Identification of the L130F mutation in the FOXC1 (forkhead box C1) gene. The chromatogram shows the genomic DNA sequence of an unaffected individual and patient 2. Patient 2 has a heterozygous C-to-T transition that results in a leucine-to-phenylalanine change at codon position 130.
The L130F mutation in the FOXC1 (forkhead box C1) gene does not affect protein stability. The Xpress (Invitrogen, Carlsbad, Calif) epitope–tagged wild-type (WT) FOXC1 and L130F, transfected into COS-7 cells, were detected by immunoblotting. The L130F protein is expressed at levels similar to those of WT FOXC1 protein. Both occurred as a doublet at approximately 65 kDa. The protein size marker is indicated to the left.
The L130F mutation in the FOXC1 (forkhead box C1) gene alters phosphorylation of wild-type (WT) FOXC1 protein. In this immunoblot, the disappearance of the higher-molecular-weight bands on incubation with calf intestinal alkaline phosphatase (CIP) and their appearance with the inhibition of CIP by sodium vanadate (NaVO3) indicated that the WT FOXC1 and L130F proteins are both phosphorylated.
The L130F mutation in the FOXC1 (forkhead box C1) gene disrupts efficient nuclear localization of the FOXC1 protein. The L130F proteins, visualized by Cy3 fluorescence (red) during microscopy, showed reduced localization to the nucleus, visualized by 4′,6-diamidine-2-phenylindole staining (blue), compared with wild-type (WT) FOXC1. A total of 480 cells and 623 cells were counted for WT FOXC1 and L130F, respectively.
The L130F mutation in the FOXC1 (forkhead box C1) gene impairs DNA binding. The wild-type (WT) FOXC1 and L130F proteins were incubated with phosphorus 32-deoxycytidine triphosphate–labeled double-stranded DNA containing FOXC1-binding sites. Unlike WT FOXC1, which formed protein-DNA complexes (*), the electrophoretic mobility shift assay showed, with this autoradiogram, that the L130F protein was unable to bind to DNA even at high concentrations.
The L130F mutation in the FOXC1 (forkhead box C1) gene impairs transcriptional activation. The L130F protein transactivated the luciferase reporter with 6× FOXC1 binding sites (BS) (above the graph) at residual levels. The data show mean luciferase values, normalized to Renilla luciferase, from a representative experiment carried out in triplicate. Error bars are the standard error of the mean. WT indicates wild type.
Molecular models and scatterplot of in silico analysis of the L130F mutation in the FOXC1 (forkhead box C1) gene. The FOXC2-derived homology model of FOXC1 shows the protein backbone (ribbon), mutated residues (gray), and unmutated residues (white). The wild-type and mutant-equivalent models were submitted to an atomic nonlocal environment assessment (ANOLEA) Swiss model server. Energy differences are in E/kT units, where E represents energy; k, the Boltzmann constant; and T, absolute temperature.
Ito YA, Footz TK, Murphy TC, Courtens W, Walter MA. Analyses of a Novel L130F Missense Mutation in FOXC1. Arch Ophthalmol. 2007;125(1):128-135. doi:10.1001/archopht.125.1.128
Copyright 2007 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2007
To understand how the novel L130F mutation, found in 2 patients with Axenfeld-Rieger syndrome, disrupts function of the forkhead box C1 protein (FOXC1).
Sequencing DNA from patients with Axenfeld-Rieger syndrome identified a novel missense mutation that results in an L130F substitution in the FOXC1 gene. Site-directed mutagenesis was used to introduce the L130F mutation into the FOXC1 complementary DNA. The level of L130F protein expression was determined by means of immunoblotting. We determined the mutant protein's ability to localize to the nucleus, bind DNA, and transactivate a reporter construct.
The FOXC1 L130F mutant protein is expressed at levels similar to those of wild-type FOXC1. The L130F protein, however, migrated at an apparent reduced molecular weight compared with the wild-type protein, suggesting that the mutant and wild-type proteins may be differentially phosphorylated. The L130F protein also had a significantly impaired capacity to localize to the nucleus, bind DNA, and transactivate reporter genes.
The disease-causing L130F mutation further demonstrates that helix 3 of the forkhead domain is important for the FOXC1 protein to properly localize to the nucleus, bind DNA, and activate gene expression.
