established in vitro and in vivo.

Results: Foxp1 overexpression in retinal progenitor cells resulted in reduced rod and increased cone photoreceptors. However, retina-specific knockout of Foxp1 was not associated with retinal differentiation abnormalities. Foxp1 was highly expressed in the lens during early development, and continued to be expressed in epithelial and cortical fiber cells until adulthood. At birth, analyses of Foxp1 lens-specific knockout (Foxp1-L-CKO) mice showed no gross morphologic changes in germinal or central epithelial cell compared to the controls. However, the numbers of proliferating and apoptotic cells were significantly increased in Foxp1-L-CKO mice. In addition, clear Y-structures were not observed in either the posterior or anterior sutures of the Foxp1-L-CKO lenses. Mature lenses of Foxp1-L-CKO mice were small and opaque. The fiber cell structure in the core and the cortical fiber cell columns were disturbed in Foxp1-L-CKO mice at postnatal day 14, potentially accounting for the opacity. In addition, epithelial cells were not aligned into columns along the transition zone in Foxp1-L-CKO mice. Taken together, these results suggest that Foxp1 has a role during lens growth in epithelial and differentiating fiber cells.

Conclusions: Loss of Foxp1 results in loss of suture and fiber cell alignment, which eventually causes lens opacity, suggesting that Foxp1 has a key role in establishing cortical lens architecture.

Forkhead box protein P1 (Foxp1) belongs to the Foxp subfamily of the Forkhead box family proteins, which are characterized by the presence of a Forkhead box DNA-binding domain.¹ The four members of the Foxp subfamily (Foxp1–4), have similar structures and characteristics, including their domain composition and configuration. Foxp1 was first cloned from mouse lung and identified as a transcriptional repressor.² It has been shown to regulate the development of various tissues. Beginning in early development, mouse Foxp1 is expressed in the lung, heart, brain, and thymus,² and has pivotal roles in their development.³ For example, Foxp1 is essential for cardiac outflow and endocardial cushion morphogenesis; therefore, embryonic lethality results from Foxp1 knockout in mice.⁴ In hematopoiesis, Foxp1 is essential for the generation of naive T cells,⁵ and negatively regulates monocyte differentiation and macrophage function.⁶ In the central nervous system, Foxp1 and Foxp2 are expressed as early as embryonic day (E) 12.5 and their expression persists into adulthood. 

Foxp1 mutation is associated with various developmental defects as well as cancer.⁷ Foxp1 is a tumor suppressor gene; consistent with this concept, it reportedly is involved in various forms of cancer.⁸ FOXP1, FOXP2, and FOXP4 are known to function in brain development and the evolution of language. The FOXP2 mutation is known to cause developmental speech and language disorders, particularly developmental verbal dyspraxia.⁹ Global developmental delay and intellectual disability in a patient with a 0.19-Mb intragenic deletion were predicted to be attributable to haploinsufficiencey of FOXP1.⁷ In addition, autism spectrum disorders reportedly are associated with FOXP genes.¹⁰ 

In zebrafish retina, Foxp1 was found to be strongly expressed in retinal progenitors; however, as eye development progressed, its signals gradually became restricted to retinal ganglion cells (RGCs).¹¹ The roles of Foxp1 and other Foxp members have not been reported in mouse retinal development. Furthermore, its roles in mouse lens development had not been investigated to our knowledge. We found that Foxp1 was strongly expressed in mouse retinal progenitors; therefore, its roles in retinal development were analyzed. Unexpectedly, obvious defects in retinal development were not detected upon retina-specific knockout of Foxp1; however, Foxp1 was essential for lens development. 

Methods

Animal

All animal experiments were approved by the Animal Care Committee of the Institute of Medical Science, University of Tokyo and conducted in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Foxp1^flox/flox mice have been described previously.⁵ Dkk3-cre mice¹² and Pax6-lens-cre mice¹³ were kindly donated by Takahisa Furukawa (Osaka University, Osaka, Japan) and Peter Gruss (Max Planck Institute for Biophysical Chemistry, Göttingen, Germany), respectively. To get retina- or lens-specific conditional knockout (CKO) mice, we crossed Foxp1 mice (Foxp1-fl/+, or -fl/fl) with either Dkk3-cre/+ or Pax6-lens-cre/+ mice, respectively, to obtain F1 mice having Foxp1-fl and cre transgenes. Then, F1 mice were intercrossed to get CKO mice, and we used mice having genotypes homozygous for flox (fl) and heterozygous for Cre. We used littermates of conditional knockout mice, which lacked Cre, as control mice. Parental strains of both transgenic lines were backcrossed with C57BL/6J at least 5 generations. We obtained ICR mice from Japan SLC, Inc. (Hamamatsu, Japan), and we confirmed that the mice were free of Rd1 mutation. We used ICR mice for expression profiling of Foxp1 and retinal explants. For all embryonic mouse tissues analyzed, day E 0.5 was defined as the date that the vaginal plug was observed. We used approximately 8-week-old mice when we indicate “adult mouse.” 

