Identification and Analysis of a Novel Mutation in the FOXC1 Forkhead Domain

Abstract

purpose. To determine the genetic and biochemical defects that underlie Axenfeld-Rieger malformations, identify the pathogenic mutation causing these malformations, and understand how these mutations alter protein function.

methods. FOXC1 was amplified from a proband with Axenfeld-Rieger malformations and the proband’s mother. PCR products were sequenced to identify the pathogenic mutation. Site-directed mutagenesis was used to introduce this mutation into the FOXC1 cDNA. A synthetic mutation at the same position was also introduced, and both natural and synthetic proteins were tested for their ability to localize to the nucleus, bind DNA, and transactivate gene expression.

results. A novel missense mutation (L86F) was identified in FOXC1 in this family. The mutation is located in α-helix 1 of the forkhead domain. Biochemical assays showed that the L86F mutation does not affect nuclear localization of FOXC1, but reduces DNA binding and significantly reduces transactivation. The severity of the disruption to FOXC1 protein activity does not appear to correspond well with the severity of the phenotype in the patient. Analogous studies using a L86P, a known α-helix breaker, severely disrupts FOXC1 function, revealing the importance of helix 1 in FOXC1 structure and function.

conclusions. A novel mutation in helix 1 of the FOXC1 forkhead domain has been identified and the importance of position 86 in FOXC1 activity demonstrated. These studies also identified the role of helix 1 in FOXC1 function and provide further evidence for the lack of strong genotype–phenotype correlation in FOXC1 pathogenesis. Normal development appears to be dependent on tight upper and lower thresholds of FOXC1 activity.

The FOX (forkhead box) proteins have been defined by the presence of the forkhead domain (FHD), a 110-amino-acid DNA-binding motif, originally identified as a region of homology between Drosophila melanogaster forkhead protein and rat hepatocyte nuclear factor 3 proteins. ¹ Since that time, the forkhead family of transcription factors has grown rapidly, establishing the FOX family as critical regulators of embryogenesis, tissue-specific cell differentiation, cell migration, tumorigenesis, and even language and speech acquisition. ^{2 3 4}  

Mutations in the FOX family member FOXC1 lead to a clinically heterogeneous condition, Axenfeld-Rieger (AR) malformations. ^{5 6 7 8 9 10 11 12 13 14} The penetrance of ocular AR malformations is highly variable. Patients may manifest iris hypoplasia, posterior embryotoxon, adhesions of the iris and the cornea, an underdeveloped or aberrantly developed angle between the cornea and iris, and corectopia. In approximately half of the patients with AR malformations, glaucoma, a progressive, blinding condition, develops. Nonocular features include dental anomalies, maxillary hypoplasia, and redundant periumbilical skin. Congenital cardiac anomalies may also rarely be found in patients with FOXC1 mutations. ^{5 6 9 15 16}  

Molecular characterization of the 553-amino-acid FOXC1 protein has shown that it contains modular transcription activation domains—one at the N terminus and one at the C terminus of the protein—as well as a transcription inhibitory domain adjacent to the FHD. ¹⁷ Whereas the FOXC1 FHD has not been found to have modular transcriptional activation properties, analysis of two disease-causing missense mutations (F112S and I126M) within the FHD of FOXC1 have shown that perturbations that do not affect the capacity of FOXC1 to bind DNA are nonetheless able to impede the transactivational ability of FOXC1. ^{17 18} Other disease-causing missense mutations appear to impede the ability of FOXC1 to bind DNA, either leaving some residual binding activity (S82T) or disrupting binding altogether (S131L). One mutation (I87M), located in helix 1 of the FHD, appears to reduce the levels of FOXC1 protein, either by destabilizing the protein or reducing protein synthesis. ¹⁸ Studies of the FHD reveal how different missense mutations disrupt specific FOXC1 activities, showing the utility of using these naturally occurring missense mutations to understand the function of FOXC1. 

We report herein the identification, patient phenotype, and molecular analysis of a novel disease-causing missense mutation in FOXC1. Molecular analysis revealed that the L86F missense mutation, located in helix 1 of the FHD, reduces the ability of FOXC1 to bind DNA and disrupts its ability to transactivate gene expression. Analysis of a synthetic L86P mutation demonstrates the importance of helix 1 in contributing to the overall function of FOXC1. 

Methods

Patient Report

This research adhered to the tenets of the Declaration of Helsinki. The proband was a male with congenital glaucoma in the right eye leading to a complete loss of vision in the eye by age 22. In the left eye, the patient had posterior embryotoxon, iris hypoplasia, iridocorneal adhesions in the angle, and mild corectopia. Interocular pressure was more than 40 mm Hg in the patient’s left eye. Systemic anomalies included short stature and obesity, a myocardial infarction that occurred at age 41, and dental anomalies. The mother had diagnoses of iris processes to Schwalbe’s line, Haab’s striae, congenital glaucoma, obesity, short stature, and hypercholesterolemia. 