The inability of FOXC1 to function owing to the L130F mutation provides further insight into how disruptions in the FOXC1 gene lead to human Axenfeld-Rieger syndrome.
The forkhead box (FOX) family of transcription factors, including FOXC1, share an evolutionarily conserved 110–amino acid sequence known as the forkhead domain (FHD).1 This DNA-binding motif is composed of 1 minor and 3 major α-helixes and 2 β-sheets.1 The 2 β-sheets form 2 loops that wrap around the DNA, giving the FHD its characteristic winglike structure.2
Patients with FOXC1 mutations mapping to chromosome 6p25 are affected with human Axenfeld-Rieger (AR) syndrome.3,4 This genetic disease is transmitted in an autosomal dominant manner and is highly penetrant. The FOXC1 protein is expressed in many of the developing ocular tissues in the anterior chamber of the mouse eye.5 Mutations in FOXC1 result in malformations in the anterior chamber of the human eye that include iridogoniodysgenesis, iris hypoplasia, corectopia, polycoria, a prominent Schwalbe line, and iridocorneal tissue adhesions.6- 8 The most serious consequence of AR syndrome is that approximately 50% of patients develop glaucoma.9 In addition to the ocular defects, patients with AR can have systemic defects, including maxillary hypoplasia, hypodontia, and a protruding umbilicus.10- 12
Correct FOXC1 expression is crucial for embyrogenesis and, in particular, for the normal development of the skeletal, cardiovascular, urogenital, and ocular tissues.13- 17 Furthermore, FOXC1 continues to be expressed in several adult tissues, including the eyes, brain, heart, and kidneys.16 Because the spatial and temporal patterns for FOXC1 expression have been observed to coincide with the differentiation of specific tissues,16 mutations in the FOXC1 gene usually result in gross morphological defects.
For normal development, not only does FOXC1 need to be expressed in the appropriate spatial and temporal patterns, but the level of FOXC1 expression must be strictly regulated.18 Thus, AR syndrome can result from either a loss-of-function mutation, in which FOXC1 expression is less than the critical lower threshold of 80% of wild-type activity levels, or a mutation causing FOXC1 expression to exceed the upper threshold of 150%.18 No strong genotype-phenotype correlation has been established, which suggests that the process for developing a particular phenotype in a patient with a mutation of the FOXC1 gene is a complex one.19
We have identified a novel FOXC1 missense mutation, L130F, in 2 related individuals with AR syndrome. The hydrophobic L130 residue is located in helix 3 of the FHD. Helix 3 is referred to as the recognition helix because it interacts with the major groove of DNA.20 Previous studies have found that missense mutations that occur within helix 3 disrupt DNA binding and subsequent transcriptional activation.21 Thus, we performed a molecular analysis to examine how the disease-causing L130F mutation disrupts FOXC1 function. Molecular analyses of FOXC1 missense mutations such as L130F will provide further insight into how disruptions in FOXC1 lead to human AR syndrome and will give new possibilities for the development of treatments for this disease.
This research adhered to the tenets of the Declaration of Helsinki. Patient samples and information were collected with informed consent as specified by the University of Alberta Ethics Board. The L130F mutation was identified in a white woman (patient 1) and her son (patient 2). Patient 1 was diagnosed as having AR syndrome at 27 years of age, after her son was diagnosed as having AR syndrome and glaucoma at 2 months of age. Patient 1 had no dental or facial abnormalities. An ophthalmological examination revealed that patient 1 had iris hypoplasia, a prominent Schwalbe line, and peripheral anterior synechiae, but no glaucoma. Patient 2 was diagnosed as having AR syndrome because he had corectopia and hypertelorism. He had a slight excess of skin at the umbilical region. Results of the ophthalmological examination disclosed a posterior embryotoxon. Patient 2 also had an intraocular pressure of 14 mm Hg, but abnormally high cup-disc ratios of 0.85 and 0.80 for the left and right eyes, respectively. Both patients were bilaterally affected. The maternal grandparents did not have the L130F mutation, indicating a de novo mutation in patient 1.