PCR, DNA Cloning, and Foxp1 Overexpression

For RT-PCR and quantitative RT-PCR (RT-qPCR), total RNA was purified from homogenized tissues of ICR mice using Trizol (Thermo Fisher Scientific, Waltham, MA, USA), and cDNA was synthesized using using ReverTra Ace pPCR RT Master Mix (Toyobo, Osaka, Japan). Quantitative PCR (qPCR) was performed using the SYBR Green-based method, using the Roche Light Cycler 96 (Roche Diagnostics, Indianapolis, IN, USA). Glyceraldehyde 3-phosphate dehydrogenase (Gapdh) was used as an internal control. Primer sequences for qPCR are listed in the Supplementary Materials. Full length cDNA for Foxp1 was kindly provided by Yoshihiro Morikawa (Wakayama Medical University, Wakayama, Japan) and subcloned into XhoI/BamHI sites of pMXs-IRES-EGFP plasmids. Retrovirus preparation and infection and retinal explant cultures prepared from ICR mice were performed as described previously.^14,15 

Cell Proliferation Assay

For cell proliferation assay in vivo, 100 μg/g body weight of bromodeoxyuridine (BrdU; Sigma-Aldrich Corp., St. Louis, MO, USA) was injected intraperitoneally into pregnant females at a concentration of 1 μg/μl, or into pups (P0, 0.5 μg/μl). Then, tissues were dissected 1 hour afterwards. For cell proliferation assay in vitro, BrdU (1 μg/ml) was added to the media after 3 days of the culture, and retinas then were harvested after 24 hours of additional culture. Incorporated BrdU was visualized using anti-BrdU antibody as described previously.¹⁶ 

Immunohistochemistry (IHC)

Cryosections for fluorescence immunohistochemistry was prepared from fixed retinal explant or fixed mouse eyes as described previously.¹⁶ Briefly, for lens whole mount preparations, lenses were dissected from the eye, and the capsule was peeled gently by forceps to leave epithelial cells. Then, isolated lenses were fixed in 4% paraformaldehyde (PFA) for 10 minutes at the room temperature, and processed for IHC. For lens cryosections, lenses from P0 or P14 mice were isolated and fixed in either 4% or 1% PFA for 10 minutes or 24 hours, respectively. Following incubation with primary antibodies, signals were visualized with appropriate secondary antibodies and nuclear staining by 4′,6-diamidino-2-phenylendole (DAPI). Samples were observed using an Axio Imager M1 and 2 (Carl Zeiss Meditec, Jena, Germany) fluorescent microscope. The antibodies used in this study are listed in the Supplementary Table. Actin was stained with fluorescent phallotoxin (Alexa Fluor 568 Phalloidin, A12380, 1:1000 dilution; Molecular Probes, Eugene, OR, USA). Cone photoreceptors were stained with peanut germ agglutinin (PNA)-FITC (L7381, 1:200 dilution; Sigma-Aldrich Corp.). Whole mount staining was performed after washing isolated retina by 0.1% Triton X-100 in PBS twice followed by incubation in 2% horse serum/0.1% Triton X-100 in PBS. Image analysis was performed by using ImageJ (https://imagej.nih.gov/ij/; provided in the public domain by the National Institutes of Health [NIH], Bethesda, MD, USA). 

Electroretinogram (ERG)

Electroretinography was performed using LS-W (Mayo Clinic, Rochester, MN, USA) with Scope and LabChart 7 software (ADInstruments, Sydney, Australia). Mice were anesthetized by a ketamine-xylazine mixture (50 and 10 mg/kg). One drop of phenylephrine (1%) was applied for pupil dilation. Scotopic ERGs were recorded with increasing intensities of light flashes in the dark-adapted state (>12 hours). Flashes were presented at approximately 30- to 60-second intervals. Photopic ERGs were recorded after light adaptation (>2 hours) with a background illumination of 35 cd/m². Flashes were presented at 1-second intervals. A total of 32 trials were averaged for single-flash responses. 

Gene Expression Analysis by Microarray

Total RNA was purified from the lenses of Foxp1 lens-specific knockout (Foxp1-L-CKO) mice and their littermate controls (8 lenses for each pool) using RNeasy (Qiagen, Hilden, Germany), and quality of purified RNA was confirmed by 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA) and stored at −80°C. Microarray was performed using SurePrint G3 Mouse GE microarray 8 × 60 K (Agilent Technologies). Briefly, synthesis of labeled cRNA was performed by using Low Input Quick Amp Labeling Kit (Agilent Technologies) according to the manufacturer's protocol. Microarrays were hybridized with labeled cRNAs, then the slide was washed and signal intensities were measured by using the SureScan Microarray Scanner (Agilent Technologies). The Feature Extraction software (Agilent Technologies) was used to produce the processed signal values, and the values were normalized by 75 percentile by using the GeneSpring software (Tomy Digital Biology, Tokyo, Japan). Ontology analysis was performed using DAVID (National Institute of Allergy and Infectious Diseases, NIH). The log2 value of Foxp1-L-CKO was subtracted by that of control littermate, and the subtracted δ values are shown as log2-fold change. MatInspector (Genomatix, Munich, Germany)¹⁷ was used to search for Foxp1 binding sites. 

