Essential structural and functional determinants within the forkhead d
http://www.100md.com
《核酸研究医学期刊》
1 Department of Medical Genetics and 2 Department of Ophthalmology, 832 Medical Sciences Building, University of Alberta, Edmonton, Alberta, Canada T6G 2H7, USA and 3 Genome Technology Branch, National Human Genome Research Institute, National Institutes of Health, Bethesda, MD, USA
* To whom correspondence should be addressed. Tel: +1 780 492 9805; Fax: +1 780 492 6934; Email: mwalter@ualberta.ca
Present address: R. A. Saleem, The Institute for Systems Biology, Seattle, WA, USA
ABSTRACT
The forkhead domain (FHD)-containing developmental transcription factor FOXC1 is mutated in patients presenting with Axenfeld–Rieger malformations. In this paper, we report the introduction of positive, negative or neutral charged amino acids into critical positions within the forkhead domain of FOXC1 in an effort to better understand the essential structural and functional determinants within the FHD. We found that FOXC1 is intolerant of mutations at I87. Additionally, alterations of amino acids within -helix 1 of the FOXC1 FHD affected both nuclear localization and transactivation. Amino acids within -helix 3 were also found to be necessary for transactivation and can have roles in correct localization. Interestingly, changing amino acids within -helix 3, particularly R127, resulted in altered DNA-binding specificity and granted FOXC1 the ability to bind to a novel DNA sequence. Given the limited topological variation of FHDs, due to the high conservation of residues, we anticipate that models of forkhead domain function derived from these data will be relevant to other members of the FOX family of transcription factors.
INTRODUCTION
Autosomal dominant mutations in the FOX family transcription factor FOXC1 underlie Axenfeld–Rieger (AR) malformations mapping to human chromosome 6p25 (1,2), thought to arise because of defects in neural crest cell migration and differentiation (3) . Patients with AR malformations present with ocular features that include iris hypoplasia, iridocorneal adhesions, posterior embryotoxon and corectopia. The most severe ocular consequence of AR malformations is the development of glaucoma, a progressively blinding condition, in 50% of the AR patients. Non-ocular features of AR malformations include dental and umbilical anomalies. Cardiac anomalies have been observed to infrequently cosegregate with AR malformations (6–8).
Work in animal models has demonstrated the importance of FOXC1 as a developmentally key transcription factor. Recombinant Foxc1 null mice die peri- or post-natally with massive skeletal, cardiac, ocular and urogenital anomalies (3,8–11). Heterozygous Foxc1+/– mice are viable but recapitulate the spectrum of ocular defects seen within humans carrying FOXC1 mutations (12). Some of these ocular anterior segment defects in heterozygous Foxc1+/– mice can be modified by Tyrosinase, a member of a family of genes involved in melanin biosynthesis (13). FOXC1 is also involved in TGF-? signaling (3,14) and is required for the transcription of another important regulator of developmental processes, the T-box gene Tbx1 (15). The activation of Tbx1 can be induced by Shh, with induction mediated by FOXC proteins (15). Together with a highly related protein, FOXC2 (Figure 1), FOXC1 is thought to act as an important regulator of somitogenesis (15–17).
Figure 1. Sequence alignment of FHDs of FOXC1 and FOXC2. The sequences shown in single letter amino acid code are human FOXC1 (Q12948 ) and FOXC2 (Q99958 ). The mismatches in the sequence between FOXC1 and FOXC2 are shown in blue. The disease-causing missense mutations of FOXC1 analyzed in this study are shown in red. The alanine scanning and charge-altered mutations introduced in FOXC1 are shown above the sequence at their respective amino acid positions. The location of the -helices defined in the solution structure of human FOXC2 (FREAC-11, pdb|1D5V) are schematically represented in the bar below the alignment.
FOXC1 contains the forkhead domain (FHD), a highly conserved sequence of approximately 110 amino acids. The three-dimensional topology of the FHD was first resolved using X-ray crystallography on Foxa3 (formerly HNF3) bound to a DNA target (18) and has since been studied using NMR analysis of FOXC2, Foxd3 and FOXO4 (19–21). This DNA-binding motif is a variant of the helix–turn–helix motif, consisting of three -helices, two ?-sheets and two large loops that form ‘wing-like’ structures (18–21). This region functions as a DNA-binding domain and contains the sequences required for nuclear localization of FOXC1 (22–24).
In vitro site selection experiments were used to determine that FOXC1 bound a nine-base-pair core sequence 5'-GTAAATAAA-3' with high affinity (25). These experiments also found that when the FOXC1 FHD binds, it bends the DNA 94°, while in the context of full length FOXC1, the bound DNA is bent 112° (26). Experiments analyzing FOX binding sites upstream of the Tbx1 gene found that FOXC1 is also able to bind the DNA sequence 5'-AAAACAAACAGGC-3' in EMSAs (15).
Outside the FHD, FOXC1 contains N- and C-terminal transactivation domains, allowing FOXC1 to act as a transcriptional activator, as well as a transcriptional inhibitory domain C-terminal to the FHD (22). FOXC1 is phosphorylated, probably in the inhibitory domain region, but the functional significance of FOXC1 phosphorylation is undetermined (22).
All of the disease-causing missense mutations within FOXC1 identified to date in patients with AR malformations are located within the FHD (Table 1). Analyses of FOXC1 missense mutations characterized through both molecular modeling and biochemical means revealed several different modes of impairment to FOXC1 function (24,26,27). These impairments include defects in localization of FOXC1 to the nucleus, defects in DNA-binding capacity, defects to transactivation and alterations to DNA-binding specificity. One mutation (I87M) was found to reduce levels of FOXC1 either by destabilizing FOXC1 or by reducing efficiency of the translation of FOXC1 mRNA. These molecular defects do not appear to be dependent upon the position of the mutations within the FHD.
Table 1. Molecular defects caused by missense mutations in FOXC1
To deepen our understanding of FHD function, we have created a molecular model of the FOXC1 FHD to predict residues that are critical for function and structure of FOXC1. Critical amino acid positions within the FOXC1 FHD were chosen and converted to an alanine (neutral charge), glutamic acid (negative) or lysine (positive charged) residue. Our model of the FOXC1 FHD was used to predict the effect that these conversions would have on the structure of the FHD. The effect of these conversions on FOXC1 function was then determined by cell biology and biochemical analyses. Our results have revealed the locations of residues that are critical for the translocation of FOXC1 to the nucleus, that guide the interactions of FOXC1 with DNA and organize the FHD in such a manner that FOXC1 is stable, competent to bind DNA and able to activate transcription. We anticipate that this information will also allow prediction of the effects on structure and function of alterations of FHD amino acids of other members belonging to the FOX family of transcriptional regulators.
MATERIALS AND METHODS
Plasmid construction and mutagenesis
The FOXC1 pcDNA His/Max4 protein expression construct has been previously described (24). For mutagenesis reactions, a fragment of FOXC1 from 106 to 777 was PCR amplified and cloned into pGEM T Easy (Promega) using primers 5'-ggc tac acc gcc atg c-3' (forward) and 5'-gct ctc gat ctt ggg cac-3' (reverse). This was then subjected to mutagenesis using the QuickchangeTM (Stratagene) mutagenesis kit according to the manufacturer's protocol and primers summarized in Table 2, with the addition of 5% dimethylsulfoxide. The primers used for mutagenesis contained degenerate oligonucleotides at the site of interest, so that all the three desired mutagenesis products could be obtained from a single reaction. Potential mutant constructs were sequenced and confirmed mutant fragments subcloned back into the FOXC1 pcDNA His/Max4 construct. The final constructs were sequenced again to ensure the integrity of the construct.
Table 2. Primers used for degenerate site directed mutagenesis
Cell culture and transfection
COS-7 and HeLa cells were grown in Dulbecco's modified Eagles medium and 10% fetal calf serum. Cells were transfected using Fugene6TM transfection reagent (Roche) according to the manufacturer's protocol. For protein expression, 2.5 μg of plasmid DNA and 20 μl of Fugene6 were used for a 100 mm transfection plate. For luciferase assays, HeLa cells were transfected in 6-well plates using 500 ng of FOXC1 pcDNA His/Max4, 50 ng of the pGL3-TK-FOXC1 binding site construct, 1 ng of the pRL construct and 3 μl of Fugene6 transfection reagent. COS-7 cells were grown and transfected directly on coverslips when used for immunofluorescence, using 1 μg of FOXC1 pcDNAHis/Max4 and 3 μl of Fugene6 transfection reagent.
Protein expression
Protein extraction was performed 48 h after transfection by washing the cells in phosphate buffered saline (PBS), collecting the cells by scraping and lysing the cells by sonication in the presence of lysis buffer (Promega). After sonicating, the lysates were centrifuged at 13 000 g for 5 min at 4°C. Proteins were resolved by SDS–PAGE, and western-blot analysis was performed using a mouse monoclonal anti-Xpress antibody (Invitrogen) against the pcDNA His/Max4 vector encoded, N-terminal Xpress tag.
Immunofluorescence
All immunofluorescence procedures were carried out at room temperature. After transfection for 24 h, cells were washed with PBS, fixed using 2% paraformaldehyde/PBS for 10 min and washed again in PBS. Cells were permeabilized using 0.05% Triton X-100/PBS for 10 min. After washing, a 5% BSA/PBS solution was applied and the cells were blocked for 1 h. The cells were then washed with a 1% BSA/PBS solution and incubated for 1 h with a 1:400 solution of the anti-Xpress antibody in 1% BSA/PBS. Following incubation, the cells were washed with the 1% BSA/PBS solution and incubated for 1 h with a 1:400 solution of an anti-mouse Cy3 conjugated secondary antibody (Jackson Immunolaboratories) in 1% BSA/PBS. The cells were then washed with 1% BSA/PBS, mounted using mounting media containing 4',6-diamidine 2-phenylindole (DAPI) and the coverslips were sealed using nail polish.
For each construct, cells were scored as showing either nuclear, cytoplasmic, or both nuclear and cytoplasmic staining. At least 100 cells were scored for each construct that was tested.
Northern-blot analysis
About 48 h after transfection, cells were washed with PBS and the RNA was extracted through the use of TRIzol (Gibco/BRL) reagent. The RNA was size separated and loading amounts were equalized on a 1x 3-(N-morpholino) propane sulfonic acid, pH 7.0, 0.66 M formaldehyde agarose gel and transferred onto a Hybond (Amersham) nylon membrane. The membrane was then probed with a 32P-labeled PCR product from the 5' end of FOXC1 (20–233 nt).
Electrophoretic mobility shift assays
Cell extracts containing recombinant FOXC1 protein were equalized for the amounts of recombinant FOXC1 by immunoblot detection. Cell extracts were incubated with 0.45 mM DTT, 0.09 μg poly(dI-dC), 0.225 μg sheared salmon sperm DNA, 13.5% glycerol and 20 000 counts per minute of 32P-labeled oligonucleotides (Table 3) at room temperature for 15 min. After pre-running 6% Tris–glycine–EDTA gels, the electrophoretic mobility shift assay (EMSA) reactions were subjected to electrophoresis.