Mutation Detection

Conditions for amplification of fragments of FOXC1 are published elsewhere. ¹⁰ PCR products were gel purified and extracted with gel-affinity columns (QIAquick; Qiagen, Valencia, CA) and sequenced directly by ³³P sequencing (Amersham, Arlington Heights, IL). 

Plasmid Construction

FOXC1 pcDNA4 His/Max B(Invitrogen) has been described previously. ¹⁸ Mutagenesis was performed with a kit (QuickChange Stratagene; La Jolla, CA) with the addition of 10% dimethylsulfoxide and the appropriate primers. The mutagenic primer sequences were as follows: L86F, forward 5′-tat agc tac atc gct ttc atc acc atg gcc atc-3′; reverse 5′-gat ggc cat ggt gat gaa agc gat gta gct ata-3′; L86P, forward 5′-tat agc tac atc gcg ccc atc acc atg gcc atc-3′; reverse 5′-gat ggc cat ggt gat ggg cgc gat gta gct ata-3′. Potential mutant constructs were sequenced with an automated sequencer (LI-COR, Lincoln, NE), and confirmed mutants were subcloned into the FOXC1 pcDNA4 His/Max vector and sequenced. 

Cell Culture

COS-7 and HeLa cells were maintained in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum. Transfections were performed with a commercial reagent (FuGene6; Roche Diagnostics, Indianapolis, IN), according to the manufacturer’s protocol, using 2 μg DNA per 100-mm plate and 0.5 μg DNA per 25-mm well. 

Immunofluorescence

COS-7 cells were grown and transfected on coverslips with 1 × 10⁵ cells/mL and placed in six-well tissue culture plates. After 24 hours, immunofluorescence detection was performed as described previously. ¹⁸  

Electrophoretic Mobility Shift Assays

COS-7 extracts containing recombinant FOXC1 were equalized for amounts of recombinant FOXC1 protein by Western analysis with the anti-Xpress antibody (Stratagene) against the vector-encoded N-terminal Xpress tag. Protein extracts were incubated with 1.25 mM dithiothreitol (DTT), 0.3 μg of sheared salmon sperm DNA, 0.125 μg of poly dIdC (Sigma-Aldrich), and 20,000 cpm of ³²P-dCTP labeled double-stranded DNA containing the FOXC1 binding site, shown in italic (forward 5′-gat cca aag taa ata aa caa cag a-3′; reverse 5′-gat ctc tgt tgt tta ttt act ttg-3′). Reactions were incubated at room temperature for 15 minutes and resolved on a 6% polyacrylamide Tris-glycine-EDTA gel. 

Dual Luciferase Reporter Assays

Transactivation assays were performed as described previously ¹⁸ using a reporter assay system (Dual Luciferase; Promega, Madison, WI). 

Results

Mutation Identification

FOXC1 was screened for mutations by direct sequence analysis of PCR products of patients’ DNA. A two-nucleotide change was identified within the FHD (Fig. 1) within a patient with AR malformations. The resultant changes were a G→T transversion at position 255 and a transition of a C to a T at position 256 (255GC→TT). HhaI restriction digests of the patient’s DNA and 100 normal control chromosomes confirmed these changes. The patient’s DNA did not encode the HhaI restriction site (data not shown). The G→T transversion results in a silent mutation at codon 85 (alanine to alanine) however, the C→T transition at codon 86 results in a missense mutation, converting the amino acid from a leucine to a phenylalanine (L86F). The same FOXC1 mutation was identified in the proband’s mother. 

Molecular Analysis

The L86F mutation occurs in helix 1 of the FHD. For an understanding of how this mutation alters FOXC1 function, site-directed mutagenesis was used to introduce the L86F mutation into the FOXC1 cDNA. As a further probe of the alterations of this position, a L86P synthetic missense mutant of FOXC1 was also constructed. Prolines are known to cause drastic perturbations of the secondary structures of α-helices, thus extending our understanding of α-helix 1 function. Immunofluorescent detection of Xpress-tagged (Stratagene) recombinant FOXC1 showed that wild-type FOXC1 and FOXC1 L86F both localize predominantly to the nuclei of COS-7 cells (Fig. 2) . FOXC1 L86P drastically reduces the nuclear localization of FOXC1 protein with only 20% of the cell population showing nuclear localization of FOXC1 L86P. The FOXC1 L86P molecules localize either within the cytoplasm or into aggregates in what appears to be the endoplasmic reticulum. 