Table 1

 

Gene Expression in Foxp1-L-CKO Lenses at P0

 

Table 2

 

Gene Expression of Eph/Ephrin Family Members in Foxp1-L-CKO

 

Results

Foxp1 Expression Changes With Development in the Retina and Lens

At first, the expression of Foxp1 in the lens, cornea, and retina at E17.5 and in adulthood eyes of ICR mice was examined (Fig. 1A). Foxp1 was expressed in all the examined samples in embryonic and adult tissues. We then examined the transition of expression of Foxp1 and other Foxp family members (2–4) in developing mouse retina by RT-qPCR (Fig. 1B). The expression of Foxp1 was intense in the early embryonic retina, and sharply decreased from E14.5 to E17.5 (Fig. 1B). Then, after birth, the expression level continuously and gradually decreased until adulthood (Fig. 1B). The expression level of other members of the Foxp family (Fogxp2–4) were low at E14.5; level of Foxp2 transiently increased at approximately E17.5, then decreased continuously and gradually. Foxp3 expression began at E17.5 and gradually increased until approximately at P5, then decreased gradually. Foxp4 showed weak expression at E14.5, then slightly decreased at E17.5 and achieved a mild peak at approximately P5 (Fig. 1B). 

Figure 1

 

Expression of Foxp1 in developing retina (Re) and lens (Le). (A) Reverse transcription-PCR of embryonic E17.5 and adult ocular tissues show that Foxp1 is expressed in lens, cornea (Co), and retina. Gapdh was used as a loading control, and Crybb1 was used to verify lens tissue, at embryonic and adult (Ad; 8 weeks) ICR mice. (B) Quantitative RT-PCR of Foxp genes in retina at different developmental stages (E14.5-Adult) reveal varying patterns of expression, with Foxp1 and Foxp2 being highly expressed at developmental stages, and Foxp3 and Foxp4 being maximally expressed at perinatal stages. Gapdh was used as a control. (C–K') Spatial expression of Foxp1 during eye development. Frozen sections of ICR mouse eyes at indicated developmental stage were immunostained for Foxp1 (green), and nuclei were stained with DAPI (red). (C–E) Cover the whole eye, and (F–H) show part of the retina with apical side to the bottom. (J, K) Show the region around the lens equator, and the anterior pole of the same samples are shown in (J') and (K'). NBL, neuroblastic layer; INL, inner nuclear layer. Scale bars: 50 μm.

We next examined the spatial expression pattern of Foxp1 protein in the retina by immunohistochemistry. Foxp1 was localized to the neuroblastic layer during early retinal development (Figs. 1C, 1D), while cells on the basal side of the neuroblastic layer were only weakly positive for Foxp1 at E13.5 (Figs. 1E, 1F; Supplementary Fig. S1A). Double staining with the anti-Ki67 antibody for proliferation showed that Foxp1 was expressed in proliferating retinal progenitor cells (Supplementary Fig. S1A). With maturation of the retina, Foxp1 expression decreased and converged on a subset of cells in the ganglion cell layer (GCL) at E17.5 and P0 (Fig. 1G; Supplementary Figs. S1C, S1D), and remained in a very small number of cells within the GCL in P10 and adult retinas (Fig. 1H; Supplementary Figs. S1E, S1F). Double staining with Brn3b (Pou4f2) showed that Foxp1 was expressed in some of the Brn3b-positive RGCs (Supplementary Figs. S1B–G), but not in choline acetyltrasferase (ChAT)–positive amacrine cells (Supplementary Fig. S1H), suggesting that Foxp1 expression in the GCL represented a subpopulation of RGCs and not displaced amacrine cells. In addition to the GCL, Foxp1 was expressed in the ciliary epithelium from the embryonic period to adulthood (Supplementary Figs. S1I–L). 

In the developing lens at E9.5, Foxp1 was expressed broadly in the lens placode in addition to the optic vesicle (Fig. 1C). At E11.5, strong Foxp1 expression was observed in the epithelial cells of the lens (Fig. 1D), while weak immunoreactivity was present in the elongating primary fiber cells. By E13.5, the expression in epithelial cells near the lens equator became more intense than that in the epithelium at the apical pole and fiber cells in the transition zone (Figs. 1E, 1I). From P0 onwards, Foxp1 was expressed in the lens epithelium and lens fiber cells near the equator (Figs. 1J, 1K; Supplementary Figs. S2A, S2B). Its expression also was observed in central epithelial cells at the anterior pole (Figs. 1J', 1K'). Almost all lens epithelium expressed Foxp1, and some coexpressed Foxp1 and Ki67 (Supplementary Fig. S2C). Incorporation of BrdU and phosphor-histone H3 staining confirmed that the Foxp1 was expressed in proliferating cells (Supplementary Figs. S2D, S2E). 