Table 3. The FOXC1 binding site and variant oligonucleotides
Dual-luciferase assays
The luciferase reporter construct, based on the pGL3 vector with a thymidine kinase promoter and six copies of FOXC1 binding site (Table 3), was described previously (24). Transfected cells (described above) were grown for 48 h and luciferase assays performed using the Dual Luciferase Assay Kit (Promega) according to the manufacturer's protocol. The reactions were replicated a minimum of three times.
Molecular modeling
Molecular models for the FHD of FOXC1 were generated using MODELLER (28) based on the NMR structure of FOXC2 (pdb|1D5V) (20).
RESULTS
FOXC1 was mutated independently to an alanine (A), a glutamate (E) or a lysine (K) residue at positions P79, L86, I87, I91, I126 and R127 by site-directed mutagenesis (Figure 1). The rationale for choosing these positions was based upon both molecular modeling and threading analyses indicating that these changes probably perturb FHD structure, as well as empirical observations from the studies of naturally occurring missense mutations at these positions (24,26,27).
Protein expression and western-blot analysis
Expression and western-blot analysis of the recombinant proteins revealed that all the plasmid constructs expressed recombinant FOXC1 with the exception of FOXC1 mutated at position 87 (Supplemental Figure 1). When I87 was converted to an A, E or K, the FOXC1 protein could not be detected by western-blot analysis (Figure 2B). Northern-blot analysis was carried out to ensure that mRNA was, in fact, being expressed from the transfected recombinant FOXC1 plasmid (Figure 2A). FOXC1 87A/E/K mRNA was detected by northern-blot analysis, but FOXC1 87A/E/K protein could not be detected by western-blot analysis. This finding is consistent with the notion that alterations of I87 drastically affect the cellular levels of soluble recombinant FOXC1 protein.
Figure 2. Conversion of I87 to an alanine, glutamate or lysine residue reduces protein levels of FOXC1. (A) Northern-blot analysis showing expression of FOXC1 and FOXC1 87 A/E/K mRNA. Blot is probed with an N terminal portion of the FOXC1 gene (see Materials and Methods). Loading amounts were equalized on a 1x 3-(N-morpholino) propane sulfonic acid, pH 7.0, 0.66 M formaldehyde agarose gel (see Materials and Methods).
Subcellular localization of the FOXC1 charge conversion proteins
Previous work has shown that missense mutations within the FHD are able to disrupt the normal nuclear localization of FOXC1 in COS-7 cells (26). Based on this observation, the subcellular localization of the FOXC1 charge conversion proteins was tested. By immunofluorescence, the subcellular localization of the FOXC1 mutant proteins was found to be highly variable and dependent upon both the position of the amino acid change and the charge of the amino acid change (Figure 3 and Table 4). With the exception of amino acid residue R127, conversion of the amino acid into the neutral alanine amino acid had the least severe disruption on nuclear localization of FOXC1. Nuclear localization was least perturbed by the presence of the neutral charged P79A (91% nuclear), with the charged amino acids P79E and P79K being more disruptive to FOXC1 localization, at 80 and 75% nuclear respectively. At position L86, the change to the uncharged alanine disrupts the nuclear localization of FOXC1 to a degree similar to what is seen with P79K. Again, when charged amino acids are substituted for leucine at position 86, subcellular localization becomes more perturbed (L86E at 51% nuclear, L86K at 64% nuclear) than what is seen with alanine substitution (72% nuclear). The localization of FOXC1 that is observed with mutations at amino acid positions 79 and 86 is predominantly nuclear within individual cells, but is not confined as strictly to the nucleus compared to cells expressing wild-type FOXC1. When charged amino acids are substituted for the isoleucine at position 91, the disruption becomes severe, with 25% of the population showing nuclear localization for I91E and only 5% of the cellular population showing nuclear staining when FOXC1 carries the I91K mutation. I91A shows levels of nuclear localization that are similar to levels seen with L86E. The differences to nuclear localization upon substitution of differently charged amino acids is most pronounced in I126 and R127, both of which are located in -helix 3 of the FHD. The substitution of charged residues at I126 prevents the strict nuclear localization of FOXC1 in any of the cells in the population, leading instead to mixed nuclear and cytoplasmic staining as well as only cytoplasmic staining. The mislocalization of FOXC1 I126E and I126K appears to be different from the mislocalization when the other amino acids are altered. As can be seen in Figure 3, a major part of the immunofluorescent signal given by these mutations appears to localize to a perinuclear space, within what is likely the endoplasmic reticulum, a site of protein degradation. Conversely, FOXC1 carrying the I126A alteration shows much stronger nuclear localization, where levels are similar to what is seen with P79K or L86A. Finally, FOXC1 R127A, at 56% nuclear localization, does have a nuclear localization defect, but this is much less severe than the disruption seen for FOXC1 R127E, in which mislocalization is complete. Interestingly, nuclear localization of FOXC1 R127K approaches wild-type levels (88%) with little defect in localization seen.
Figure 3. A, E and K conversions in the FOXC1 FHD differentially disrupt the nuclear localization of FOXC1. COS-7 cells were transiently transfected with recombinant XpressTM epitope-tagged constructs. Cy3 fluorescence indicates the position of the Xpress epitope-tagged FOXC1 proteins within the cell. DAPI staining indicates the position of the nuclei.
Table 4. Summary of subcellular localization of wild-type and converted FOXC1 molecules
Charge conversion mutations differentially perturb the DNA-binding capacity of FOXC1
As the FHD is the domain through which FOXC1–DNA interactions are mediated, it was of interest to test how the altering of the amino acid and charge at different positions affects the ability of FOXC1 to bind an in vitro derived FOXC1 binding site (Table 3) (24,25). As seen in Figure 4, FOXC1 with alterations to amino acids P79, L86 or I91 is still able to bind the FOXC1 binding site, although in some cases this binding capacity is reduced. The binding capacity of FOXC1 P79A and P79K is at wild-type levels or near wild-type levels, while P79E appears to bind with a 3- to 4-fold reduction in affinity for the FOXC1 binding site (Figure 4A). The binding capacity of FOXC1 L86A, E or K is at wild-type FOXC1 levels or near wild-type FOXC1 levels (Figure 4B). Binding capacity of FOXC1 with alterations at I91 is reduced in the cases of I91A and I91K with 3- to 4-fold reductions in affinity for the FOXC1 binding site (Figure 4C). FOXC1 I91E binds the FOXC1 binding site with wild-type FOXC1 affinity.
Figure 4. FOXC1 A, E and K mutations disrupt the DNA-binding capacity of FOXC1 to different extents depending on position and charge. Equalized recombinant FOXC1 containing cell extracts were incubated with 32P-labeled FOXC1 binding site (see Table 3) and resolved by native PAGE. Position of the predominant FOXC1–DNA complex is indicated with the filled gray arrowhead.
The deficiencies in FOXC1 binding are more pronounced when A, E or K are introduced at positions I126 or R127. Binding capacity of FOXC1 is completely disrupted by the presence of a lysine or glutamic acid at 126 (Figure 4D). When an alanine is present, binding capacity is retained; however, there is an approximate 4-fold increase in the amount of FOXC1 I126A protein required for wild-type levels of DNA binding. Two of the R127 changes, R127A and R127K, show residual binding capacity, but the levels of this capacity are reduced to well below 10x wild-type FOXC1 levels (Figure 4E). The R127E mutation shows no capacity for DNA binding by this assay.
Amino acids within -helix 3 of the FHD regulate FOXC1 DNA-binding specificity
It has been shown previously that mutations within -helix 3 are able to alter the binding site specificity of FOXC1 (24). It is also known that the specificity of many protein–DNA interactions rely upon the interactions of charged amino acid side chains such as arginines, lysines and glutamic acids, and the charges inherent in the DNA (29). We therefore utilized a series of FOXC1 binding site variants (Table 3) to test whether, in fact, any of the mutations tested in this study were able to alter the binding site specificity of FOXC1. All the altered FOXC1 molecules appeared to bind the variant oligonucleotides in a pattern similar to that seen for FOXC1 (data not shown), with the exception of R127A and R127K (Figure 5) which display altered binding site specificity in comparison with wild-type FOXC1. R127A has a reduced binding affinity for variant oligonucleotides 1 and 2 compared to wild-type FOXC1 affinity for these oligonucleotides (Figure 5A). Interestingly, while wild-type FOXC1 is unable to bind oligonucleotide 7, the R127A FOXC1 recombinant protein is able to weakly bind this variant oligonucleotide. In a manner that is strikingly different from wild-type FOXC1 or any of the other mutant FOXC1 proteins tested in this and other studies, R127K has an affinity for variant oligonucleotide 7 that is stronger than the affinity of FOXC1 R127K for the FOXC1 binding site (Figure 5B). These data indicate that helix 3 of the FHD in FOXC1 may play a strong role in regulating the specificity of FOXC1–DNA interactions.
Figure 5. Conversion of R127 into an alanine or lysine residue alters the binding specificity of FOXC1. FOXC1 containing cell extracts were incubated with 32P-labeled FOXC1 binding site or variant oligonucleotides (see Table 3) and resolved by native PAGE. Position of the predominant FOXC1–DNA complex is indicated with the filled gray arrowhead, an asterisk indicates oligonucleotides that the FOXC1 constructs bind with altered affinity. FOXC1 127A no longer binds oligonucleotides 1 and 2 while binding oligonucleotide 7. FOXC1 127K binds oligonucleotide 7 with a greater affinity than its affinity for the FOXC1 binding site.
Transactivation assays
FOXC1 is able to potently drive expression of a reporter construct when the FOXC1 binding site is present in front of the construct promoter (see Figure 6 inset). The effect of the different charges at the different positions on the ability of FOXC1 to drive transactivation was tested. The majority of constructs tested had severely reduced transactivation activity, including those FOXC1 mutations that retained DNA-binding activity (FOXC1 79A/E, 86A/E/K, 91A/E/K, 126A). Notable exceptions were FOXC1 P79K, R127A and R127K. FOXC1 P79K retained an average transactivation capacity of 66% of wild-type activity, while FOXC1 R127A and R127K had transactivation capacities of 36 and 38% of wild-type FOXC1 activity, respectively.
Figure 6. Disruption of transactivation of a luciferase reporter construct by FOXC1 A/E/K proteins. The thick black bars represent the mean of the values; the Y error bar indicates the standard deviation of the values. The inset shows a schematic of the FOXC1 binding site, thymidine kinase promoter, luciferase gene reporter construct.
DISCUSSION
These experiments provide an in-depth analysis of critical positions within the FHD delineating the nature of essential structural and functional determinants within the FOXC1 FHD (Table 5). This study represents the largest analysis of the FHD domain undertaken for any FOX class protein to date. Since the FHD is a highly conserved domain, with limited three-dimensional topological variations, this study should facilitate the modeling of missense mutations within the FHD for all the members of the FOX family of transcription factors.