These mutations did not affect the overall stability of FOXC1, as recombinant proteins of the correct molecular weight were detected by Western analysis (Fig. 3A) . COS-7 cell extracts containing L86F or L86P recombinant FOXC1 were used in EMSAs to determine whether these mutations have an effect on the ability of FOXC1 to bind the previously described DNA-binding site. ^{18 19} FOXC1 with the L86F mutation showed an approximate 50% reduction in capacity for DNA binding in comparison with wild-type FOXC1, whereas, in contrast, the L86P mutation completely abolished DNA binding in FOXC1 (Fig. 3B) . 

FOXC1 is known to transactivate gene expression from a reporter construct. ^{17 18} The effect of the L86F and L86P mutations on the transactivation potential of FOXC1 was therefore investigated. Using a luciferase reporter construct with FOXC1 binding sites positioned upstream of a thymidine kinase promoter, we found the transactivation ability of FOXC1 to be severely perturbed by either the L86F or L86P mutation (Fig. 3C) . FOXC1 in which the FHD was deleted (ΔFHD) activated luciferase expression at only 3% of wild-type FOXC1. Similarly, FOXC1 L86F transactivated only the expression of the luciferase reporter at 6% of wild-type FOXC1 levels, showing that although able to bind the FOXC1 binding site in vitro, FOXC1 L86F was unable to transactivate gene expression. The presence of FOXC1 L86P led to 16% luciferase transactivation compared with wild-type FOXC1. 

To test the ability of either FOXC1 L86F or L86P to act in a dominant negative manner, we cotransfected wild-type FOXC1 with either FOXC1 L86F or L86P and measured the effect on transactivation. The presence of either mutant FOXC1 protein did not alter the levels of activation of the reporter construct by wild-type FOXC1 (data not shown). 

Discussion

The identification of L86F, a novel, disease-causing mutation in FOXC1, is helpful in refining our molecular model of the FHD of FOXC1. The mutation of two adjacent nucleotides is rare (CG→TT), and the mechanism for this change is unknown. The transversion of the C toT at nucleotide 257 is a silent mutation and may be a rare polymorphism. However, this silent mutation has been reported only in this proband and his mother, probably reflecting that the change is specific to this pedigree. 

Previous characterization has shown that an isoleucine mutation at position 87 leads to a drastic reduction in the levels of FOXC1 protein, presumably by destabilizing the protein. ¹⁸ That mutations at amino acid 86, including the relatively severe proline mutation, do not affect the stability of FOXC1 protein, offers insight into the role of residues within helix 1 of the FHD. An I87M mutation of FOXC1, one amino acid downstream, destabilizes the protein, demonstrating that amino acid position 87 has an important intramolecular role, contributing to the overall structure and stability of the FOXC1 protein. ¹⁸ Although the L86F and L86P mutations result in stable protein products, mutations at this position still can disrupt FOXC1 function. FOXC1 L86F is localized to cell nuclei and retains the ability to bind the FOXC1 binding site in EMSA experiments. The binding of L86F FOXC1 to the FOXC1 binding site is impaired in comparison with the wild type, but clearly the FHD retains enough of its overall structure to bind DNA. 

Compared with FOXC1 L86F, the L86P synthetic mutation severely perturbs the nuclear localization of FOXC1. Given that FOXC1 L86F localizes to the nucleus at wild-type levels, it seems likely that the inclusion of a proline within helix 1 disrupts the helical structure, perturbing the overall structure of the FHD of FOXC1 enough that the transport machinery is no longer able to accumulate FOXC1 in the nucleus. The L86P alteration is also severe enough to abolish FOXC1 DNA binding. 

Both the L86F and the L86P mutations severely impair the transactivational ability of FOXC1. In the case of L86F, although the mutation still has appreciable DNA binding capacity, the mutation severely perturbs the ability of FOXC1 to drive transcription of the luciferase reporter, possibly due to failed interactions with the transcriptional machinery. These data indicate that position 86 has an intermolecular role, as evidenced by the diminished activity of FOXC1 L86F. As for FOXC1 L86P, this mutant shows a serious disruption in nuclear localization and did not appear to bind the FOXC1 binding site in EMSAs. Although the disruption to transactivation is severe in comparison to wild-type FOXC1 transactivation, there is still residual transactivation above that in the ΔFHD and L86F FOXC1 proteins. It is possible that the EMSAs, which showed an absence of L86P binding to the FOXC1 binding site, may not be sensitive enough to detect the small amount of binding that is sufficient for transactivation above the baseline. It is also possible that some binding of FOXC1 L86P occurs when the FOXC1 binding site is presented in the context of the reporter plasmid and not on the smaller FOXC1 binding site oligonucleotide used in the EMSAs. Nevertheless, the severe disruption of the transactivation ability of FOXC1 L86F and L86P illustrates the importance of this residue to FOXC1 function. These data support the idea that while position 86 has an intermolecular function, as evidenced by the transactivation defect of FOXC1, helix 1 overall plays a strong intramolecular role, demonstrated by the severe DNA binding defects seen with FOXC1 L86P. 