Foxp1 Gain-of-Function Leads to Fewer Rods and More Cones

To examine the biological function of Foxp1 during retinal development, we first attempted to overexpress Foxp1 using retrovirus-mediated gene transduction in a retinal explant culture. A retrovirus encoding Foxp1 and enhanced green fluorescent protein (EGFP) was introduced into isolated retinas at E17.5, and the retinas were cultured as explants for 2 weeks. Overexpression of the Foxp1 protein was confirmed by immunostaining (Fig. 2A). Cells positive for EGFP were distributed mostly in the outer nuclear layer (ONL), but the proportion of cells in the ONL was lower in Foxp1-overexpressing retina and that in the inner nuclear layer was increased (Fig. 2B). Furthermore, Foxp1-expressing cells tended to localize more on the internal side of the ONL (Fig. 2C). Cone cell body tended to be localized on the internal side of the ONL in the explant culture; in fact, immunostaining with retinal subtype markers showed that Foxp1 overexpression suppressed PNR (Nr2e3)–positive rod photoreceptor differentiation and promoted the appearance of Rxrγ-positive cone cells (Fig. 2D; Supplementary Figs. S3A, S3B). The number of amacrine cells, detected using the HuC/D (Elavl3/4) marker, was slightly increased (Fig. 2D; Supplementary Fig. S3A). The number of Vsx2-positive bipolar cells and Glutamine synthetase (GS)–positive Müller glia was not different or rather decreased in Foxp1-expressing retina (Fig. 2D), respectively. Changes in Foxp1 expression did not result in significant changes in BrdU incorporation or the proportion of cells expressing Ki67 (Figs. 2E, 2F), suggesting that Foxp1 overexpression at this stage did not affect the proliferative activity of retinal progenitor cells. 

Figure 2

 

Overexpression of Foxp1 in retinal progenitors decreased rod and increased cone photoreceptors. (A) Immunolabeling of frozen cryosections from retinas transduced with Foxp1-EGFP or EGFP only (control) retroviruses at E17.5 showed efficient transduction and effective induction of ectopic Foxp1 expression in the outer nuclear layer after 2 weeks of culture. (B) Quantification of EGFP-positive cells showed differences in the relative numbers in the outer and inner nuclear layers of control and Foxp1 transduced retinas. (C) Number of EGFP-positive cells in ONL in three horizontally divided regions of the ONL was counted. Population in each subregion of ONL was expressed as percent population in total EGFP-positive cells in ONL, showing differential distribution of EGFP-positive cells in the ONL. (D) After 2 weeks of culture, frozen sectioned retinas were stained with various retinal subtype markers. Markers are photoreceptor-specific nuclear receptor (PNR; Nr2e3) for rod, Rxrγ for cone, GS for Müller glia, Vsx2 for bipolar cells, and HuC/D for amacrine cells. Marker/EGFP double positive cells in total EGFP-positive cells are shown. Decreased PNR-positive cells and increased Rxrγ-positive cells were observed. (E, F) Proliferation activities after 4 days of culture of retinal explants prepared from E17.5 retina were examined by BrdU incorporation or Ki67 staining. Population of proliferating EGFP-positive cells in total EGFP-positive cells are shown in (F); PNR-positive cells already were much lower in Foxp1 expressing samples. Nuclei were visualized with DAPI (blue) in (A) and (E). ONBL, outer neuroblastic layer. Scale bars: 50 μm. Cells were counted in more than 2 photos of at least 3 independent explants for each category in (B–D, F). Data are expressed as average value with standard deviation. *P < 0.05 and **P < 0.01 (Student's t-test with 2-tails).

No Obvious Retinal Development Defects in Mice With Retina-Specific Knockout of Foxp1

We next performed a Foxp1 loss-of-function analysis using mice with Foxp1 retina-specific knockout (Foxp1-R-CKO). Foxp1-flox mice⁵ were crossed with Dkk3-cre mice,¹² which expressed Cre in retinal progenitor cells. Foxp1 expression was absent from the retinas derived from Foxp1-R-CKO (Fig. 3A), as expected. The three retinal nuclear layers formed appropriately (Fig. 3B), and the thickness of each layer was similar between the control and Foxp1-R-CKO mouse retinas (Fig. 3C). In addition, all of the examined subtype markers were similarly expressed in the retinas of the control and Foxp1-R-CKO mice (Supplementary Fig. S4A). Since Foxp1 was expressed strongly in E11.5–13.5 retinas, which corresponds to the period in which amacrine cells are generated, we examined the differentiation of subtypes of amacrine cells in more detail (Figs. 3D, 3E; Supplementary Fig. S4B). Again, no significant difference was observed. Foxp1 also is strongly expressed in RGCs, but no significant difference was observed in the distribution and number of RGCs between control and Foxp1-R-CKO mouse retinas following whole-mount immunostaining (Fig. 3F; Supplementary Figs. S4C, S4D). During early development, the proliferation of retinal progenitors was comparable between control and Foxp1-R-CKO mouse retinas (Supplementary Fig. S4E). Furthermore, the differentiation of RGCs at E14.5 was indistinguishable (Supplementary Fig. S4F). 