Table 5. Summary of the molecular consequences of conversion of these critical amino acid positions into an alanine, glutamic acid or lysine residue
Wild-type FOXC1 localizes completely to the nucleus in almost all cells within a given cell population (24,26,27). To date, no mechanism has been identified whereby FOXC1 is differentially localized in response to environmental stimuli or post-translational modifications, such as the insulin-stimulated phosphorylation of FOXO proteins by PKB and subsequent redistribution of FOXO proteins to the cytosol . When analyzing these mutations of FOXC1, mixed localization of FOXC1 to both the nucleus and the cytoplasm of the cell is often seen. This incomplete localization could be due to inefficient transport of defective FOXC1 molecules into the nucleus, or, alternatively, may reflect a defect in the ability of FOXC1 to be retained within the nucleus.
P79 is not involved in the formation of the FOXC1–DNA complex
Alterations at position 79 located N-terminal to -helix 1 cause the mildest disruptions of the positions tested herein, to the nuclear localization of FOXC1 (Table 4). P79 is oriented in such a way that the amino acid residue extends away from the DNA (Figure 7A) and thus is unlikely to be critical for formation of the FOXC1–DNA complex. These data would indicate that the FHD must be largely intact and the native structure preserved, as none of the alterations at position 79 appeared to alter either the affinity of FOXC1 for the FOXC1 binding site or the specificity of FOXC1 affinity for the FOXC1 binding site. Changes from a proline at amino acid 79, however, do appear to change the overall structure of FOXC1 sufficiently so that problems with not only the distribution of FOXC1, but also its transactivation do occur. It is interesting that while FOXC1 P79K shows more of a defect in FOXC1 subcellular localization (25% mislocalization; Table 4) compared with P79A/E it is the only one of the three that retains the capacity for transactivation, even at 66% of wild-type levels (Figure 6). Clearly, while the conformation of FOXC1 P79K is altered suffiently as to result in a defect in the way this mutant protein interacts with the nuclear transport machinery, the molecule is intact enough to both bind the FOXC1 binding site and participate in transcription from a reporter construct.
Figure 7. Molecular model of FOXC1 forkhead domain. (A) A ribbon model showing the positions of the disease-causing missense mutations in the FHD of FOXC1. (B) A wire diagram summary of the functional sub-domains of the FOXC1 FHD. Regions 1 and 2: the N-terminal portion of the FHD and -helix 1 function in the organization of the FHD for transactivation, DNA binding and nuclear localization. Region 3: -helix 3 is involved in DNA-binding specificity of the FHD and also plays a role in the organization of the FHD with respect to nuclear localization and high efficiency binding. Region 4: the wing 2 region functions in the organization of the FHD for transactivation and DNA binding.
With respect to the P79K mutation, and, in fact, all the FOXC1 mutant proteins that retain transactivation capabilities, it is unknown whether the ability to bind or transactivate in high-order chromatin structures is affected.
Residues in -helix 1 are required for the transcriptionally competency of FOXC1
The mutations located within -helix 1 can have mild (86 A/E/K and 91 A/E/K) or more severe (87 A/E/K) effects. Previous work has shown that a naturally occurring I87M missense mutation is able to reduce the levels of FOXC1 protein produced in COS-7 cells (24), similar to what is seen in this study with the I87 A/E/K mutations. These data imply that I87 is critical in the formation of a stable FHD and FOXC1 molecule with alterations of the isoleucine at position 87 reducing the levels of soluble recombinant FOXC1 protein.
The FOXC1 86 A/E/K and 91 A/E/K mutations impede the exclusive localization of FOXC1 to the nucleus (Figure 3 and Table 4). The localization defects to FOXC1 91 A/E/K are interesting for two reasons. Previous characterizations of naturally occurring FOXC1 L86F, I91S, I91T mutations indicate that mutations to position 91 severely impair nuclear localization, while the L86F mutation shows wild-type nuclear localization of FOXC1 (26,27). Previous work has also defined a nuclear localization accessory signal that lies within amino acid residues 78–93, encompassing the I91 position (22). Molecular modeling predicts that I91 is oriented inwards towards the hydrophobic core of the FHD (Figure 7A). Taken together, these data indicate that I91 plays a critical role in organizing the FHD with respect to interactions between the nuclear transport machinery and FOXC1.
All six of these helix 1 mutations severely disrupt FOXC1 transactivation but retain the ability to bind the FOXC1 binding site, though DNA binding is reduced with some of the mutations. Thus, while the FHD is grossly intact, these helix 1 mutations must have effects that extend beyond the FHD, either increasing the potency of the inhibitory domain C-terminal to the FHD and/or preventing function of both the N-terminal and C-terminal transactivation domains of FOXC1 (22). Amino acids L86 and I91 therefore appear to be critical in organizing the structure of the FHD in such a manner that the FOXC1 molecule as a whole is transcriptionally competent.
I126 is critical for correct nuclear localization and transactivation of FOXC1
Mutations in helix 3 appear to have drastic effects on the function of FOXC1 with respect to localization, transactivation and/or DNA-binding specificity. Conversion of I126 to a neutrally charged alanine residue perturbs both the subcellular distribution of FOXC1 and DNA-binding capacity of FOXC1 to some extent, but the FHD domain retains enough wild-type structure to facilitate FOXC1 localization and DNA binding. Molecular modeling of the FHD predicts that I126 is an important residue in the formation of a hydrophobic core bounded by the three primary -helices (24). The substitution to an alanine is likely to preserve the hydrophobicity of the core, allowing the basic structure of the FHD to remain intact, while the charged substitutions (FOXC1 126E and 126K) would introduce hydrophilic amino acid residues into the hydrophobic core, disrupting the FOXC1–DNA interactions. All three of the FOXC1 126 constructs fail to transactivate gene expression (Figure 6). These data imply that although FOXC1 I126A retains enough native structure to allow both FOXC1 interactions with the nuclear transport machinery and FOXC1–DNA interactions, the overall structure of FOXC1 is altered to an extent that perturbs the transactivation domains that lie at the N and C terminals of the FOXC1 protein. It is also possible that the change in conformation enhances the inhibitory effects of the inhibitory domain rather than diminishing the function of the transactivation domains.
R127 is a critical amino acid with respect to the FOXC1–DNA-binding specificity
The R127 position is a highly conserved position in FOX proteins and again, conservation of the amino acid charge causes the least disruption of FOXC1 localization. Interestingly, both FOXC1 R127A and K, while still binding the FOXC1 binding site with reduced affinity, were found to have alterations to binding-site specificity compared to both wild-type FOXC1 and each other (Figure 5). The specificity of binding of FOXC1 R127A appears to become less stringent with binding of variant oligonucleotides 6–9, while conversely, the binding specificity of R127K appears to be altered so that this FOXC1 molecule now has a higher affinity for variant oligonucleotide 7 (gtaaattaa) than for the FOXC1 binding site (gtaaataaa) (Figure 5).
Resolution of the crystal structure of FOXA2 (18) and NMR of the solution structure of FOXC2 (20), the latter containing a FHD 97% identical to that of FOXC1, indicates that helix 3 lies across the DNA in the major groove, possibly functioning as a recognition helix. Previous work has shown that an I126M missense mutation is able to alter the binding specificity of FOXC1 (24), though not to the same extent as what is seen with 127K. The crystal and NMR resolved structures show that the R127 side chain extends away from the amphipathic helix 3 and is likely to be a contact point between FOXC1 and DNA. A molecular model, based on threading experiments in which FOXC1 was threaded through FOXC2 coordinates, shows the R127 residue extending away from -helix 3, into the major groove of the DNA (Figure 7A). Thus, the positively charged 127K would attract the negatively charged DNA, whereas 127A would neither attract nor repulse. Hence, both these FOXC1 constructs would still be able to bind. The presence of the negative charge at this contact point would create repulsion between FOXC1 R127E and the DNA.
In spite of showing a severe reduction in binding to the FOXC1 binding site, both FOXC1 R127A and R127K are able to drive expression of a reporter construct, at levels that are 36% and 38% of wild-type FOXC1 levels, respectively (Figure 6). It would appear that, based on the levels of transcription relative to the levels of binding, the transactivation domains are properly organized in FOXC1 R127A/K molecules. These data indicate that the key role of R127 is to function in the proper recognition and binding of the FOXC1 binding site, as well as indicating that R127 does not function to organize the FHD in such a manner as to make FOXC1 transcriptionally competent.
It remains a formal possibility that the FOXC1 mutants that are able to bind DNA by EMSA, but show a transactivation defect in in vivo luciferase assays, may have an in vivo defect in DNA binding that underlies the transactivation defect. However, data from the FOXC1 R127A and R127K mutants are not consistent with this possibility. These two mutants retain sufficient DNA binding to facilitate transactivation, indicating that EMSA experiments may be more sensitive to defects in DNA binding than the transactivation assays. Further investigation of the consequences of FOXC1 mutations upon FOXC1–DNA binding and transactivation functions will necessitate utilization of bona fide FOXC1 target genes in the context of native chromatin. Such investigations are currently underway.
A molecular model of the FHD of FOXC1
The FHD is highly conserved, and structural studies have shown limited three-dimensional variation between different FOX FHDs which implies limited variation in modes of DNA recognition. The structural basis of differences in DNA sequence recognition remains undetermined; however, there is evidence that the third helix, which is positioned in the major groove of DNA (31), and the second wing (25) guide protein–DNA interactions. Rather than alterations in the gross topology of the FHD being the basis of DNA recognition, it is thought that differences in charge along the interface between the FHD and DNA provide the basis for sequence specificity (21). Differences in charge along the interface between the FOXC1 binding site and helix-3 of the FOXC1 FHD are likely to underlie the alterations in DNA-binding specificities that are seen in FOXC1 127A and 127K mutant proteins.
A ribbon model showing the positions of the disease-causing missense mutations in the FHD of FOXC1 is shown in Figure 7A. A wire diagram summary of the functional sub-domains of the FOXC1 FHD is shown in Figure 7B. As amino acids predicted to be critical for FOXC1 structure on the basis of this model were actually found to be key determinants of FOXC1 structure and function by our subsequent biochemical analyses, we have confidence that this FOXC1 model has further predictive value.
Finally, this study illustrates the success in merging predictive bioinformatic analyses with experimental cell biology to achieve testable models of protein structure. These studies demonstrate the functional importance of not only the individual amino acids within the FHD of FOXC1 but also demonstrate how -helix 1 and -helix 3 secondary structures in the FHD are involved in FOXC1 function. Helix 1 appears to be involved in the organization of the FHD in such a manner that FOXC1 can be translocated to or retained within the nucleus and is transcriptionally active. Helix 3 appears to also be involved in organization of the FHD so that FOXC1 nuclear translocation or retention occurs and FOXC1 is transcriptionally competent; it also plays a major role in controlling FOXC1–DNA interactions.
Given the high degree of conservation between FHD domains within the FOX family, with respect to the three dimensional topology (18–21), this combination of biological and computational analysis is a framework which will allow for predictive modeling of changes within the FHDs not only for members of the C class of FOX proteins but also for FHDs of other FOX classes as well.