The FOXC1 gene was amplified as previously described.2 Polymerase chain reaction products were gel purified, extracted on separation columns (Qiagen, Valencia, Calif), and sequenced directly by using a phosphorus 33–labeled terminator cycle sequencing kit (Amersham Biosciences, Baie d’Urfe, Quebec). In addition, the polymerase chain reaction products were sequenced using a 3130 × l genetic analyzer (Applied Biosystems Inc, Foster City, Calif) to generate chromatograms and to confirm the observations from the manual sequencing experiments.
The FOXC1 pcDNA4 His/Max B (Invitrogen, Carlsbad, Calif) has been described previously.18 Site-directed mutagenesis was performed using a mutagenesis kit (QuickChange; Strategene, La Jolla, Calif) with the addition of 10% dimethylsulfoxide. The mutagenic primer sequences for L130F were as follows: forward, 5′-agc atc cgc cac aac ttc tcg ctc aac gag tgc-3′; reverse, 5′-gca ctc gtt gag cga gaa gtt gtg gcg gat gct-3′. Potential mutant constructs were sequenced with the 3130 × l genetic analyzer. Confirmed mutants were subcloned into the FOXC1 pcDNA4 His/Max vector and resequenced.
We cultured COS-7 cells and HeLa cells in Dulbecco modified Eagle medium and 10% fetal bovine serum at 37°C.
The COS-7 cells (106 cells per 100-mm plate) were transfected with 4 μg of FOXC1 tagged with Xpress epitope (Invitrogen) using commercially available reagent (Fugene 6; Roche, Indianapolis, Ind). Forty-eight hours after transfection, the proteins were extracted and resolved on a 10% sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) gel. The proteins were detected by immunoblotting using an anti-Xpress antibody (1:10 000 dilution; Invitrogen) against the pcDNA4 His/Max vector–encoded N-terminal Xpress tag and visualized with chemiluminescent substrate (Supersignal West Pico; Pierce Biotechnology, Rockford, Ill). Protein extracts were incubated with 20 U of calf intestinal alkaline phosphatase (Invitrogen) with or without 11μM sodium vanadate for 1 hour in a 37°C water bath. An equal amount of 2× SDS-PAGE loading buffer was added, and the proteins were then resolved on a 10% SDS-PAGE gel and visualized by immunoblotting as described.
The COS-7 cells (2 × 105 cells per 35-mm well) were grown and transfected directly on coverslips with 1 μg of Xpress-epitope–tagged FOXC1. The COS-7 cells were also transfected with the pcDNA4 His/Max vector as a control experiment. Twenty-four hours after transfection, the localization of the FOXC1 protein was visualized by incubating the coverslips with anti-Xpress antibody and antimouse Cy3–conjugated secondary antibody. The position of the nucleus was visualized by staining with 4′,6-diamidine-2-phenylindole. First, a total of 480 cells transfected with wild-type FOXC1 were scored for nuclear or cytoplasmic staining or both. Then, a total of 623 cells transfected with L130F were scored.
The amount of L130F protein in the COS-7 cell extracts was equalized to wild-type FOXC1 levels by inspection of the proteins detected by immunoblotting. The protein extracts were incubated for 30 minutes at room temperature with 1.25mM dithiothreitol, 0.3 μg of sheared salmon sperm DNA, 0.125 μg of poly dI/dC (Sigma-Aldrich Corp, St Louis, Mo), and 80 000 counts per minute of phosphorus 32 (32P)–deoxycytidine triphosphate–labeled double-stranded DNA containing the following FOXC1 binding site (shown underlined): forward, 5′-gatccaaa gtaaataaa caacaga-3′; reverse, 5′-gatctctgttg tttatttac tttg-3′.18 After prerunning the 6% polyacrylamide gel containing Tris-glycine-EDTA buffer for 15 minutes, the electrophoretic mobility shift assay (EMSA) reaction products were subjected to electrophoresis. As a control, the COS-7 cells were transfected with the pcDNA4 His/Max vector. Also, a mock EMSA reaction was carried out with just the 32P-deoxycytidine triphosphate–labeled double-stranded DNA to ensure that the EMSA was specific for the ability of the FOXC1 protein to bind to DNA.