Figure 3

 

Retina development of Foxp1-retina specific conditional knockout (Foxp1-R-CKO) was indistinguishable from that of control mouse. (A) The expression of Foxp1 in developing retinas derived from Foxp1-R-CKO and control. Frozen sections of the retinas of Foxp1-R-CKO and control were immunostained with Foxp1 antibody (green). Arrowheads indicate immune-signal–positive cells. (B). Staining pattern with DAPI of adult retinas of Foxp1-R-CKO and control. Scale bars: 50 μm (A, B). (C) Thickness of retinal three layers of adult Foxp1-R-CKO and control was examined and expressed as percent to the whole thickness. (D, E) Populations of cells positive labelled for calbindin (D, horizontal cells and a subset of amacrine cells) and ChAT (E, cholinergic amacrine cells) in each retinal sublayer are shown. (F) Number of Brn3b-positive RGCs was counted on whole mount retina. ns, not significant, calculated by Student's t-test with 2-tails (C–F). All samples contain more than 3 retinas from independent mice. (G–K) Two-month-old Foxp1-R-CKO or control littermates of either sex were analyzed with ERG under scotopic (G, H) or photopic conditions (I). Traces showing scotopic b-wave (G) representing rod bipolar depolarization, a-wave of cone-rod mixed maximum response (H) representing rod photoreceptor depolarization, and photopic response (I) representing cone pathways. (J, K) Stimulus strength-amplitude relationship was analyzed for scotopic a- and b-waves, respectively. The experiments were done with 3 mice for each genotype, and representative data are shown for (G–I), and average with standard deviation are shown in (J) and (K).

Finally, retinal function in Foxp1-R-CKO mice was examined using ERG. Foxp1-R-CKO mice exhibited comparable scotopic b-waves and a-wave with those of their control littermates (rod activity; Figs. 3G, 3H). In photopic ERG, which reflected the cone pathway, the a-, b-, and op-waves in Foxp1-R-CKO mice were indistinguishable from those in control mice (Fig. 3I). The responses of scotopic a- and b-waves to a stepwise stimulus increase were similar, (Figs. 3J, 3K, respectively), suggesting that Foxp1-knockout retina retained fully functional rod and cone pathways. 

Severe Perturbation of Lens Development in Foxp1-L-CKO Mice

To investigate the role of Foxp1 in lens development, we crossed Foxp1-flox mice⁵ with lens-specific Pax6-Cre transgenic mice¹³ and obtained Foxp1-L-CKO mice. These Foxp1-L-CKO mice were born in accordance with the Mendelian law of inheritance, and appeared healthy. At P0, expression of Foxp1 protein in the lens was almost completely abolished in Foxp1-L-CKO (Supplementary Figs. S5A, S5B'). The eyes of Foxp1-L-CKO mice and littermate controls opened approximately 2 weeks after birth as expected, and appeared normal; Foxp1-L-CKO eyes did not exhibit redness or abnormal growth around the eye, neither showed signs of clouding of the cornea (Fig. 4A and data not shown). Two weeks after birth (P14), the eyes of Foxp1-L-CKO mice were smaller than those of their littermate control group (Fig. 4B). The lenses of these Foxp1-L-CKO mice were also smaller and exhibited nuclear opacity (Figs. 4C–E). Control and heterozygous (Foxp1-fl/+, Cre+) mice lens did not develop such phenotype even at 8 weeks after birth (Fig. 4C'). Sections of lens also confirmed smaller diameter of the lens of Foxp1-L-CKO (Figs. 4D, 4E). We then examined the actin skeleton using phalloidin. When viewed perpendicular to the optical axis, lens fiber cells are normally tightly packed into columns in a highly ordered fashion. In Foxp1-L-CKO mouse lenses, the epithelial cell layer and outside capsule appeared to be intact (Figs. 4F, 4H, left side); however, the lens fiber cells in the cortex of Foxp1-L-CKO mouse lenses appeared disorganized in their alignment (Fig. 4H). The hexagonal cross-sectional shape of lens fiber cells also is disturbed as evident by phalloidin staining, and this may represent abnormal fiber cell membrane architecture. In the lens core, the phalloidin staining in the control lenses showed packed cells in the center (Fig. 4G); however, in the Foxp1-L-CKO lens core, cell borders appeared to be ruptured or merged into what appears to be a large vacuole (Fig. 4G). Such changes in the lens core may account for the opacities observed in the intact Foxp1-L-CKO lens. As fiber cells differentiate from epithelial cells, disorganization of fiber cell alignment in the lens column observed in Foxp1-L-CKO could result from abnormalities of the epithelium. We examined the alignment of differentiating lens epithelium in whole-mount preparations. While in control lenses, epithelial cells align in meridional rows near the equator (Figs. 4J, 4J'). This alignment was disrupted in Foxp1-L-CKO mice (Figs. 4K, 4K'). 