SUPPLEMENTARY MATERIAL
ACKNOWLEDGEMENTS
We thank the Ocular Genetics Laboratory for critical reading of this manuscript. We thank May Yu for technical assistance. This work was supported by the Alberta Heritage Foundation for Medical Research (AHFMR) and the Canadian Institutes for Health Research (CIHR). R.A.S. is supported by a University of Alberta Dissertation Fellowship. M.A.W. is an AHFMR senior scholar and a CIHR investigator.
REFERENCES
Mears,A.J., Jordan,T., Mirzayans,F., Dubois,S., Kume,T., Parlee,M., Ritch,R., Koop,B., Kuo,W.L., Collins,C. et al. ( (1998) ) Mutations of the forkhead/winged-helix gene, FKHL7, in patients with Axenfeld–Rieger anomaly. Am. J. Hum. Genet., , 63, , 1316–1328.
Nishimura,D.Y., Swiderski,R.E., Alward,W.L., Searby,C.C., Patil,S.R., Bennet,S.R., Kanis,A.B., Gastier,J.M., Stone,E.M. and Sheffield,V.C. ( (1998) ) The forkhead transcription factor gene FKHL7 is responsible for glaucoma phenotypes which map to 6p25. Nature Genet., , 19, , 140–147.
Kume,T., Deng,K.Y., Winfrey,V., Gould,D.B., Walter,M.A. and Hogan,B.L. ( (1998) ) The forkhead/winged helix gene Mf1 is disrupted in the pleiotropic mouse mutation congenital hydrocephalus. Cell, , 93, , 985–996.
Lines,M.A., Kozlowski,K. and Walter,M.A. ( (2002) ) Molecular genetics of Axenfeld–Rieger malformations. Hum. Mol. Genet., , 11, , 1177–1184.
Alward,W.L. ( (2000) ) Axenfeld–Rieger syndrome in the age of molecular genetics. Am. J. Ophthalmol., , 130, , 107–115.
Swiderski,R.E., Reiter,R.S., Nishimura,D.Y., Alward,W.L., Kalenak,J.W., Searby,C.S., Stone,E.M., Sheffield,V.C. and Lin,J.J. ( (1999) ) Expression of the Mf1 gene in developing mouse hearts: implication in the development of human congenital heart defects. Dev. Dyn., , 216, , 16–27.
Suzuki,T., Takahashi,K., Kuwahara,S., Wada,Y., Abe,T. and Tamai,M. ( (2001) ) A novel (Pro79Thr) mutation in the FKHL7 gene in a Japanese family with Axenfeld–Rieger syndrome. Am. J. Ophthalmol., , 132, , 572–575.
Winnier,G.E., Kume,T., Deng,K., Rogers,R., Bundy,J., Raines,C., Walter,M.A., Hogan,B.L. and Conway,S.J. ( (1999) ) Roles for the winged helix transcription factors MF1 and MFH1 in cardiovascular development revealed by nonallelic noncomplementation of null alleles. Dev. Biol., , 213, , 418–431.
Kume,T., Deng,K. and Hogan,B.L. ( (2000) ) Minimal phenotype of mice homozygous for a null mutation in the forkhead/winged helix gene, Mf2. Mol. Cell Biol., , 20, , 1419–1425.
Hong,H.K., Lass,J.H. and Chakravarti,A. ( (1999) ) Pleiotropic skeletal and ocular phenotypes of the mouse mutation congenital hydrocephalus (ch/Mf1) arise from a winged helix/forkhead transcriptionfactor gene. Hum. Mol. Genet., , 8, , 625–637.
Kidson,S.H., Kume,T., Deng,K., Winfrey,V. and Hogan,B.L. ( (1999) ) The forkhead/winged-helix gene, Mf1, is necessary for the normal development of the cornea and formation of the anterior chamber in the mouse eye. Dev. Biol., , 211, , 306–322.
Smith,R.S., Zabaleta,A., Kume,T., Savinova,O.V., Kidson,S.H., Martin,J.E., Nishimura,D.Y., Alward,W.L., Hogan,B.L. and John,S.W. ( (2000) ) Haploinsufficiency of the transcription factors FOXC1 and FOXC2 results in aberrant ocular development. Hum. Mol. Genet., , 9, , 1021–1032.
Libby,R.T., Smith,R.S., Savinova,O.V., Zabaleta,A., Martin,J.E., Gonzalez,F.J. and John,S.W. ( (2003) ) Modification of ocular defects in mouse developmental glaucoma models by tyrosinase. Science, , 299, , 1578–1581.
Zhou,Y., Kato,H., Asanoma,K., Kondo,H., Arima,T., Kato,K., Matsuda,T. and Wake,N. ( (2002) ) Identification of FOXC1 as a TGF-beta1 responsive gene and its involvement in negative regulation of cell growth. Genomics, , 80, , 465–472.
Yamagishi,H., Maeda,J., Hu,T., McAnally,J., Conway,S.J., Kume,T., Meyers,E.N., Yamagishi,C. and Srivastava,D. ( (2003) ) Tbx1 is regulated by tissue-specific forkhead proteins through a common Sonic hedgehog-responsive enhancer. Genes Dev., , 17, , 269–281.
Topczewska,J.M., Topczewski,J., Shostak,A., Kume,T., Solnica-Krezel,L. and Hogan,B.L. ( (2001) ) The winged helix transcription factor Foxc1a is essential for somitogenesis in zebrafish. Genes Dev., , 15, , 2483–2493.
Kume,T., Jiang,H., Topczewska,J.M. and Hogan,B.L. ( (2001) ) The murine winged helix transcription factors, Foxc1 and Foxc2, are both required for cardiovascular development and somitogenesis. Genes Dev., , 15, , 2470–2482.
Clark,K.L., Halay,E.D., Lai,E. and Burley,S.K. ( (1993) ) Co-crystal structure of the HNF-3/fork head DNA-recognition motif resembles histone H5. Nature, , 364, , 412–420.
Jin,C., Marsden,I., Chen,X. and Liao,X. ( (1999) ) Dynamic DNA contacts observed in the NMR structure of winged helix protein–DNA complex. J. Mol. Biol., , 289, , 683–690.
van Dongen,M.J., Cederberg,A., Carlsson,P., Enerback,S. and Wikstrom,M. ( (2000) ) Solution structure and dynamics of the DNA-binding domain of the adipocyte-transcription factor FREAC-11. J. Mol. Biol., , 296, , 351–359.
Weigelt,J., Climent,I., Dahlman-Wright,K. and Wikstrom,M. ( (2001) ) Solution structure of the DNA binding domain of the human forkhead transcription factor AFX (FOXO4). Biochemistry, , 40, , 5861–5869.
Berry,F.B., Saleem,R.A. and Walter,M.A. ( (2002) ) FOXC1 transcriptional regulation is mediated by N- and C-terminal activation domains and contains a phosphorylated transcriptional inhibitory domain. J. Biol. Chem., , 277, , 10292–10297.
Pierrou,S., Hellqvist,M., Samuelsson,L., Enerback,S. and Carlsson,P. ( (1994) ) Cloning and characterization of seven human forkhead proteins: binding site specificity and DNA bending. EMBO J., , 13, , 5002–5012.
Saleem,R.A., Banerjee-Basu,S., Berry,F.B., Baxevanis,A.D. and Walter,M.A. ( (2001) ) Analyses of the effects that disease-causing missense mutations have on the structure and function of the winged-helix protein FOXC1. Am. J. Hum. Genet., , 68, , 627–641.
Pierrou,S., Enerback,S. and Carlsson,P. ( (1995) ) Selection of high-affinity binding sites for sequence-specific, DNA binding proteins from random sequence oligonucleotides. Anal. Biochem., , 229, , 99–105.
Saleem,R.A., Banerjee-Basu,S., Berry,F.B., Baxevanis,A.D. and Walter,M.A. ( (2003) ) Structural and functional analyses of disease causing missense mutations in the forkhead domain of FOXC1. Hum. Mol. Genet., , 12, , 2993–3005.
Saleem,R.A., Murphy,T.C., Liebmann,J.M. and Walter,M.A. ( (2003) ) A novel mutation in the FOXC1 forkhead domain causes Axenfeld–Rieger malformations and disrupts FOXC1 function. Invest. Ophthalmol. Vis. Sci., , 44, , 4608–4612.
Sali,A., Potterton,L., Yuan,F., van Vlijmen,H. and Karplus,M. ( (1995) ) Evaluation of comparative protein modeling by MODELLER. Proteins, , 23, , 318–326.
Shimoni,L. and Glusker,J.P. ( (1995) ) Hydrogen bonding motifs of protein side chains: descriptions of binding of arginine and amide groups. Protein Sci., , 4, , 65–74.
van der Heide,L.P., Hoekman,M.F. and Smidt,M.P. ( (2004) ) The ins and outs of FoxO shuttling: mechanisms of FoxO translocation and transcriptional regulation. Biochem. J., , 380, , 297–309.
Overdier,D.G., Porcella,A. and Costa,R.H. ( (1994) ) The DNA-binding specificity of the hepatocyte nuclear factor 3/forkhead domain is influenced by amino-acid residues adjacent to the recognition helix. Mol. Cell. Biol., , 14, , 2755–2766.
Nishimura,D.Y., Searby,C.C., Alward,W.L., Walton,D., Craig,J.E., Mackey,D.A., Kawase,K., Kanis,A.B., Patil,S.R., Stone,E.M. and Sheffield,V.C. ( (2001) ) A spectrum of FOXC1 mutations suggests gene dosage as a mechanism for developmental defects of the anterior chamber of the eye. Am. J. Hum. Genet., , 68, , 364–372.
Kawase,C., Kawase,K., Taniguchi,T., Sugiyama,K., Yamamoto,T., Kitazawa,Y., Alward,W.L., Stone,E.M., Nishimura,D.Y. and Sheffield,V.C. ( (2001) ) Screening for mutations of Axenfeld–Rieger syndrome caused by FOXC1 gene in Japanese patients. J. Glaucoma., , 10, , 477–482.
Raymond,V., Dubois,S., Rodrigure,M., Couture,F., Anctil,J., Cote,G., Amyot,M., Blondeau,P., Bergeron,E., Walter,M. and Morissette,J. ( (2001) ) Chromosomal duplication at the IRID1 locus on 6p25 associated with wide variability of the glaucoma phenotype. Invest. Ophthalmol. Vis. Sci., , 69, (Suppl.), 628.
Panicker,S.G., Sampath,S., Mandal,A.K., Reddy,A.B., Ahmed,N. and Hasnain,S.E. ( (2002) ) Novel mutation in FOXC1 wing region causing Axenfeld–Rieger anomaly. Invest. Ophthalmol. Vis. Sci., , 43, , 3613–3616.