The HeLa cells (4 × 104 cells per 15-mm well) were cotransfected with 100 ng of the FOXC1 pcDNA4 His/Max construct, 20 ng of the pGL3-TK construct with 6× FOXC1 binding sites,18 and 1 ng of the pRL-TK control plasmid. Forty-eight hours after transfection, the luciferase assays were carried out using the dual luciferase assay kit (Promega Corp, Madison, Wis). Each experiment was done in triplicate and was performed 3 times. As a control, the HeLa cells were transfected with the pcDNA4 His/Max empty vector instead of the FOXC1 pcDNA4 His/Max construct.
In silico mutagenesis of the FOXC2 FHD model was performed using the Swiss-PdbViewer (http://ca.expasy.org/spdbv/), and the models were evaluated with the atomic nonlocal environment assessment (ANOLEA) Swiss model server (http://swissmodel.expasy.org/anolea/), as described previously.22
The FOXC1 gene was screened for mutations by direct sequence analysis of polymerase chain reaction products of patient DNA. A heterozygous C-to-T transition at codon position 130 (388C>T; L130F) that results in a leucine-to-phenylalanine change was detected in patients 1 and 2 (Figure 1). Sequencing of the patients' DNA and 100 healthy control chromosomes confirmed that the L130F mutation is not present in the healthy population.
Immunoblotting indicated that a plasmid containing a complementary DNA encoding FOXC1 with the L130F mutation and transfected into COS-7 cells was capable of expressing the L130F protein, which was approximately the same size as the wild-type protein (approximately 65 kDa) (Figure 2). Thus, the relative stability of L130F protein expression was similar to that of the wild-type protein. However, the L130F construct repeatedly produced a pattern of immunoreactive degradation products that differed from the wild-type construct pattern (data not shown). Similar to the wild-type sample, the L130F protein bands occurred as a doublet (Figure 2). However, 1 band was slightly shifted so that it had a lower molecular weight than either of the wild-type bands, suggesting that the L130F protein was modified differently than the wild-type protein. Because FOXC1 is known to be phosphorylated,23 we examined whether the L130F mutation affected protein phosphorylation. When the L130F protein was incubated with calf intestinal alkaline phosphatase, the higher-molecular-weight band was eliminated (Figure 3). When the L130F protein was incubated with the phosphatase inhibitor sodium vanadate, the doublet was restored. This finding indicated that phosphorylation of the L130F protein was responsible for the occurrence of the immunoreactive doublet bands.
Immunofluorescent microscopy was performed to determine whether the L130F protein was able to localize to the nucleus. Only 33.7% of the L130F proteins localized exclusively to the nucleus, compared with 92.5% for the wild-type proteins (Figure 4). The COS-7 cells transfected with the pcDNA4 His/Max vector showed no staining with Cy3 (data not shown), indicating that the immunofluorescence observed was specific for the FOXC1 protein. These data indicate that the L130F mutation severely disrupts the ability of the FOXC1 protein to localize to the nucleus.
The DNA-binding ability of the L130F protein, expressed in the COS-7 cells, was determined by EMSA results. For the wild-type protein, the amount of protein-DNA complexes that formed increased as the amount of protein was increased (Figure 5). In contrast, the L130F protein showed a greatly reduced capacity to bind DNA, even when the amount of protein added to the EMSA reaction was increased (Figure 5), indicating that this mutation significantly disrupts the normal DNA-binding capacity of the FOXC1 protein.
The ability of the L130F protein, expressed in the HeLa cells, to activate expression of a luciferase reporter containing 6 consensus FOXC1 binding sites was determined. Wild-type FOXC1 was able to activate expression of the luciferase reporter (Figure 6). The transactivation potential of the mutant protein was reduced 3-fold compared with wild-type FOXC1 (Figure 6). The transactivation potential of L130F is comparable to the transactivation potential of the empty expression vector, indicating that the L130F mutation severely disrupts the ability of the FOXC1 protein to activate a reporter gene.