Figure 4

 

Lenses of Foxp1-L-CKO mice are smaller and have abnormal fiber cell morphologies. Photograph of mice eyes (A), isolated whole eyes (B), and isolated lenses (C, C') from Foxp1-L-CKO mice and littermate control (Foxp1-fl/fl without Cre transgene) at P14. (C') Photos of isolated lenses from adult control and heterozygous (Foxp1-fl/+, Cre+) mice. (D–I') Lenses from P14 control littermates or Foxp1-L-CKO mice were fixed and frozen cryosectioned near the equator, perpendicular to the visual axis. (D, E) Show bright field photos. Letters (F, G, H, I) in (D) and (E) roughly indicate location of magnified panels (F, G–I), respectively. (F–I) Sections of lenses (F–I) were stained with phalloidin (F–I) and DAPI (F, H). Lens epithelium is located to the left of (F) and (H). Phalloidin signals were expressed in black and white in (G) and (I). (J–K') Whole mount preparations of lenses were stained with phalloidin and DAPI. The staining patterns revealed the alignment of equatorial epithelial cells into meridional rows (J, J'), which are disrupted in the Foxp1-L-CKO lenses (K, K'). Lens epithelium is located to the left of (J–K'). Note only epithelial cells near transition zone in control lenses start to align in order (*), but that is not observed in Foxp1-L-CKO. Scale bars: 200 μm (D, E), 25 μm (F, H, J/J', K/K'), and 50 μm (G, I).

Effect of Foxp1 Depletion on Cell Cycle Regulation in Lens Epithelial Cells

Based on the finding of abnormal lens fiber cells at 2 weeks after birth, we next examined younger Foxp1-L-CKO mouse lenses. At P0, no morphologic changes were evident in the nucleus of Foxp1-L-CKO mouse lenses (Supplementary Figs. S5C, S5D); however, phalloidin staining signals in Foxp1-L-CKO were not straight as observed in controls, indicating that epithelial cell alignment already was perturbed (Supplementary Figs. 5C–F'). We then analyzed cell proliferation and apoptosis. At P0, the proportion of BrdU-positive epithelial cells was slightly increased in Foxp1-L-CKO mouse lenses (Figs. 5A, 5B, 5I). Similar trends were observed for Ki67 and phospho histone 3 (PH3), although the changes were not statistically significant (Figs. 5C–F, 5I). The distribution of proliferating cells was similar between Foxp1-L-CKO and control (Supplementary Figs. S6A–C). The proportion of cells stained by active caspase 3 (AC3) also was significantly increased in Foxp1-L-CKO (Figs. 5G, 5H, 5J), and they were observed near the equator or near the anterior pole (Supplementary Fig. S6D). We observed that, in Foxp1-L-CKO mouse lens, more proliferating cells were scattered near the pole regions and AC3-positive cells were adjacent to the BrdU-positive cells (Figs. 5K, 5L). We also performed similar analysis of lens at E14.5 (Figs. 5O–R). Foxp1 expression in Foxp1-L-CKO already was reduced selectively in the lens but not in the retina at E14.5 (Fig. 5P), compared to control (Fig. 5O). However, there was no significant labelling for AC-3 positive cells in either control or Foxp1-L-CKO (Figs. 5Q, 5R). Taken together, these findings suggested that while fiber cells appeared to be relatively normal at P0, epithelial cell proliferation and apoptotic cell death were increased in Foxp1-L-CKO mouse lenses at P0. 

Figure 5

 

Increased epithelial cell death in Foxp1-L-CKO lenses at P0 but not at E14.5. Lenses from control littermates (A, C, E, G, O, Q) and FoxP1-L-CKO (B, D, F, H, K, L, P, R) lenses at P0 (A–L) or at E14.5 (O–R) were fixed, sectioned parallel to the visual axis, sections collected near the center and stained with BrdU (A, B, green), Ki67 (C, D, green), PH3 (E, F, green), active caspase-3 (G, H, green), and counterstained with DAPI (blue). The transition zone is marked as “TZ” in (A–H). The number of epithelial cells positively stained was counted and normalized by number of DAPI-positive epithelial cells (I, J). The average proportion of positively stained cells of BrdU (n = 6), Ki67 (n = 5), PH3 (n = 5), AC3 (n = 6 and 5) are shown. Data represent means ± standard deviation. *P < 0.05 and **P < 0.01 (Student's t-test with 2-tails). (K, L) Epithelial cells were triple stained with BrdU (red), AC3 (green), and DAPI (blue) to show that apoptotic cells were all BrdU-negative but often were found next to the BrdU-positive cells. (O–R) E14.5 Control and Foxp1-L-CKO lenses were immunostained for Foxp1 (O, P, green), Ki67 (Q, R, red) AC3 (Q, R, green), and counterstained with DAPI (blue). Closed arrows in (B, H, L) indicate some of positively stained cells. Scale bars: 100 μm (A–H), 25 μm (K–L'), 50 μm (O–R).

Loss of Foxp1 Disturbs Lens Fiber Cell Maturation

Having established that epithelial proliferation was affected in Foxp1-deficient lens at P0, we further analyzed Foxp-L-CKO lenses by focusing on the fiber cells in equatorial and axial sections at P0. When we performed axial sectioning, although the distribution of nuclei in the control lenses at the bow region showed a bow-like appearance (Figs. 6A, 6C), the nuclei in Foxp1-L-CKO lenses were distributed above and below the lens equator, even after the fiber cells had migrated from the equatorial region of the lens into the interior of the lens (Figs. 6B–6D). 