Murphy,T.C., Saleem,R.A., Footz,T., Ritch,R., McGillivray,B. and Walter,M.A. ( (2004) ) The wing 2 region of the FOXC1 forkhead domain is necessary for normal DNA-binding and transactivation functions. Invest. Ophthalmol. Vis. Sci., , 45, , 2531–2538.(R. A. Saleem1, S. Banerjee-Basu3, T. C. )
* To whom correspondence should be addressed. Tel: +1 780 492 9805; Fax: +1 780 492 6934; Email: mwalter@ualberta.ca
Present address: R. A. Saleem, The Institute for Systems Biology, Seattle, WA, USA
ABSTRACT
The forkhead domain (FHD)-containing developmental transcription factor FOXC1 is mutated in patients presenting with Axenfeld–Rieger malformations. In this paper, we report the introduction of positive, negative or neutral charged amino acids into critical positions within the forkhead domain of FOXC1 in an effort to better understand the essential structural and functional determinants within the FHD. We found that FOXC1 is intolerant of mutations at I87. Additionally, alterations of amino acids within -helix 1 of the FOXC1 FHD affected both nuclear localization and transactivation. Amino acids within -helix 3 were also found to be necessary for transactivation and can have roles in correct localization. Interestingly, changing amino acids within -helix 3, particularly R127, resulted in altered DNA-binding specificity and granted FOXC1 the ability to bind to a novel DNA sequence. Given the limited topological variation of FHDs, due to the high conservation of residues, we anticipate that models of forkhead domain function derived from these data will be relevant to other members of the FOX family of transcription factors.
INTRODUCTION
Autosomal dominant mutations in the FOX family transcription factor FOXC1 underlie Axenfeld–Rieger (AR) malformations mapping to human chromosome 6p25 (1,2), thought to arise because of defects in neural crest cell migration and differentiation (3) . Patients with AR malformations present with ocular features that include iris hypoplasia, iridocorneal adhesions, posterior embryotoxon and corectopia. The most severe ocular consequence of AR malformations is the development of glaucoma, a progressively blinding condition, in 50% of the AR patients. Non-ocular features of AR malformations include dental and umbilical anomalies. Cardiac anomalies have been observed to infrequently cosegregate with AR malformations (6–8).
Work in animal models has demonstrated the importance of FOXC1 as a developmentally key transcription factor. Recombinant Foxc1 null mice die peri- or post-natally with massive skeletal, cardiac, ocular and urogenital anomalies (3,8–11). Heterozygous Foxc1+/– mice are viable but recapitulate the spectrum of ocular defects seen within humans carrying FOXC1 mutations (12). Some of these ocular anterior segment defects in heterozygous Foxc1+/– mice can be modified by Tyrosinase, a member of a family of genes involved in melanin biosynthesis (13). FOXC1 is also involved in TGF-? signaling (3,14) and is required for the transcription of another important regulator of developmental processes, the T-box gene Tbx1 (15). The activation of Tbx1 can be induced by Shh, with induction mediated by FOXC proteins (15). Together with a highly related protein, FOXC2 (Figure 1), FOXC1 is thought to act as an important regulator of somitogenesis (15–17).
Figure 1. Sequence alignment of FHDs of FOXC1 and FOXC2. The sequences shown in single letter amino acid code are human FOXC1 (Q12948 ) and FOXC2 (Q99958 ). The mismatches in the sequence between FOXC1 and FOXC2 are shown in blue. The disease-causing missense mutations of FOXC1 analyzed in this study are shown in red. The alanine scanning and charge-altered mutations introduced in FOXC1 are shown above the sequence at their respective amino acid positions. The location of the -helices defined in the solution structure of human FOXC2 (FREAC-11, pdb|1D5V) are schematically represented in the bar below the alignment.
FOXC1 contains the forkhead domain (FHD), a highly conserved sequence of approximately 110 amino acids. The three-dimensional topology of the FHD was first resolved using X-ray crystallography on Foxa3 (formerly HNF3) bound to a DNA target (18) and has since been studied using NMR analysis of FOXC2, Foxd3 and FOXO4 (19–21). This DNA-binding motif is a variant of the helix–turn–helix motif, consisting of three -helices, two ?-sheets and two large loops that form ‘wing-like’ structures (18–21). This region functions as a DNA-binding domain and contains the sequences required for nuclear localization of FOXC1 (22–24).
In vitro site selection experiments were used to determine that FOXC1 bound a nine-base-pair core sequence 5'-GTAAATAAA-3' with high affinity (25). These experiments also found that when the FOXC1 FHD binds, it bends the DNA 94°, while in the context of full length FOXC1, the bound DNA is bent 112° (26). Experiments analyzing FOX binding sites upstream of the Tbx1 gene found that FOXC1 is also able to bind the DNA sequence 5'-AAAACAAACAGGC-3' in EMSAs (15).
Outside the FHD, FOXC1 contains N- and C-terminal transactivation domains, allowing FOXC1 to act as a transcriptional activator, as well as a transcriptional inhibitory domain C-terminal to the FHD (22). FOXC1 is phosphorylated, probably in the inhibitory domain region, but the functional significance of FOXC1 phosphorylation is undetermined (22).
All of the disease-causing missense mutations within FOXC1 identified to date in patients with AR malformations are located within the FHD (Table 1). Analyses of FOXC1 missense mutations characterized through both molecular modeling and biochemical means revealed several different modes of impairment to FOXC1 function (24,26,27). These impairments include defects in localization of FOXC1 to the nucleus, defects in DNA-binding capacity, defects to transactivation and alterations to DNA-binding specificity. One mutation (I87M) was found to reduce levels of FOXC1 either by destabilizing FOXC1 or by reducing efficiency of the translation of FOXC1 mRNA. These molecular defects do not appear to be dependent upon the position of the mutations within the FHD.
Table 1. Molecular defects caused by missense mutations in FOXC1
To deepen our understanding of FHD function, we have created a molecular model of the FOXC1 FHD to predict residues that are critical for function and structure of FOXC1. Critical amino acid positions within the FOXC1 FHD were chosen and converted to an alanine (neutral charge), glutamic acid (negative) or lysine (positive charged) residue. Our model of the FOXC1 FHD was used to predict the effect that these conversions would have on the structure of the FHD. The effect of these conversions on FOXC1 function was then determined by cell biology and biochemical analyses. Our results have revealed the locations of residues that are critical for the translocation of FOXC1 to the nucleus, that guide the interactions of FOXC1 with DNA and organize the FHD in such a manner that FOXC1 is stable, competent to bind DNA and able to activate transcription. We anticipate that this information will also allow prediction of the effects on structure and function of alterations of FHD amino acids of other members belonging to the FOX family of transcriptional regulators.
MATERIALS AND METHODS
Plasmid construction and mutagenesis
The FOXC1 pcDNA His/Max4 protein expression construct has been previously described (24). For mutagenesis reactions, a fragment of FOXC1 from 106 to 777 was PCR amplified and cloned into pGEM T Easy (Promega) using primers 5'-ggc tac acc gcc atg c-3' (forward) and 5'-gct ctc gat ctt ggg cac-3' (reverse). This was then subjected to mutagenesis using the QuickchangeTM (Stratagene) mutagenesis kit according to the manufacturer's protocol and primers summarized in Table 2, with the addition of 5% dimethylsulfoxide. The primers used for mutagenesis contained degenerate oligonucleotides at the site of interest, so that all the three desired mutagenesis products could be obtained from a single reaction. Potential mutant constructs were sequenced and confirmed mutant fragments subcloned back into the FOXC1 pcDNA His/Max4 construct. The final constructs were sequenced again to ensure the integrity of the construct.
Table 2. Primers used for degenerate site directed mutagenesis
Cell culture and transfection
COS-7 and HeLa cells were grown in Dulbecco's modified Eagles medium and 10% fetal calf serum. Cells were transfected using Fugene6TM transfection reagent (Roche) according to the manufacturer's protocol. For protein expression, 2.5 μg of plasmid DNA and 20 μl of Fugene6 were used for a 100 mm transfection plate. For luciferase assays, HeLa cells were transfected in 6-well plates using 500 ng of FOXC1 pcDNA His/Max4, 50 ng of the pGL3-TK-FOXC1 binding site construct, 1 ng of the pRL construct and 3 μl of Fugene6 transfection reagent. COS-7 cells were grown and transfected directly on coverslips when used for immunofluorescence, using 1 μg of FOXC1 pcDNAHis/Max4 and 3 μl of Fugene6 transfection reagent.
Protein expression
Protein extraction was performed 48 h after transfection by washing the cells in phosphate buffered saline (PBS), collecting the cells by scraping and lysing the cells by sonication in the presence of lysis buffer (Promega). After sonicating, the lysates were centrifuged at 13 000 g for 5 min at 4°C. Proteins were resolved by SDS–PAGE, and western-blot analysis was performed using a mouse monoclonal anti-Xpress antibody (Invitrogen) against the pcDNA His/Max4 vector encoded, N-terminal Xpress tag.
Immunofluorescence
All immunofluorescence procedures were carried out at room temperature. After transfection for 24 h, cells were washed with PBS, fixed using 2% paraformaldehyde/PBS for 10 min and washed again in PBS. Cells were permeabilized using 0.05% Triton X-100/PBS for 10 min. After washing, a 5% BSA/PBS solution was applied and the cells were blocked for 1 h. The cells were then washed with a 1% BSA/PBS solution and incubated for 1 h with a 1:400 solution of the anti-Xpress antibody in 1% BSA/PBS. Following incubation, the cells were washed with the 1% BSA/PBS solution and incubated for 1 h with a 1:400 solution of an anti-mouse Cy3 conjugated secondary antibody (Jackson Immunolaboratories) in 1% BSA/PBS. The cells were then washed with 1% BSA/PBS, mounted using mounting media containing 4',6-diamidine 2-phenylindole (DAPI) and the coverslips were sealed using nail polish.
For each construct, cells were scored as showing either nuclear, cytoplasmic, or both nuclear and cytoplasmic staining. At least 100 cells were scored for each construct that was tested.
Northern-blot analysis
About 48 h after transfection, cells were washed with PBS and the RNA was extracted through the use of TRIzol (Gibco/BRL) reagent. The RNA was size separated and loading amounts were equalized on a 1x 3-(N-morpholino) propane sulfonic acid, pH 7.0, 0.66 M formaldehyde agarose gel and transferred onto a Hybond (Amersham) nylon membrane. The membrane was then probed with a 32P-labeled PCR product from the 5' end of FOXC1 (20–233 nt).
Electrophoretic mobility shift assays
Cell extracts containing recombinant FOXC1 protein were equalized for the amounts of recombinant FOXC1 by immunoblot detection. Cell extracts were incubated with 0.45 mM DTT, 0.09 μg poly(dI-dC), 0.225 μg sheared salmon sperm DNA, 13.5% glycerol and 20 000 counts per minute of 32P-labeled oligonucleotides (Table 3) at room temperature for 15 min. After pre-running 6% Tris–glycine–EDTA gels, the electrophoretic mobility shift assay (EMSA) reactions were subjected to electrophoresis.