The clinical features of the patient carrying the L86F FOXC1 mutation provide further evidence for FOXC1’s being an important regulator of mesodermal development. Experiments in murine models have shown that deletions of Foxc1 can cause severe ocular phenotypes ^{11 15 20 21 22 23 24} including iris hypoplasia and maldevelopment of the iridocorneal angle. ²⁴ Findings of ocular anomalies in our patient, in association with the L86F FOXC1 mutation, is a strong demonstration of mouse models that recapitulate human defects. Murine models have also demonstrated that Foxc1 plays an important role in cardiac development. ^{6 15} Our patient had a myocardial infarction. Although some other patients with AR malformation have been reported to have cardiac defects, ^{5 6 9 15 16} such clinical cardiac findings are infrequently present in them. The cardiac anomalies in our patient may simply be a comorbidity or may reflect still poorly understood stochastic and genetic background influences on FOXC1-associated human disease. 

The analysis of the naturally occurring FOXC1 L86F disease-causing missense mutation again demonstrates that there is a lack of strong genotype–phenotype correlations in FOXC1 pathogenesis. Within individuals with AR malformation due to FOXC1 mutations, it is not possible to predict the severity of the disease, or whether the disease will manifest in anomalous or syndromic form, from the degree of molecular deficiency in FOXC1 function. The available evidence implies that FOXC1 activity is strictly regulated and has a tight threshold with both upper and lower limits. FOXC1 duplications cause AR-like malformations ^{7 25 26} as do mutations that cause functional nulls of FOXC1. ^{7 8 9 10 18}  

AR malformations can also result from mutations in the homeobox gene PITX2 on the long arm of human chromosome 4, region 25. ²⁷ However, in a manner that is quite different from FOXC1 mutations, mutations in PITX2 have strong genotype–phenotype correlations. ^{28 29} It appears that the greater the disruption to PITX2 activity, the more severe the phenotype of the disease. Mutations in PITX2 follow a progressive model where severity of AR phenotypes is directly correlated to diminished protein activity. In contrast, mutations in FOXC1 appear to follow a threshold model where normal development is reliant on wild-type levels of FOXC1 activity, with any reduction or increase in activity leading to maldevelopment. Although the pathways and/or mechanisms disrupted by FOXC1 or PITX2 mutations are clearly regulated differently, both result in AR malformations. Thus AR malformations may represent a point of convergence of different pathways and mechanisms in eye development. 

 Supported by Summer Studentship Award (TCM). MAW is an Alberta Heritage Fund for Medical Research Senior Scholar and a Canadian Institutes of Health Research (CIHR) investigator.

 Submitted for publication January 28, 2003; revised June 19, 2003; accepted June 30, 2003.

 Disclosure: R.A. Saleem, None; T.C. Murphy, None; J.M. Liebmann, None; M.A. Walter, None

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

 

Figure 1.

 

Mutation identification in FOXC1. The genomic DNA sequence of a wild-type control (left) and a patient with the FOXC1 mutation (right) is shown. The G→T transversion results in a silent mutation, whereas the C→T transition results in a leucine-to-phenylalanine missense mutation.

Figure 2.

 

Immunofluorescence of recombinant FOXC1 constructs in COS-7 cells and the percentage of exclusively nuclear localization. Whereas FOXC1 and FOXC1 L86F localize to the nucleus, nuclear localization of FOXC1 L86P is severely disrupted.

Figure 3.

 

L86F and L86P mutations disrupt FOXC1 function. (A) Western blot of Xpress epitope-tagged FOXC1, L86F, and L86P detected with an anti-Xpress antibody. (B) L86F binds the FOXC1 DNA-binding site, whereas FOXC1 DNA binding is disrupted by the L86P mutation. (C) Transactivation of a luciferase reporter construct by FOXC1 is impaired by both the L86F and the L86P mutation. At the bottom is a schematic of the reporter construct with six copies of the FOXC1 DNA binding site located upstream of the thymidine kinase (TK) promoter, driving expression of the luciferase gene. Error bars, SE.

The authors thank Farideh Mirzayans, Jody Marshall for technical assistance, and Robert Ritch for assistance with the patients’ reports.