Figure 6

 

Foxp1 CKO lens fiber cells have defective suture formation and nuclei migration. (A–D) Lenses from littermate control (A) and Foxp1-L-CKO (B) mice at P0 were fixed and sectioned parallel to the visual axis, stained with DAPI. Nuclei distribution relative to the distance from the center of the lens (a = 0), was quantified as average number of nuclei per 0.1 × 0.1 (a/r) square and shown as heat map (C). In (C), the regions indicated by stripes, which are located in left most side, were excluded from the analysis because this region contains epithelial cell nuclei, and we wanted to focus on the location of fiber cell nuclei only. (D). Average number of differentiating fiber cell nuclei in the bow regions of littermate control and Foxp1-L-CKO lenses counted on the cryosection. (*P < 0.05 calculated by Student's t-test with 2-tails, n = 6 for control and n = 5 for Foxp1-L-CKO). (E–J) Analysis of lens suture at P0. Lenses from littermate control (E, G, I) and Foxp1-L-CKO (F, H, J) mice at P0 were fixed, serial sectioned from anterior pole to posterior pole perpendicular to the visual axis, and stained with phalloidin. Scale bars: 25 μm (E–H), 50 μm (I, J).

We next focused on the formation of the sutures, a characteristic structure formed as fiber cells from opposing hemispheres meet and form junctions at the apical and posterior poles of the lens, resulting in a characteristic Y-shaped suture in murine lens.¹⁸ While Y-shaped sutures were clearly evident in control lenses at P0, fiber cells in Foxp1-L-CKO lenses appeared to form junctions with other fiber cells, but the characteristic Y-shape was not evident and the fiber cell alignment appeared disorganized (Figs. 6E–J). We considered the possibility that this was due to the loss of some junction proteins in the lens, such as β-catenin, N-cadherin (Cdh2), and Zo-1 (Tjp1); however, their expression in Foxp1-L-CKO lenses appeared normal (Supplementary Figs. S7A–J). Pax6 signals were observed in the epithelium and germinative zone, and the expression patterns in Foxp1-L-CKO showed stronger signals in fiber cells of Foxp1-L-CKO (Supplementary Figs. 7K, 7M). When we quantified the signal intensity of Pax6 in the lens epithelium and fiber cells, average of intensity in Foxp1-L-CKO in fiber cells showed a much larger value than that for control littermates, but the difference was not statistically significant (Supplementary Fig. S7O). Prox1 signals were subjacent to the germinative zone epithelium in the Foxp1-L-CKO (Supplementary Figs. S7L, S7N), suggesting abnormal elongation and differentiation of these cells at the equator and transitional zone. Finally, Pericentrin, which is a component of centrosomes and expressed in juxtanuclear spots in cells, also was not perturbed in Foxp1-L-CKO mice (Supplementary Figs. S8A–K). 

Changes of Gene Expression in Lens From Foxp1-L-CKO

To investigate whether gene expression changes take place or not in Foxp1-L-CKO lenses, we isolated total RNA from control littermates and Foxp1-L-CKO lenses from 4 male mice for each sample at P0 and performed transcriptome analysis by the microarray. We focused on the expression of Crystallins and some lens structural genes^19,20 in addition to the transcription regulators,²¹ but these genes were not greatly affected by the loss of Foxp1 as summarized in Table 1. The levels of some members of Eph and Ephrin fluctuated in Foxp1-L-CKO sample (Table 2), but we could not find consistent tendency in these changes, suggesting less possibility of participation of Eph/Ephrin members in the lens phenotype observed in Foxp1-L-CKO. Further ontology analysis of approximately 1000 downregulated and approximately 270 upregulated genes, which met the condition (expression >100 and >2-fold changes), revealed a category of biological adhesion in the second largest population of terms in genes upregulated in Foxp1-L-CKO lens (Table 3). The genes, which were down- or upregulated in the Foxp1-L-CKO lenses to the control, as well as those having a signal value more than 1000 are summarized (Table 4). Analysis using MatInspector (Genomatix) showed that many of these genes contain putative Foxp1 recognition motif. The other subsets (Ckmt1, Cntn1, Gm1987, Tnfaip6, and Hdgfrp3) have putative recognition site of the forehead box. 

Table 3

 

Ontology Analysis of Genes Down- or Upregulated in Foxp1-L-CKO Lenses

 

Table 4

 

List of Genes Down- or Upregulated in Foxp1-CKO Lenses

 

* Potential Foxp1 binding site in the promoter region was analyzed by MatInspector: For genes that contained recognition sequence for Foxp1 in their 5′ promoter sequence, the P value calculated by MatInspector (the probability to obtain an equal or greater number of recognition sequences for Forkhead box family genes in a random sequence of the same size) is shown. No, no Foxp1 binding site prediected on the promoter region by MatInspector; N/A, pseudogenes whose promoter information was unavailable.

† P values lower than 0.05.