Table 3. The FOXC1 binding site and variant oligonucleotides
Dual-luciferase assays
The luciferase reporter construct, based on the pGL3 vector with a thymidine kinase promoter and six copies of FOXC1 binding site (Table 3), was described previously (24). Transfected cells (described above) were grown for 48 h and luciferase assays performed using the Dual Luciferase Assay Kit (Promega) according to the manufacturer's protocol. The reactions were replicated a minimum of three times.
Molecular modeling
Molecular models for the FHD of FOXC1 were generated using MODELLER (28) based on the NMR structure of FOXC2 (pdb|1D5V) (20).
RESULTS
FOXC1 was mutated independently to an alanine (A), a glutamate (E) or a lysine (K) residue at positions P79, L86, I87, I91, I126 and R127 by site-directed mutagenesis (Figure 1). The rationale for choosing these positions was based upon both molecular modeling and threading analyses indicating that these changes probably perturb FHD structure, as well as empirical observations from the studies of naturally occurring missense mutations at these positions (24,26,27).
Protein expression and western-blot analysis
Expression and western-blot analysis of the recombinant proteins revealed that all the plasmid constructs expressed recombinant FOXC1 with the exception of FOXC1 mutated at position 87 (Supplemental Figure 1). When I87 was converted to an A, E or K, the FOXC1 protein could not be detected by western-blot analysis (Figure 2B). Northern-blot analysis was carried out to ensure that mRNA was, in fact, being expressed from the transfected recombinant FOXC1 plasmid (Figure 2A). FOXC1 87A/E/K mRNA was detected by northern-blot analysis, but FOXC1 87A/E/K protein could not be detected by western-blot analysis. This finding is consistent with the notion that alterations of I87 drastically affect the cellular levels of soluble recombinant FOXC1 protein.
Figure 2. Conversion of I87 to an alanine, glutamate or lysine residue reduces protein levels of FOXC1. (A) Northern-blot analysis showing expression of FOXC1 and FOXC1 87 A/E/K mRNA. Blot is probed with an N terminal portion of the FOXC1 gene (see Materials and Methods). Loading amounts were equalized on a 1x 3-(N-morpholino) propane sulfonic acid, pH 7.0, 0.66 M formaldehyde agarose gel (see Materials and Methods).
Subcellular localization of the FOXC1 charge conversion proteins
Previous work has shown that missense mutations within the FHD are able to disrupt the normal nuclear localization of FOXC1 in COS-7 cells (26). Based on this observation, the subcellular localization of the FOXC1 charge conversion proteins was tested. By immunofluorescence, the subcellular localization of the FOXC1 mutant proteins was found to be highly variable and dependent upon both the position of the amino acid change and the charge of the amino acid change (Figure 3 and Table 4). With the exception of amino acid residue R127, conversion of the amino acid into the neutral alanine amino acid had the least severe disruption on nuclear localization of FOXC1. Nuclear localization was least perturbed by the presence of the neutral charged P79A (91% nuclear), with the charged amino acids P79E and P79K being more disruptive to FOXC1 localization, at 80 and 75% nuclear respectively. At position L86, the change to the uncharged alanine disrupts the nuclear localization of FOXC1 to a degree similar to what is seen with P79K. Again, when charged amino acids are substituted for leucine at position 86, subcellular localization becomes more perturbed (L86E at 51% nuclear, L86K at 64% nuclear) than what is seen with alanine substitution (72% nuclear). The localization of FOXC1 that is observed with mutations at amino acid positions 79 and 86 is predominantly nuclear within individual cells, but is not confined as strictly to the nucleus compared to cells expressing wild-type FOXC1. When charged amino acids are substituted for the isoleucine at position 91, the disruption becomes severe, with 25% of the population showing nuclear localization for I91E and only 5% of the cellular population showing nuclear staining when FOXC1 carries the I91K mutation. I91A shows levels of nuclear localization that are similar to levels seen with L86E. The differences to nuclear localization upon substitution of differently charged amino acids is most pronounced in I126 and R127, both of which are located in -helix 3 of the FHD. The substitution of charged residues at I126 prevents the strict nuclear localization of FOXC1 in any of the cells in the population, leading instead to mixed nuclear and cytoplasmic staining as well as only cytoplasmic staining. The mislocalization of FOXC1 I126E and I126K appears to be different from the mislocalization when the other amino acids are altered. As can be seen in Figure 3, a major part of the immunofluorescent signal given by these mutations appears to localize to a perinuclear space, within what is likely the endoplasmic reticulum, a site of protein degradation. Conversely, FOXC1 carrying the I126A alteration shows much stronger nuclear localization, where levels are similar to what is seen with P79K or L86A. Finally, FOXC1 R127A, at 56% nuclear localization, does have a nuclear localization defect, but this is much less severe than the disruption seen for FOXC1 R127E, in which mislocalization is complete. Interestingly, nuclear localization of FOXC1 R127K approaches wild-type levels (88%) with little defect in localization seen.
Figure 3. A, E and K conversions in the FOXC1 FHD differentially disrupt the nuclear localization of FOXC1. COS-7 cells were transiently transfected with recombinant XpressTM epitope-tagged constructs. Cy3 fluorescence indicates the position of the Xpress epitope-tagged FOXC1 proteins within the cell. DAPI staining indicates the position of the nuclei.
Table 4. Summary of subcellular localization of wild-type and converted FOXC1 molecules
Charge conversion mutations differentially perturb the DNA-binding capacity of FOXC1
As the FHD is the domain through which FOXC1–DNA interactions are mediated, it was of interest to test how the altering of the amino acid and charge at different positions affects the ability of FOXC1 to bind an in vitro derived FOXC1 binding site (Table 3) (24,25). As seen in Figure 4, FOXC1 with alterations to amino acids P79, L86 or I91 is still able to bind the FOXC1 binding site, although in some cases this binding capacity is reduced. The binding capacity of FOXC1 P79A and P79K is at wild-type levels or near wild-type levels, while P79E appears to bind with a 3- to 4-fold reduction in affinity for the FOXC1 binding site (Figure 4A). The binding capacity of FOXC1 L86A, E or K is at wild-type FOXC1 levels or near wild-type FOXC1 levels (Figure 4B). Binding capacity of FOXC1 with alterations at I91 is reduced in the cases of I91A and I91K with 3- to 4-fold reductions in affinity for the FOXC1 binding site (Figure 4C). FOXC1 I91E binds the FOXC1 binding site with wild-type FOXC1 affinity.
Figure 4. FOXC1 A, E and K mutations disrupt the DNA-binding capacity of FOXC1 to different extents depending on position and charge. Equalized recombinant FOXC1 containing cell extracts were incubated with 32P-labeled FOXC1 binding site (see Table 3) and resolved by native PAGE. Position of the predominant FOXC1–DNA complex is indicated with the filled gray arrowhead.
The deficiencies in FOXC1 binding are more pronounced when A, E or K are introduced at positions I126 or R127. Binding capacity of FOXC1 is completely disrupted by the presence of a lysine or glutamic acid at 126 (Figure 4D). When an alanine is present, binding capacity is retained; however, there is an approximate 4-fold increase in the amount of FOXC1 I126A protein required for wild-type levels of DNA binding. Two of the R127 changes, R127A and R127K, show residual binding capacity, but the levels of this capacity are reduced to well below 10x wild-type FOXC1 levels (Figure 4E). The R127E mutation shows no capacity for DNA binding by this assay.
Amino acids within -helix 3 of the FHD regulate FOXC1 DNA-binding specificity
It has been shown previously that mutations within -helix 3 are able to alter the binding site specificity of FOXC1 (24). It is also known that the specificity of many protein–DNA interactions rely upon the interactions of charged amino acid side chains such as arginines, lysines and glutamic acids, and the charges inherent in the DNA (29). We therefore utilized a series of FOXC1 binding site variants (Table 3) to test whether, in fact, any of the mutations tested in this study were able to alter the binding site specificity of FOXC1. All the altered FOXC1 molecules appeared to bind the variant oligonucleotides in a pattern similar to that seen for FOXC1 (data not shown), with the exception of R127A and R127K (Figure 5) which display altered binding site specificity in comparison with wild-type FOXC1. R127A has a reduced binding affinity for variant oligonucleotides 1 and 2 compared to wild-type FOXC1 affinity for these oligonucleotides (Figure 5A). Interestingly, while wild-type FOXC1 is unable to bind oligonucleotide 7, the R127A FOXC1 recombinant protein is able to weakly bind this variant oligonucleotide. In a manner that is strikingly different from wild-type FOXC1 or any of the other mutant FOXC1 proteins tested in this and other studies, R127K has an affinity for variant oligonucleotide 7 that is stronger than the affinity of FOXC1 R127K for the FOXC1 binding site (Figure 5B). These data indicate that helix 3 of the FHD in FOXC1 may play a strong role in regulating the specificity of FOXC1–DNA interactions.
Figure 5. Conversion of R127 into an alanine or lysine residue alters the binding specificity of FOXC1. FOXC1 containing cell extracts were incubated with 32P-labeled FOXC1 binding site or variant oligonucleotides (see Table 3) and resolved by native PAGE. Position of the predominant FOXC1–DNA complex is indicated with the filled gray arrowhead, an asterisk indicates oligonucleotides that the FOXC1 constructs bind with altered affinity. FOXC1 127A no longer binds oligonucleotides 1 and 2 while binding oligonucleotide 7. FOXC1 127K binds oligonucleotide 7 with a greater affinity than its affinity for the FOXC1 binding site.
Transactivation assays
FOXC1 is able to potently drive expression of a reporter construct when the FOXC1 binding site is present in front of the construct promoter (see Figure 6 inset). The effect of the different charges at the different positions on the ability of FOXC1 to drive transactivation was tested. The majority of constructs tested had severely reduced transactivation activity, including those FOXC1 mutations that retained DNA-binding activity (FOXC1 79A/E, 86A/E/K, 91A/E/K, 126A). Notable exceptions were FOXC1 P79K, R127A and R127K. FOXC1 P79K retained an average transactivation capacity of 66% of wild-type activity, while FOXC1 R127A and R127K had transactivation capacities of 36 and 38% of wild-type FOXC1 activity, respectively.
Figure 6. Disruption of transactivation of a luciferase reporter construct by FOXC1 A/E/K proteins. The thick black bars represent the mean of the values; the Y error bar indicates the standard deviation of the values. The inset shows a schematic of the FOXC1 binding site, thymidine kinase promoter, luciferase gene reporter construct.
DISCUSSION
These experiments provide an in-depth analysis of critical positions within the FHD delineating the nature of essential structural and functional determinants within the FOXC1 FHD (Table 5). This study represents the largest analysis of the FHD domain undertaken for any FOX class protein to date. Since the FHD is a highly conserved domain, with limited three-dimensional topological variations, this study should facilitate the modeling of missense mutations within the FHD for all the members of the FOX family of transcription factors.
