Protein-Protein Interaction Among the FoxP Family Members and their Regulation of Two Target Genes, VLDLR and CNTNAP2 in the Zebra Finch Song System

Abstract

The Forkhead transcription factor FOXP2 is implicated in speech perception and production. The avian homolog, FoxP2 contributes to song learning and production in birds. In human cell lines, transcriptional activity of FOXP2 requires homo-dimerization or dimerization with paralogs FOXP1 or FOXP4. Whether FoxP dimerization occurs in the brain is unknown. We recently showed that FoxP1, FoxP2 and FoxP4 (FoxP1/2/4) proteins are co-expressed in neurons of Area X, a song control region in zebra finches. We now report on dimer- and oligomerization of zebra finch FoxPs and how this affects transcription. In cell lines and in the brain we identify homo- and hetero-dimers, and an oligomer composed of FoxP1/2/4. We further show that FoxP1/2 but not FoxP4 bind to the regulatory region of the target gene Contactin-associated protein-like 2 (CNTNAP2). In addition, we demonstrate that FoxP1/4 bind to the regulatory region of very low density lipoprotein receptor (VLDLR), as has been shown for FoxP2 previously. Interestingly, FoxP1/2/4 individually or in combinations regulate the promoters for SV40, zebra finch VLDLR and CNTNAP2 differentially. These data exemplify the potential for complex transcriptional regulation of FoxP1/2/4, highlighting the need for future functional studies dissecting their differential regulation in the brain.

Keywords: FoxP1; FoxP2; FoxP4; forkhead transcription factors; protein interactions; speech; transcription factors; zebra finch.

Figure 1

Western blots after co-immunoprecipitations show that all FoxP1/2/4 zebra finch proteins can homodimerize (A–C) and heterodimerize (D–F) in vitro. HeK293 cells were transfected with combinations of expression vectors encoding FoxP1/2/4 proteins that were tagged with FLAG or V5. The FLAG monoclonal antibody was used to immunoprecipitate proteins from cell extracts. Immunoprecipitated proteins were resolved on SDS-polyacrylamide gels, transferred to nitrocellulose and analyzed by sequential immunoblotting with V5 and FLAG antibodies. In all co-immunoprecipitations (A–F), from left to the right, the first lane shows the empty vector (E.V.) as a negative control; lanes 2 and 3 show protein extracts of transfections with only one of the two tagged proteins as further controls; lane 4 shows the protein extract from transfection with both proteins. The upper panels show the V5 detection, and the lower panels the subsequent FLAG detection. For all conditions and detections we show the input proteins, co-immunoprecipitated proteins and the supernatant after co-immunoprecipitation. In all cases, there is a V5 protein co-immunoprecipitated with the FLAG antibody in the lane where both proteins are present and a reduction of the co-immunoprecipitated protein in the supernatant, showing an interaction of both proteins marked with an asterisk (*).

Figure 2

Western blots demonstrating the specificity of the different FoxP antibodies by detecting only one FoxP protein in extracts of HeK293 cells transfected with an over-expression vector carrying either empty vector (E.V.), or the three FLAG-tagged FoxP proteins. Proteins were resolved by SDS-PAGE, transferred to nitrocellulose, and first detected (top panels) with anti-FoxP1 (A), or anti-FoxP2 (B), or anti FoxP4 (C) and sequentially detected FLAG/b-actin antibodies as loading controls (middle panels) (A–C). Bottom panels show detection of actin in the samples as a loading control. In all cases, the specificity of the antibody is evident from a single band in the expected lane (*).

Figure 3

FoxP1/2/4 can hetero-dimerize in the brain. The three panels depict representative co-immunoprecipitation experiments from nuclear protein extracts of adult zebra finch brains with anti-FoxP2 (A,C), anti-FoxP4 (B), or nonspecific IgG (A–C) under non-denaturing conditions. Proteins were resolved by SDS-PAGE, transferred to nitrocellulose, and analyzed by sequential immunoblotting with anti-FoxP1 (A-left and B-right, Westerns), or anti-FoxP2 (A,C left Westerns), or anti FoxP4 (B,C right panels). In all cases, a hetero-dimer was co-immunoprecipitated with the specific antibodies from whole forebrain lysate and no signal of the same size was detected in the IgG control, suggesting an interaction of FoxP proteins.

Figure 4

FoxP1/2/4 zebra finch proteins can oligomerize in vitro. (A) Schematic representation of possible combinations of dimers (left panel) or a multimer (right panel) of FoxP1/2/4 in IPs. (B) Western blots of double-immunoprecipitation of HeK293 cells transfected with combinations of expression vectors encoding FoxP1 FLAG-tagged, FoxP2 Myc-tagged, and FoxP4 V5-tagged proteins, revealing a FoxP1/2/4 multimer. The left panel of the Western blot shows, from left to right, in the first lane the input protein extract after transfection with all three proteins, followed by lane 2 showing the immunoprecipitated FoxP2-Myc (lower panel, asterisk) with the co-immunoprecipitated FoxP1-FLAG and FoxP4-V5 (upper panel, asterisk). Lane 3 shows the absence of immunoprecipitated proteins when using IgG (rabbit), lane 4 shows the subsequent immunoprecipitation of FoxP1-FLAG (upper panel, asterisk) and co-immunprecipitated FoxP4 and FoxP2 (upper and lower panels respectively, asterisks). In the right Western blot the first lane is again the input, followed by the supernatant of the first immunoprecipitation with Myc and IgG rabbit in lanes 2 and 3, and then the supernatant of the subsequent immunoprecipitation with FLAG.

