Background

Dysfunctions of the human FOXP2 gene have been implicated in speech and language impairments [1]-[6]. As a transcription factor, the FOXP2 protein controls the expression of hundreds of downstream genes, many of which play important roles in nervous system development and function [7]-[11]. An adequate amount of functional FOXP2 protein is thought to be critical for the proper development of the distributed neural circuits underlying speech and language [12],[13]. However, how the expression of the human FOXP2 gene is regulated is not clearly understood. miRNAs are small nonprotein-coding RNA molecules that regulate gene expression post-transcriptionally by targeting specific sequences in the 3'-UTRs of mRNAs, leading to mRNA degradation and/or translational suppression [14],[15]. The FOXP2 mRNA has an approximately 4-kb long 3' UTR, twice as long as its protein coding region [1], raising the possibility that FOXP2 expression is regulated by miRNAs. Here we report the identification of multiple miRNAs that downregulate the expression of the human FOXP2 gene by targeting specific sequences in its 3' UTR.

Results

Identification of miRNA binding sites in the human FOXP2 3' UTR

Using cDNAs made from human fetal brains combined with polymerase chain reaction (PCR), we cloned a 3845nt sequence fragment that matched 100% to the human FOXP2 3' UTR sequence annotated in the NCBI database. We searched this FOXP2 3' UTR sequence for miRNA binding sites, requiring a perfect match to the 7-nt seed sequence in a miRNA [15],[16]. This search predicted numerous miRNAs that potentially target the FOXP2 3' UTR. Some of these miRNAs have more than one binding site and the binding site sequences are conserved in higher vertebrate species. We selected 12 miRNAs: miR-9, miR-19b, miR-27b, miR-92a, miR-140-5p, miR-190, miR-200a, let-7a, miR-129-5p, miR-582-5p, miR-892a, and miR-1237 (Figure1) and tested whether they downregulate FOXP2 expression in cell culture systems.

Figure 1
figure 1

Predicted miRNA binding sites in the human FOXP2 3' UTR. The numbers in the parentheses indicate the positions of miRNA binding sites in the human FOXP2 3' UTR. The first nucleotide downstream from the stop codon is denoted as +1.

Full size image

Multiple miRNAs downregulate human FOXP2 protein and mRNA expression

We transfected mimics of each candidate miRNA into human HEK293 cells (where FOXP2 is expressed endogenously) and measured FOXP2 protein levels 72hours later by Western blot analysis. We found that miR-9, miR-19b, miR-140-5p, miR-200a, let-7a, miR-129-5p, miR-582-5p, and miR-892a reduced FOXP2 protein levels significantly (Figure2A). Of these miRNAs, let-7a, miR-9, and miR-129-5p were among the most effective regulators, reducing FOXP2 protein by 70-90%; they are also known to be abundantly expressed in vertebrate brains [17],[18]. We focused on these three miRNAs and tested whether their regulatory effects were dose-dependent at three dosages: 2, 6.5, and 20 nM. At 2 nM, comparing to the control, let-7a and miR-9 each decreased FOXP2 protein levels by about 50%, while miR-129-5p decreased FOXP2 protein by less than 10% (p<0.01 for all). At 20 nM, let-7a and miR-9 decreased FOXP2 protein levels by 90%, and miR-129-5p decreased FOXP2 protein by 70% (p<0.001 for all, Figure2B and C). The dose-dependent downregulation was significant for each miRNA (p<0.0018 for let-7a and miR-9; p<0.0006 for miR-129-5p, Jonckheere-Terpstra test). Using quantitative real time PCR (qRT-PCR), we also found that FOXP2 mRNA level was downregulated by let-7a, miR-9, and miR-129-5p in similar transfection experiments (Figure2D).

Figure 2
figure 2

Multiple miRNAs downregulate human FOXP2 expression. (A) Representative Western blot images and quantitative results showing that miR-9, miR-19b, miR-140-5p, miR-200a, let-7a, miR-129-5p, miR-582-5p, and miR-892a downregulated human FOXP2 protein expression when transfected into HEK293 cells. (B and C) The downregulatory effects of let-7a, miR-9, and miR-129-5p were dosage-dependent. (D) Let-7a, miR-9, and miR-129-5p downregulated FOXP2 mRNA expression in similar transfection experiments as measured by qRT-PCR. For both Western blot and qRT-PCR, Actin was used as an internal control; a miRNA mimic with a scrambled sequence (Ambion) was used as a negative control; and a siRNA against FOXP2 siFOXP2 was used as a positive control. Transfection was performed at least three times (n'='3) for each miRNA, and quantifications were averaged from the three transfections, two Western blot or two qRT-PCR experiments per transfection. *p<0.05, **p<0.01, and ***p<0.001.