Table 5. Summary of the molecular consequences of conversion of these critical amino acid positions into an alanine, glutamic acid or lysine residue
Wild-type FOXC1 localizes completely to the nucleus in almost all cells within a given cell population (24,26,27). To date, no mechanism has been identified whereby FOXC1 is differentially localized in response to environmental stimuli or post-translational modifications, such as the insulin-stimulated phosphorylation of FOXO proteins by PKB and subsequent redistribution of FOXO proteins to the cytosol . When analyzing these mutations of FOXC1, mixed localization of FOXC1 to both the nucleus and the cytoplasm of the cell is often seen. This incomplete localization could be due to inefficient transport of defective FOXC1 molecules into the nucleus, or, alternatively, may reflect a defect in the ability of FOXC1 to be retained within the nucleus.
P79 is not involved in the formation of the FOXC1–DNA complex
Alterations at position 79 located N-terminal to -helix 1 cause the mildest disruptions of the positions tested herein, to the nuclear localization of FOXC1 (Table 4). P79 is oriented in such a way that the amino acid residue extends away from the DNA (Figure 7A) and thus is unlikely to be critical for formation of the FOXC1–DNA complex. These data would indicate that the FHD must be largely intact and the native structure preserved, as none of the alterations at position 79 appeared to alter either the affinity of FOXC1 for the FOXC1 binding site or the specificity of FOXC1 affinity for the FOXC1 binding site. Changes from a proline at amino acid 79, however, do appear to change the overall structure of FOXC1 sufficiently so that problems with not only the distribution of FOXC1, but also its transactivation do occur. It is interesting that while FOXC1 P79K shows more of a defect in FOXC1 subcellular localization (25% mislocalization; Table 4) compared with P79A/E it is the only one of the three that retains the capacity for transactivation, even at 66% of wild-type levels (Figure 6). Clearly, while the conformation of FOXC1 P79K is altered suffiently as to result in a defect in the way this mutant protein interacts with the nuclear transport machinery, the molecule is intact enough to both bind the FOXC1 binding site and participate in transcription from a reporter construct.
Figure 7. Molecular model of FOXC1 forkhead domain. (A) A ribbon model showing the positions of the disease-causing missense mutations in the FHD of FOXC1. (B) A wire diagram summary of the functional sub-domains of the FOXC1 FHD. Regions 1 and 2: the N-terminal portion of the FHD and -helix 1 function in the organization of the FHD for transactivation, DNA binding and nuclear localization. Region 3: -helix 3 is involved in DNA-binding specificity of the FHD and also plays a role in the organization of the FHD with respect to nuclear localization and high efficiency binding. Region 4: the wing 2 region functions in the organization of the FHD for transactivation and DNA binding.
With respect to the P79K mutation, and, in fact, all the FOXC1 mutant proteins that retain transactivation capabilities, it is unknown whether the ability to bind or transactivate in high-order chromatin structures is affected.
Residues in -helix 1 are required for the transcriptionally competency of FOXC1
The mutations located within -helix 1 can have mild (86 A/E/K and 91 A/E/K) or more severe (87 A/E/K) effects. Previous work has shown that a naturally occurring I87M missense mutation is able to reduce the levels of FOXC1 protein produced in COS-7 cells (24), similar to what is seen in this study with the I87 A/E/K mutations. These data imply that I87 is critical in the formation of a stable FHD and FOXC1 molecule with alterations of the isoleucine at position 87 reducing the levels of soluble recombinant FOXC1 protein.
The FOXC1 86 A/E/K and 91 A/E/K mutations impede the exclusive localization of FOXC1 to the nucleus (Figure 3 and Table 4). The localization defects to FOXC1 91 A/E/K are interesting for two reasons. Previous characterizations of naturally occurring FOXC1 L86F, I91S, I91T mutations indicate that mutations to position 91 severely impair nuclear localization, while the L86F mutation shows wild-type nuclear localization of FOXC1 (26,27). Previous work has also defined a nuclear localization accessory signal that lies within amino acid residues 78–93, encompassing the I91 position (22). Molecular modeling predicts that I91 is oriented inwards towards the hydrophobic core of the FHD (Figure 7A). Taken together, these data indicate that I91 plays a critical role in organizing the FHD with respect to interactions between the nuclear transport machinery and FOXC1.
All six of these helix 1 mutations severely disrupt FOXC1 transactivation but retain the ability to bind the FOXC1 binding site, though DNA binding is reduced with some of the mutations. Thus, while the FHD is grossly intact, these helix 1 mutations must have effects that extend beyond the FHD, either increasing the potency of the inhibitory domain C-terminal to the FHD and/or preventing function of both the N-terminal and C-terminal transactivation domains of FOXC1 (22). Amino acids L86 and I91 therefore appear to be critical in organizing the structure of the FHD in such a manner that the FOXC1 molecule as a whole is transcriptionally competent.
I126 is critical for correct nuclear localization and transactivation of FOXC1
Mutations in helix 3 appear to have drastic effects on the function of FOXC1 with respect to localization, transactivation and/or DNA-binding specificity. Conversion of I126 to a neutrally charged alanine residue perturbs both the subcellular distribution of FOXC1 and DNA-binding capacity of FOXC1 to some extent, but the FHD domain retains enough wild-type structure to facilitate FOXC1 localization and DNA binding. Molecular modeling of the FHD predicts that I126 is an important residue in the formation of a hydrophobic core bounded by the three primary -helices (24). The substitution to an alanine is likely to preserve the hydrophobicity of the core, allowing the basic structure of the FHD to remain intact, while the charged substitutions (FOXC1 126E and 126K) would introduce hydrophilic amino acid residues into the hydrophobic core, disrupting the FOXC1–DNA interactions. All three of the FOXC1 126 constructs fail to transactivate gene expression (Figure 6). These data imply that although FOXC1 I126A retains enough native structure to allow both FOXC1 interactions with the nuclear transport machinery and FOXC1–DNA interactions, the overall structure of FOXC1 is altered to an extent that perturbs the transactivation domains that lie at the N and C terminals of the FOXC1 protein. It is also possible that the change in conformation enhances the inhibitory effects of the inhibitory domain rather than diminishing the function of the transactivation domains.
R127 is a critical amino acid with respect to the FOXC1–DNA-binding specificity
The R127 position is a highly conserved position in FOX proteins and again, conservation of the amino acid charge causes the least disruption of FOXC1 localization. Interestingly, both FOXC1 R127A and K, while still binding the FOXC1 binding site with reduced affinity, were found to have alterations to binding-site specificity compared to both wild-type FOXC1 and each other (Figure 5). The specificity of binding of FOXC1 R127A appears to become less stringent with binding of variant oligonucleotides 6–9, while conversely, the binding specificity of R127K appears to be altered so that this FOXC1 molecule now has a higher affinity for variant oligonucleotide 7 (gtaaattaa) than for the FOXC1 binding site (gtaaataaa) (Figure 5).
Resolution of the crystal structure of FOXA2 (18) and NMR of the solution structure of FOXC2 (20), the latter containing a FHD 97% identical to that of FOXC1, indicates that helix 3 lies across the DNA in the major groove, possibly functioning as a recognition helix. Previous work has shown that an I126M missense mutation is able to alter the binding specificity of FOXC1 (24), though not to the same extent as what is seen with 127K. The crystal and NMR resolved structures show that the R127 side chain extends away from the amphipathic helix 3 and is likely to be a contact point between FOXC1 and DNA. A molecular model, based on threading experiments in which FOXC1 was threaded through FOXC2 coordinates, shows the R127 residue extending away from -helix 3, into the major groove of the DNA (Figure 7A). Thus, the positively charged 127K would attract the negatively charged DNA, whereas 127A would neither attract nor repulse. Hence, both these FOXC1 constructs would still be able to bind. The presence of the negative charge at this contact point would create repulsion between FOXC1 R127E and the DNA.
In spite of showing a severe reduction in binding to the FOXC1 binding site, both FOXC1 R127A and R127K are able to drive expression of a reporter construct, at levels that are 36% and 38% of wild-type FOXC1 levels, respectively (Figure 6). It would appear that, based on the levels of transcription relative to the levels of binding, the transactivation domains are properly organized in FOXC1 R127A/K molecules. These data indicate that the key role of R127 is to function in the proper recognition and binding of the FOXC1 binding site, as well as indicating that R127 does not function to organize the FHD in such a manner as to make FOXC1 transcriptionally competent.
It remains a formal possibility that the FOXC1 mutants that are able to bind DNA by EMSA, but show a transactivation defect in in vivo luciferase assays, may have an in vivo defect in DNA binding that underlies the transactivation defect. However, data from the FOXC1 R127A and R127K mutants are not consistent with this possibility. These two mutants retain sufficient DNA binding to facilitate transactivation, indicating that EMSA experiments may be more sensitive to defects in DNA binding than the transactivation assays. Further investigation of the consequences of FOXC1 mutations upon FOXC1–DNA binding and transactivation functions will necessitate utilization of bona fide FOXC1 target genes in the context of native chromatin. Such investigations are currently underway.
A molecular model of the FHD of FOXC1
The FHD is highly conserved, and structural studies have shown limited three-dimensional variation between different FOX FHDs which implies limited variation in modes of DNA recognition. The structural basis of differences in DNA sequence recognition remains undetermined; however, there is evidence that the third helix, which is positioned in the major groove of DNA (31), and the second wing (25) guide protein–DNA interactions. Rather than alterations in the gross topology of the FHD being the basis of DNA recognition, it is thought that differences in charge along the interface between the FHD and DNA provide the basis for sequence specificity (21). Differences in charge along the interface between the FOXC1 binding site and helix-3 of the FOXC1 FHD are likely to underlie the alterations in DNA-binding specificities that are seen in FOXC1 127A and 127K mutant proteins.
A ribbon model showing the positions of the disease-causing missense mutations in the FHD of FOXC1 is shown in Figure 7A. A wire diagram summary of the functional sub-domains of the FOXC1 FHD is shown in Figure 7B. As amino acids predicted to be critical for FOXC1 structure on the basis of this model were actually found to be key determinants of FOXC1 structure and function by our subsequent biochemical analyses, we have confidence that this FOXC1 model has further predictive value.
Finally, this study illustrates the success in merging predictive bioinformatic analyses with experimental cell biology to achieve testable models of protein structure. These studies demonstrate the functional importance of not only the individual amino acids within the FHD of FOXC1 but also demonstrate how -helix 1 and -helix 3 secondary structures in the FHD are involved in FOXC1 function. Helix 1 appears to be involved in the organization of the FHD in such a manner that FOXC1 can be translocated to or retained within the nucleus and is transcriptionally active. Helix 3 appears to also be involved in organization of the FHD so that FOXC1 nuclear translocation or retention occurs and FOXC1 is transcriptionally competent; it also plays a major role in controlling FOXC1–DNA interactions.
Given the high degree of conservation between FHD domains within the FOX family, with respect to the three dimensional topology (18–21), this combination of biological and computational analysis is a framework which will allow for predictive modeling of changes within the FHDs not only for members of the C class of FOX proteins but also for FHDs of other FOX classes as well.