Finally, molecular modeling of the FOXC1 FHD was performed to predict which amino acid residue contacts would be disrupted by the L130F mutation. The first model layer of the nuclear magnetic resonance–solved structure file of FOXC224 was used as a homology model for FOXC1 because of its near-perfect sequence identity over the FHDs; the only differences are that FOXC1 contains aspartate residues at positions 96 and 117, whereas FOXC2 has glutamate residues. In this homology model, L130 was mutated to phenylalanine in silico to predict structural defects in the L130F molecule via the ANOLEA mean force potential calculations. We compared the results with those for I87M (Figure 7) and indicate that, although the side chains of I87 and L130 are normally involved in the same hydrophobic cluster, mutations at these positions may produce different effects. When I87 was changed to a methionine residue, the ANOLEA scores for M87, I104, I126, L130, F136, and W152 were all affected, whereas the effect of L130F was limited to positions 87, 130, and 152. Although recombinant FOXC1 harboring a mutation at position I87 does not produce a stable protein,25 recombinant FOXC1 harboring an L130F mutation was recoverable in whole-cell extracts (Figure 2). Thus, the sum of the ANOLEA energy differences for any single mutation model does not necessarily predict the degree of stability of the expressed recombinant protein. Nevertheless, the model is able to predict that the L130F mutation will result in severe disruptions to FOXC1 function.
The disruption of FOXC1 function by the novel disease-causing L130F mutation demonstrates the importance of helix 3 in FOXC1 function. Helix 3 of the FOXC1 protein interacts with the major groove of DNA and confers DNA-binding specificity. However, molecular modeling predicts that the L130 residue does not make direct contact with DNA (data not shown). Rather, the L130 residue is thought to be oriented toward the hydrophobic core and forms a pocket with other residues, including I87, I104, I126, F136, and W152.25 Missense mutations that substitute a differently charged amino acid, such as R127H, appear to disrupt the electrostatic charge of the FHD and thus greatly reduce the affinity of the mutant protein for DNA.19 Missense mutations such as I126E and I126K, which introduce a hydrophilic residue into the hydrophobic core, also disrupt FOXC1-DNA interactions.25 Although the interaction of the phenylalanine residue in place of the leucine residue at codon position 130 preserves the neutrally charged and hydrophobic nature of this position, the EMSA results indicate that the L130F mutation nevertheless reduces the ability of the mutant protein to bind DNA (Figure 5). The phenylalanine residue is bulkier and thus it is likely that the L130F mutation disrupts the helix 3 structure so that this helix can no longer fit into the major groove of the DNA. As a result, helix 3 of the FOXC1 protein may no longer be able to interact with DNA. The L130F mutation also reduces the transactivation potential to residual levels (Figure 5). This is consistent with the EMSA results that indicate that the L130F protein cannot bind DNA because DNA binding is a prerequisite for transactivation.
The FOXC1 protein is thought to be tightly regulated by posttranslational modifications.26 One of the ways that the FOXC1 protein is regulated is by phosphorylation.23 Previous research has determined that the phosphorylated residues of FOXC1 lie within the inhibitory domain, which is located within amino acid residues 215 through 366.23,27 Recently, the ERK1/2 mitogen-activated protein kinase–dependent phosphorylation of FOXC1 at the S272 residue was determined to stabilize FOXC1 by preventing the recruitment of degradation factors or, conversely, by recruiting stabilization factors.26 Because leucine and phenylalanine are amino acids that cannot become phosphorylated, the L130F missense mutation will not directly affect the phosphorylation state of the residue at position 130. However, the L130F protein was found to migrate at an apparently reduced molecular weight compared with the wild-type protein (Figure 2), suggesting that the mutant and wild-type proteins are differentially phosphorylated. The bulkier nature of the phenylalanine residue appears to cause enough localized structural distortion to prevent the linear amino acid sequence from folding properly. Thus, the altered topology of the L130F protein may hinder the normal recognition and regulation by protein kinases.
Mutations in FOXC2 cause hereditary lymphedema with distichiasis.21 The FHD of FOXC1 and FOXC2 have 98% sequence homology.21 An R121H missense mutation in helix 3 of the FOXC2 FHD displayed a similar migration pattern to that of the L130F mutation in FOXC1.21 The R121H and L130F proteins displayed a faster migration than did wild-type FOXC2 and wild-type FOXC1, respectively.21 Also, when treated with calf intestinal alkaline phosphatase, R121H and L130F displayed mobilities equal to those of wild-type FOXC2 and wild-type FOXC1, respectively.21 In both cases, the mutant proteins are predicted to not be phosphorylated to the full extent of the wild-type proteins. This similarity demonstrates how small changes in helix 3 of the FHD can result in great changes by altering the overall structure of the FOXC2 or FOXC1 protein.