Full size image

let-7a, miR-9, and miR-129-5p target specific sequences in the human FOXP2 3' UTR

Next, we tested whether the regulatory effects of candidate miRNAs were mediated by the 3' UTR of the human FOXP2 gene using luciferase reporter assays. We co-transfected miRNA mimics and a luciferase reporter construct containing a wild type FOXP2 3' UTR or no 3' UTR into SH-SY5Y cells, where no endogenous FOXP2 was expressed, and performed luciferase reporter assays. Compared to a control miRNA with scrambled sequence, the eight candidate miRNAs, miR-9, miR-19b, miR-140-5p, miR-200a, let-7a, miR-129-5p, miR-582-5p, and miR-892a significantly repressed luciferase activity when co-transfected with a reporter construct containing a wild type FOXP2 3' UTR (p<0.01 for all), but had no effect when co-transfected with a reporter construct containing no FOXP2 3' UTR. In contrast, miR-27b, miR-92a, miR-190, and miR-1237 did not repress luciferase activity whether the reporter constructs contained a FOXP2 3' UTR or not (Figure3A). These results were consistent with the Western blot results obtained from HEK293 cells as described in Figure2A, indicating that the regulatory functions of these miRNAs were independent of the cell lines we used.

Figure 3
figure 3

The downregulatory effects of let-7a, miR-9, and miR-129-5p are mediated via specific sequences in the human FOXP2 3' UTR. (A) miR-9, miR-19b, miR-140-5p, miR-200a, let-7a, miR-129-5p, miR-582-5p, and miR-892a significantly repressed luciferase activity when co-transfected with a reporter construct containing a wild-type FOXP2 3' UTR, but had no effects when co-transfected with a reporter construct containing no 3' UTR. (B) Mutations in the binding sites for let-7a, miR-9, and miR-129-5p in the human FOXP2 3' UTR. Mutated nucleotides are in lower case letters. (C) Mutations in both the two binding sites for let-7a, miR-9, and miR-129-5p in the FOXP2 3' UTR abolished the repressive effects of these miRNAs in luciferase reporter assays. Transfection was performed three times (n'='3) for each miRNA, and luciferase assays were performed in duplicate. To control for transfection efficiency, Renilla luciferase activity was normalized to the internal control firefly luciferase activity. Error bars'='SEM; *p<0.05, **p<0.01, and ***p<0.001, Tukeys Honest Significant Difference test.

Full size image

We focused on let-7a, miR-9, and miR-129-5p and further tested whether their regulatory effects were sequence-specific. Each of these miRNAs has two binding sites in the human FOXP2 3' UTR. For each miRNA, we made three mutant luciferase reporter constructs with sequences in binding site 1, binding site 2, and binding sites 1+2 mutated respectively by site-directed mutagenesis (Figure3B). Using luciferase reporter assays, we showed that each of these miRNAs significantly repressed luciferase activity when only one binding site was mutated (p<0.05 or p<0.01 for all three miRNAs), but their regulatory effects were completely abolished when both binding sites were mutated (p'>'0.98 for all three miRNAs, Figure3C). In comparison, the negative control miRNA had no effect on all reporter constructs whether it had no 3' UTR, wild type 3' UTR, or mutated 3' UTRs (Figure3C). These results suggested that the downregulatory effects of these miRNAs were mediated through specific sequences in the FOXP2 3' UTR, and each of their respective two binding sites was functional.

Let-7a, miR-9, and miR-129-5p are expressed in the cerebellum of the human fetal brain

FOXP2 is known to be expressed in several brain regions in the human fetal brain, including the cerebellum, the basal ganglia, the thalamus, and the cortical plate [19]. miRNAs that regulate FOXP2 expression and play a role in brain development are expected to be expressed in similar brain regions. We examined the expression of let-7a, miR-9, and miR-129-5p in human fetal brain tissue by in situ hybridization using Locked Nucleic Acid (LNA) modified miRNA detection probes. Due to limited tissue availability, only the cerebellum (16weeks old) was examined. We found that all these three miRNAs, let-7a, miR-9, and miR-129-5p, were expressed in the cerebellum of the human fetal brain (Figure4). These observations were consistent with the roles of these miRNAs in regulating FOXP2 expression during human cerebellum development.

Figure 4
figure 4

Expression of let-7a, miR-9, and miR-129-5p in the cerebellum of the human fetal brain. In all images, miRNA hybridization signals are in red (Cy3) and DAPI stained cell nuclei are in blue. (A and B) A cerebellum section hybridized with a LNA modified let-7a detection probe. Note the cytoplasmic localization of the let-7a signal. (C) A cerebellum section hybridized with a miR-9 detection probe. (D) A miR-9 probe with 4 nucleotides mutated (miR-9m) detected no hybridization signal. (E) A cerebellum section hybridized with a miR-129-5p detection probe. (F) A negative control probe with scrambled sequence detected no hybridization signal. The scale bars are 40'm in A and 20'm in B through F.