SUPPLEMENTARY MATERIAL
ACKNOWLEDGEMENTS
We thank the Ocular Genetics Laboratory for critical reading of this manuscript. We thank May Yu for technical assistance. This work was supported by the Alberta Heritage Foundation for Medical Research (AHFMR) and the Canadian Institutes for Health Research (CIHR). R.A.S. is supported by a University of Alberta Dissertation Fellowship. M.A.W. is an AHFMR senior scholar and a CIHR investigator.
REFERENCES
Mears,A.J., Jordan,T., Mirzayans,F., Dubois,S., Kume,T., Parlee,M., Ritch,R., Koop,B., Kuo,W.L., Collins,C. et al. ( (1998) ) Mutations of the forkhead/winged-helix gene, FKHL7, in patients with Axenfeld–Rieger anomaly. Am. J. Hum. Genet., , 63, , 1316–1328.
Nishimura,D.Y., Swiderski,R.E., Alward,W.L., Searby,C.C., Patil,S.R., Bennet,S.R., Kanis,A.B., Gastier,J.M., Stone,E.M. and Sheffield,V.C. ( (1998) ) The forkhead transcription factor gene FKHL7 is responsible for glaucoma phenotypes which map to 6p25. Nature Genet., , 19, , 140–147.
Kume,T., Deng,K.Y., Winfrey,V., Gould,D.B., Walter,M.A. and Hogan,B.L. ( (1998) ) The forkhead/winged helix gene Mf1 is disrupted in the pleiotropic mouse mutation congenital hydrocephalus. Cell, , 93, , 985–996.
Lines,M.A., Kozlowski,K. and Walter,M.A. ( (2002) ) Molecular genetics of Axenfeld–Rieger malformations. Hum. Mol. Genet., , 11, , 1177–1184.
Alward,W.L. ( (2000) ) Axenfeld–Rieger syndrome in the age of molecular genetics. Am. J. Ophthalmol., , 130, , 107–115.
Swiderski,R.E., Reiter,R.S., Nishimura,D.Y., Alward,W.L., Kalenak,J.W., Searby,C.S., Stone,E.M., Sheffield,V.C. and Lin,J.J. ( (1999) ) Expression of the Mf1 gene in developing mouse hearts: implication in the development of human congenital heart defects. Dev. Dyn., , 216, , 16–27.
Suzuki,T., Takahashi,K., Kuwahara,S., Wada,Y., Abe,T. and Tamai,M. ( (2001) ) A novel (Pro79Thr) mutation in the FKHL7 gene in a Japanese family with Axenfeld–Rieger syndrome. Am. J. Ophthalmol., , 132, , 572–575.
Winnier,G.E., Kume,T., Deng,K., Rogers,R., Bundy,J., Raines,C., Walter,M.A., Hogan,B.L. and Conway,S.J. ( (1999) ) Roles for the winged helix transcription factors MF1 and MFH1 in cardiovascular development revealed by nonallelic noncomplementation of null alleles. Dev. Biol., , 213, , 418–431.
Kume,T., Deng,K. and Hogan,B.L. ( (2000) ) Minimal phenotype of mice homozygous for a null mutation in the forkhead/winged helix gene, Mf2. Mol. Cell Biol., , 20, , 1419–1425.
Hong,H.K., Lass,J.H. and Chakravarti,A. ( (1999) ) Pleiotropic skeletal and ocular phenotypes of the mouse mutation congenital hydrocephalus (ch/Mf1) arise from a winged helix/forkhead transcriptionfactor gene. Hum. Mol. Genet., , 8, , 625–637.
Kidson,S.H., Kume,T., Deng,K., Winfrey,V. and Hogan,B.L. ( (1999) ) The forkhead/winged-helix gene, Mf1, is necessary for the normal development of the cornea and formation of the anterior chamber in the mouse eye. Dev. Biol., , 211, , 306–322.
Smith,R.S., Zabaleta,A., Kume,T., Savinova,O.V., Kidson,S.H., Martin,J.E., Nishimura,D.Y., Alward,W.L., Hogan,B.L. and John,S.W. ( (2000) ) Haploinsufficiency of the transcription factors FOXC1 and FOXC2 results in aberrant ocular development. Hum. Mol. Genet., , 9, , 1021–1032.
Libby,R.T., Smith,R.S., Savinova,O.V., Zabaleta,A., Martin,J.E., Gonzalez,F.J. and John,S.W. ( (2003) ) Modification of ocular defects in mouse developmental glaucoma models by tyrosinase. Science, , 299, , 1578–1581.
Zhou,Y., Kato,H., Asanoma,K., Kondo,H., Arima,T., Kato,K., Matsuda,T. and Wake,N. ( (2002) ) Identification of FOXC1 as a TGF-beta1 responsive gene and its involvement in negative regulation of cell growth. Genomics, , 80, , 465–472.
Yamagishi,H., Maeda,J., Hu,T., McAnally,J., Conway,S.J., Kume,T., Meyers,E.N., Yamagishi,C. and Srivastava,D. ( (2003) ) Tbx1 is regulated by tissue-specific forkhead proteins through a common Sonic hedgehog-responsive enhancer. Genes Dev., , 17, , 269–281.
Topczewska,J.M., Topczewski,J., Shostak,A., Kume,T., Solnica-Krezel,L. and Hogan,B.L. ( (2001) ) The winged helix transcription factor Foxc1a is essential for somitogenesis in zebrafish. Genes Dev., , 15, , 2483–2493.
Kume,T., Jiang,H., Topczewska,J.M. and Hogan,B.L. ( (2001) ) The murine winged helix transcription factors, Foxc1 and Foxc2, are both required for cardiovascular development and somitogenesis. Genes Dev., , 15, , 2470–2482.
Clark,K.L., Halay,E.D., Lai,E. and Burley,S.K. ( (1993) ) Co-crystal structure of the HNF-3/fork head DNA-recognition motif resembles histone H5. Nature, , 364, , 412–420.
Jin,C., Marsden,I., Chen,X. and Liao,X. ( (1999) ) Dynamic DNA contacts observed in the NMR structure of winged helix protein–DNA complex. J. Mol. Biol., , 289, , 683–690.
van Dongen,M.J., Cederberg,A., Carlsson,P., Enerback,S. and Wikstrom,M. ( (2000) ) Solution structure and dynamics of the DNA-binding domain of the adipocyte-transcription factor FREAC-11. J. Mol. Biol., , 296, , 351–359.
Weigelt,J., Climent,I., Dahlman-Wright,K. and Wikstrom,M. ( (2001) ) Solution structure of the DNA binding domain of the human forkhead transcription factor AFX (FOXO4). Biochemistry, , 40, , 5861–5869.
Berry,F.B., Saleem,R.A. and Walter,M.A. ( (2002) ) FOXC1 transcriptional regulation is mediated by N- and C-terminal activation domains and contains a phosphorylated transcriptional inhibitory domain. J. Biol. Chem., , 277, , 10292–10297.
Pierrou,S., Hellqvist,M., Samuelsson,L., Enerback,S. and Carlsson,P. ( (1994) ) Cloning and characterization of seven human forkhead proteins: binding site specificity and DNA bending. EMBO J., , 13, , 5002–5012.
Saleem,R.A., Banerjee-Basu,S., Berry,F.B., Baxevanis,A.D. and Walter,M.A. ( (2001) ) Analyses of the effects that disease-causing missense mutations have on the structure and function of the winged-helix protein FOXC1. Am. J. Hum. Genet., , 68, , 627–641.
Pierrou,S., Enerback,S. and Carlsson,P. ( (1995) ) Selection of high-affinity binding sites for sequence-specific, DNA binding proteins from random sequence oligonucleotides. Anal. Biochem., , 229, , 99–105.
Saleem,R.A., Banerjee-Basu,S., Berry,F.B., Baxevanis,A.D. and Walter,M.A. ( (2003) ) Structural and functional analyses of disease causing missense mutations in the forkhead domain of FOXC1. Hum. Mol. Genet., , 12, , 2993–3005.
Saleem,R.A., Murphy,T.C., Liebmann,J.M. and Walter,M.A. ( (2003) ) A novel mutation in the FOXC1 forkhead domain causes Axenfeld–Rieger malformations and disrupts FOXC1 function. Invest. Ophthalmol. Vis. Sci., , 44, , 4608–4612.
Sali,A., Potterton,L., Yuan,F., van Vlijmen,H. and Karplus,M. ( (1995) ) Evaluation of comparative protein modeling by MODELLER. Proteins, , 23, , 318–326.
Shimoni,L. and Glusker,J.P. ( (1995) ) Hydrogen bonding motifs of protein side chains: descriptions of binding of arginine and amide groups. Protein Sci., , 4, , 65–74.
van der Heide,L.P., Hoekman,M.F. and Smidt,M.P. ( (2004) ) The ins and outs of FoxO shuttling: mechanisms of FoxO translocation and transcriptional regulation. Biochem. J., , 380, , 297–309.
Overdier,D.G., Porcella,A. and Costa,R.H. ( (1994) ) The DNA-binding specificity of the hepatocyte nuclear factor 3/forkhead domain is influenced by amino-acid residues adjacent to the recognition helix. Mol. Cell. Biol., , 14, , 2755–2766.
Nishimura,D.Y., Searby,C.C., Alward,W.L., Walton,D., Craig,J.E., Mackey,D.A., Kawase,K., Kanis,A.B., Patil,S.R., Stone,E.M. and Sheffield,V.C. ( (2001) ) A spectrum of FOXC1 mutations suggests gene dosage as a mechanism for developmental defects of the anterior chamber of the eye. Am. J. Hum. Genet., , 68, , 364–372.
Kawase,C., Kawase,K., Taniguchi,T., Sugiyama,K., Yamamoto,T., Kitazawa,Y., Alward,W.L., Stone,E.M., Nishimura,D.Y. and Sheffield,V.C. ( (2001) ) Screening for mutations of Axenfeld–Rieger syndrome caused by FOXC1 gene in Japanese patients. J. Glaucoma., , 10, , 477–482.
Raymond,V., Dubois,S., Rodrigure,M., Couture,F., Anctil,J., Cote,G., Amyot,M., Blondeau,P., Bergeron,E., Walter,M. and Morissette,J. ( (2001) ) Chromosomal duplication at the IRID1 locus on 6p25 associated with wide variability of the glaucoma phenotype. Invest. Ophthalmol. Vis. Sci., , 69, (Suppl.), 628.
Panicker,S.G., Sampath,S., Mandal,A.K., Reddy,A.B., Ahmed,N. and Hasnain,S.E. ( (2002) ) Novel mutation in FOXC1 wing region causing Axenfeld–Rieger anomaly. Invest. Ophthalmol. Vis. Sci., , 43, , 3613–3616.
Murphy,T.C., Saleem,R.A., Footz,T., Ritch,R., McGillivray,B. and Walter,M.A. ( (2004) ) The wing 2 region of the FOXC1 forkhead domain is necessary for normal DNA-binding and transactivation functions. Invest. Ophthalmol. Vis. Sci., , 45, , 2531–2538.(R. A. Saleem1, S. Banerjee-Basu3, T. C. )