Discussion

In this study, we found that Foxp1 had a specific function in lens development, but not in the retina. There are four members of the Foxp family with similar primary structure.¹ Whether they have distinct or redundant functions has been examined in various ways. Foxp1 and Foxp2 contain an additional polyglutamine tract, which binds to CtBP1.²² A phylogenetic analysis suggested that Foxp2 and Foxp4 were more closely related to Foxp1, and Foxp3 appeared to be the most diverse member of the Foxp subfamily.²³ The lens expressed Foxp1 and Foxp4 during development, with a higher expression level of Foxp4 than Foxp1.²⁴ In lung, Foxp1, 2, and 4 are highly expressed in distinct patterns in the developing airway epithelium.²⁵ Foxp1 and Foxp2 cooperatively regulate lung and esophagus development, and both are necessary for the proper development of the lung,³ supporting the idea of their nonredundant roles. Therefore, we hypothesized that Foxp1 has specific roles in the lens that cannot be compensated for by the presence of Foxp4. By contrast, although Foxp1 is expressed strongly in the immature retina, it is dispensable. The developing retina expressed all four Foxp family members, and Foxp1 appeared at the earliest point in retinal development. However, our results suggested their redundant roles in retinal development. In contrast, gain-of-function of Foxp1 resulted in the changes of cell fate from the late born retinal subsets to the early born retinal subsets, such as cone and amacrine cells. Since proliferation activity of retinal progenitor cells was not affected, this observation may not be caused by earlier exit from the cell cycle. Early phase retinal progenitors express endogenous Foxp1 strongly, and they produce early born retinal subsets, leading us to hypothesize that expression of Foxp1 in progenitors is a permissive environment for cone and amacrine differentiation. Alternatively, Foxp1 has suppressive effects for rod cell lineage differentiation. Examination of effects of ectopic expression of Foxp1 and other members in different developmental stages of retinal progenitor cells may give critical information to clarify molecular mechanisms. 

At 2 weeks old, Foxp1-L-CKO mice exhibited a smaller lens, central fiber cell abnormality, disorganization of the lens fiber column in the lens cortex, and the misalignment of epithelial cells near the equator. These features were not evident at P0, but subtle increases in the proliferation and apoptosis of epithelial cells were observed at this time, along with malformation of the lens sutures. These seemingly separate phenotypes are probably related with regard to unique cell homeostasis in the lens. As such, cell disruption is likely to be due to the disruption of fiber cell columns in the lens cortex. It is well established that the loss of gap-junction-mediated cell coupling along the fiber cell column results in lens core abnormalities.^26,27 Disruption of the fiber cell column in turn probably results in the loss of cell alignment during the differentiation of epithelial cells, which also was observed in this study. Microarray analysis of the gene expression signature of the Foxp1-L-CKO lens and control littermates showed that approximately 6% of upregulated genes in Foxp1-L-CKO were categorized into biological adhesion. Examination of relationships between such genes and the phenotype of Foxp1-L-CKO is our next goal toward fully revealing the roles of Foxp1 for lens development. 

In terms of the regulation of proliferation and apoptosis, FOXP1 has been implicated in the regulation of tumorigenesis of various human cancers by functioning as a cancer driver or as a tumor suppressor, depending on the cell type.²⁸ As one of the mechanisms behind the oncogenic features of FOXP1, direct suppression of a wide range of proapoptotic genes in B-cell lymphomas has been reported.²⁹ Our microarray analysis showed that most of the genes were rather downregulated in Foxp1-L-CKO (Supplementary Fig. S9). However, Hrk, BH3 domain-only protein, was upregulated, and contribution of Hrk in the phenotype induced by the loss-of-Foxp1 in lens is an interesting next question. 

Plausible target genes of Foxp1 may include those that are pivotal for maintaining lens epithelium identity, since Foxp1 is a transcriptional suppressor. It should be noted, however, that Foxp1 does not appear to be essential for lens induction, as the loss of transcription factors important for lens induction would result in more marked changes in the lens at an earlier stage of development.²¹ Although the level of Lhx3 was slightly decreased in the Foxp1-CKO lens in the microarray, Foxp1 is known to suppress Lhx3 expression and promote Hox6, 9, or 10 in motor neurons,³⁰ In the retina, Lhx3 promotes the expression of genes that are involved in the early stages of differentiation, but may suppress the expression of genes that are required in mature photoreceptors in chicken retina.³¹ In addition, although the involvement of these particular members of Hox in retinal and lens development had not been reported, these molecules are candidates to clarify mechanisms of roles of Foxp1 in the eye. 

Foxp subfamily proteins are widely expressed in various tissues, but little has been reported about their functions in the eye. The results of this study implied that, although Foxp1 is essential during lens development, other members of the Foxp subfamily also may coordinate the development of the retina. Further study on their functions and mechanisms of action should contribute to revealing the transcriptional regulatory network behind the development of this complicated tissue. 

Acknowledgments

The authors thank Saigetsu Sho for initiation of this work and technical support. 

Supported by a grant-in-aid from the Ministry of Education, Culture, Sports, Science, and Technology of Japan and Research on Applying Health Technology, Health and Labour Sciences by the Ministry of Health, Labour and Welfare, Japan. 

Disclosure: H. Suzuki-Kerr, None; Y. Baba, None; A. Tsuhako, None; H. Koso, None; J.D. Dekker, None; H.O. Tucker, None; H. Kuribayashi, None; S. Watanabe, None