The immunofluorescence results indicate that most of the L130F proteins are unable to localize to the nucleus (Figure 4), which is surprising because L130F is not in the regions of FOXC1 known to be directly involved in nuclear localization.23 Also, previous experiments have shown that, in many cases, substitution of a differently charged amino acid at any position disrupts the ability of the protein to localize to the nucleus to a greater extent than a substitution involving amino acids with the same charge.22 For example, a FOXC1 missense mutation involving 2 neutrally charged amino acid residues, I126A, resulted in the localization of 77% of the I126A proteins to the nucleus.25 However, when the I126 residue was replaced with a negatively charged glutamic acid residue or a positively charged lysine residue, none of the mutant proteins localized to the nucleus.25 This was not the case with the L130F mutation. Although leucine and phenylalanine are both neutrally charged, immunofluorescence showed that the L130F mutation severely disrupted the normal localization of the protein to the nucleus. Only 33.7% of the L130F proteins are able to localize exclusively to the nucleus. Because of the L130F mutation, the overall topology of the L130F protein may be altered in a manner that prevents the nuclear localization signal and nuclear localization accessory signal from being properly detected. However, the alteration in the phosphorylation pattern in the L130F protein may also contribute to the reduction in the transport of the mutant protein to the nucleus, because phosphorylation appears to regulate the nuclear transport of many transcription factors.28,29
Consistent with findings from previous studies,3,4 a single missense mutation in FOXC1 identified within a single family had variable phenotypic consequences. In the case of the 2 related individuals with the L130F mutation in FOXC1, both individuals were diagnosed as having AR syndrome. However, the mother had a mild form of the disease, whereas her son was severely affected and was diagnosed as having glaucoma at just 2 months of age. Stochastic events during development are likely to result in variable expression of downstream targets of FOXC1 in regard to the timing, location, and level of expression. Thus, although tight regulation of FOXC1 is essential for proper development, environmental factors and modifier genes may also contribute to phenotypic variability.18
Investigation of the L130F mutation also gives important insight into the likely effects of mutations of other FOX genes. A mutation equivalent to L130F has been found in the FOXL2 gene,30 where a C-to-T transition at codon position 106 (553C>T; L106F) was detected in a patient with blepharophimosis-ptosis-epicanthus inversus syndrome.30 The leucine residue at position 130 in FOXC1, which is equivalent to that at position 106 in FOXL2, is highly conserved in other FOX genes. There are at least 43 members in the human FOX gene family,31 and this L130 residue is found in 24 of those genes.30 As for L130F, we predict that the L106F FOXL2 mutation is also likely to disrupt protein function severely. A mutation of this conserved leucine residue is likely to lead to adverse functional consequences in all FOX proteins.
In this study, we report the identification of a novel missense mutation in FOXC1, L130F. The severe disruption of FOXC1 function as a result of the disease-causing L130F mutation is consistent with previously studied missense mutations located within helix 3.21 The L130F FOXC1 mutation is one of the most disruptive FOXC1 mutations studied because this mutation disrupts nuclear localization and impairs DNA binding, which subsequently impedes transcriptional activation. Thus, the analysis of the L130F missense mutation provides further insight into how disruptions in the FOXC1 FHD lead to human AR syndrome.
Correspondence: Michael A. Walter, PhD, Department of Medical Genetics, 839 Medical Sciences Bldg, University of Alberta, Edmonton, Alberta, Canada T6G 2H7 (firstname.lastname@example.org).
Submitted for Publication: June 30, 2006; final revision received September 13, 2006; accepted September 20, 2006.
Financial Disclosure: None reported.
Funding/Support: This study was supported by grant G118160216 from the Canadian Institute for Health Research (Dr Walter).
Acknowledgment: We thank May Yu for tissue culture expertise, Wim Wuyts, MD, for providing phenotypic information about the patients, and Fred Berry, PhD, for many enlightening discussions.