Full size image

Discussion

A heterozygous missense mutation in the human FOXP2 protein coding region (i.e., R553H) results in a transcription factor with compromised DNA binding and transcriptional regulatory activity [1],[20]. Presumably, reduced functional dosage of cellular FOXP2 protein causes speech and language impairments [12]. Our results, together with recent reports that miR-9, miR-132, and miR-140-5p regulate Foxp2/FoxP2 expression (Foxp2 and FoxP2 denote respective rodent and avian genes) in animal models [21],[22], highlight the importance of the FOXP2 3' UTR sequence and the roles for miRNAs in regulating FOXP2 expression. These results raise the possibility that sequence variations in the FOXP2 3' UTR, particularly in miRNA binding sites, due to mutations and/or polymorphism may contribute to dysregulation and/or functional variations of the FOXP2 gene. In addition, various genetic, physiological, and environmental factors may influence the expression of miRNAs, thus indirectly impacting FOXP2 expression.

Human subjects carrying FOXP2 mutations exhibit structural and functional abnormalities, presumably due to FOXP2 malfunction, in several brain regions, including the cerebellum [23], where FOXP2 is known to be expressed [19]. The expression of let-7a, miR-9, and miR-129-5p in the human fetal cerebellum is consistent with their roles in regulating FOXP2 expression during early cerebellum development in humans. Recently, dysregulation of miR-129-5p is found in the cerebellum of autistic brains, albeit only a limited number of brains were examined [24]. Further investigation of in vivo functions of these miRNAs will bring insights into their roles in speech and language development and related neural developmental disorders.

Methods

Cloning of the human FOXP2 3' UTR and prediction of miRNA binding sites

Human fetal brain poly(A)+RNA (Clontech) was reverse-transcribed into cDNA using an oligo-dT primer. Using primers designed based on the human FOXP2 3' UTR sequence in the NCBI database [NCBI: NM148898, variant II], we PCR amplified two overlapping fragments representing the FOXP2 3' UTR. The two fragments were ligated at a BamH1 site to obtain a 3845-nt full length human FOXP2 3' UTR. This sequence was used for miRNA binding sites prediction using the software TargetScan. Sequences in the FOXP2 3' UTR matching perfectly to the 7-nt seed sequence of a miRNA were accepted as putative miRNA binding sites [16].

Transfection, Western blotting, and qRT-PCR

miRNA mimics (Ambion) were transfected into human HEK293 cells (60 nM) using Lipofectamine 2000 (Invitrogen) and cells were harvested 72hours later for protein and mRNA assays. Proteins were separated by electrophoresis on 10% SDSPAGE gel. After blotting, membranes were probed with an antibody against the human FOXP2 protein (sc-21069, Santa Cruz Biotechnology). Protein bands were visualized using ECL-Plus and quantified with Image J. For qRT-PCR, total RNA was extracted from transfected cells and reverse-transcribed (iScript cDNA synthesis kit, Bio-Rad). qRT-PCR was performed using the SYBR Green Supermix (Bio-Rad). Primer sequences are listed in the Additional file 1: Figure S1. The specificity of PCR amplification products were validated by electrophoresis and the melting-curve analysis. For each miRNA, transfection was performed at least three times. For each transfection, Western blot and qRT-PCT were performed at least two times, and qRT-PCR was performed in triplicate.

Luciferase reporter assays

Plasmid constructs carrying wild-type or mutant human FOXP2 3' UTRs in the psiCHECK vector (Promega, 100ng) were co-transfected with miRNA mimics (100 nM) into SH-SY5Y cells. Cells were harvested 48h later and luciferase activity was assayed using the Dual-Luciferase Reporter System (Promega). To control for transfection efficiency, Renilla luciferase activity was normalized to the internal control firefly luciferase activity. For each miRNA, transfection was performed three times and luciferase activity was assayed in duplicate.

Mutagenesis of miRNA binding sites

The full length human FOXP2 3' UTR sequence was used as a template in site-directed mutagenesis of miRNA binding sites using the QuikChange II XL Mutagenesis Kit (Agilent). Specific PCR primers were designed using the QuikChangePrimer Design tool (see Additional file 1: Figure S1) and synthesized by Integrated Device Technology (IDT: http://www.idt.com). In some cases, to mutate three nucleotides in a binding site, we performed 2 rounds of mutagenesis. For example, we mutated 2 nucleotides in the first round, and used the mutated plasmid as a template in the second round to mutate the third nucleotide. All mutated binding sites were sequence verified.

In situ hybridization

In situ hybridization was performed as described previously [22]. Briefly, fixed brain sections (10'm thick) were hybridized with LNA modified miRNA detection probes at 38-45C overnight. After washing and blocking, slides were incubated with an Anti-Digoxigenin antibody for 1h, followed by signal amplification using the TSA Plus Cy3 System (PerkinElmer). LNA modified miRNA detection probes were purchased from Exiqon: let-7a probe (1800001); miR-9 probe (8807805); a customer designed mutant miR-9 probe (miR-9m: 5'-TCATAG AGCTAC ATAACCAT AC A-3', underlined are mutated nucleotides); miR-129-5p probe (3848215); negative control probe (9900401). The human fetal brain (16weeks old) cerebellum sections were obtained from a commercial source (Biochain Institute, CA).

Authors contributions

XCL designed the experiment; LF, ZS, WT carried out the experiments and analyzed the results; GZL and XJW contributed to bioinformatics analysis; ZF performed statistical analysis; XCL wrote the manuscript. All authors read and approved the manuscript.

Additional file