Figure 5

Hetero-oligomerization of FoxP1/2/4 occurs in the brain, in song nucleus Area X. Co-immunoprecipitation performed on nuclear protein extracts from microbiopsies of Area X. The protein extract was split into equal amounts. One was immunoprecipitated with FoxP2 antibody and subsequentially with FoxP4 antibody, the other half was immunoprecipitated with rabbit IgG two times sequentially. Immunoprecipitated proteins were resolved on SDS-polyacrylamide gels, transferred to nitrocellulose, and analyzed by sequential immunoblotting with FoxP2 and FoxP4 or FoxP1 antibodies.

Figure 6

Luciferase assays demonstrate transactivation properties of different combinations of FoxP1/2/4 proteins on the SV40 promoter. FoxP1/2/4 as well as their combinations significantly repressed the pGL4.13-promoter transcriptional activity through a specific DNA-binding site in the SV40 promoter. Significance levels from all combinations to the empty vector control are represented by stars, **p < 0.001–0.01; ***p < 0.001. One way ANOVA; F = 28.79; DF = 7; n = 8; followed by Tukey’s multiple comparison test; Bars show mean of means ± SEM of five independent transfections for each of the eight conditions, presented as luciferase/renilla ratio (RLU), corrected for transfection by pGL4.75 Renilla luciferase activity. 1x = 125 ng of overexpressing vector per well, 2x = 250 ng of overexpressing vector per well. The control transfection value was obtained with the empty expression vector (pcDNA3.1).

Figure 7

FoxP1/2/4 bind and activate the very low density lipoprotein receptor (VLDLR) promoter. DNA binding assays with FoxP1 (A) and FoxP4 (B). Nuclear extracts (1 μg) from HeK293 cells were incubated with the digoxigenin labeled probe (0.8 ng) representing the 27-bp of the VLDLR FoxP2 binding site. Shown in each case are protein lysate of HeK293 cells transiently transfected with empty vector and labeled probe (lane 1), shift in the presence of nuclear extract of FoxPs (lane 2), and complex formation in the presence of 200-fold molar excess of specific un-labeled probe (lane 3) and supershift in the presence of labeled probe, FoxPs protein extract and monoclonal V5 antibody (1 mg/ml; lane 4). In all cases arrows point to free oligo, non specific shift (n.s.), FoxP shift and supershift. (C) Luciferase assays were carried out in HeK293 cells to measure effects of a FoxP1/2/4 alone or in combinations on the VLDLR promoter. Significance levels from all combinations to the empty vector control are represented by asterisk, **p < 0.001–0.01; ***p < 0.001. One way analysis of variance (ANOVA); F = 26.09; DF = 7 and n = 8; followed by Tukey’s multiple comparison. Bars show mean of means ± SEM of four independent transfections, presented as luciferase/renilla ratio (RLU), corrected for transfection by pGL4.75 Renilla luciferase activity. 1x = 125 ng of overexpressing vector per well, 2x = 250 ng of overexpressing vector per well. The control transfection value was obtained with the empty expression vector (pcDNA3.1).

Figure 8

FoxP1 activated, FoxP2 repressed and FoxP4 did not bind or regulate the Contactin-associated protein-like 2 (CNTNAP2) promoter. (A) Schematic of the CNTNAP2 promoter region. Arrows show the region of the primers used to clone the CNTNAP2 promoter region. The location of the predicted transcription start site (TSS), TATA-box, CpG island (blue box), GAP (white box), 5′ UTR (gray box), coding sequence (CDS, black box) and FOXP2 binding sites (red shapes) are denoted by lines. The fragment used for the EMSA experiments (described in Adam et al., in review) is illustrated by the vertical line labeled “EMSA oligo” on the 5′ UTR region. DNA binding assays with FoxP1 (B), FoxP2 (C) and FoxP4 (D) with the CNTNAP2 oligo. Nuclear extracts (1 μg) from HeK293 cells were incubated with the digoxigenin labeled probe (0.8 ng) representing the 46-bp of the CNTNAP2 FoxP2 binding site. Shown in each case are protein lysate of HeK293 cells transiently transfected with empty vector and labeled probe (lane 1), shift in the presence of nuclear extract of FoxPs (lane 2), and complex formation in the presence of 200-fold molar excess of specific un-labeled probe (lane 3) and supershift in the presence of labeled probe, FoxPs protein extract and monoclonal V5 antibody (1 mg/ml; lane 4). In all cases arrows point at free oligo, non specific shift (n.s.), FoxP shift and supershift. (E) Luciferase assays were carried out in HeK293 cells to measure effects of FoxP1/2/4 alone or in combinations on the CNTNAP2 promoter. Significance levels from all combinations to the empty vector control are represented by asterisks, **p < 0.001–0.01; ***p < 0.001. One way ANOVA; F = 21.66; DF = 7 and n = 8; followed by a Tukey’s multiple comparison test. FoxP1 and FoxP2 single transfections significantly activated or repressed the pGL4-CNTNAP2 transcriptional activity through a specific DNA-binding site in the CNTNAP2 promoter (One way ANOVA; Tukey’s multiple comparison test; **P < 0.005; ***P < 0.0001). FoxP4 as well as all other combinations did not regulate the CNTNAP2 promoter. Bars show mean of means ± SEM of seven independent transfections presented as luciferase/renilla ratio (RLU), corrected for transfection by pGL4.75 Renilla luciferase activity. 1× = 125 ng of overexpressing vector pro well, 2× = 250 ng of overexpressing vector per well. The control transfection value was obtained with the empty expression vector (pcDNA3.1).