Cut-like homeobox 1 and nuclear factor I/B mediate ENGRAILED2 autism spectrum disorder-associated haplotype function

Abstract

Both common and rare variants contribute to autism spectrum disorder (ASD) risk, but few variants have been established as functional. Previously we demonstrated that an intronic haplotype (rs1861972-rs1861973 A-C) in the homeobox transcription factor ENGRAILED2 (EN2) is significantly associated with ASD. Positive association has also been reported in six additional data sets, suggesting EN2 is an ASD susceptibility gene. Additional support for this possibility requires identification of functional variants that affect EN2 regulation or activity. In this study, we demonstrate that the A-C haplotype is a transcriptional activator. Luciferase (luc) assays in mouse neuronal cultures determined that the A-C haplotype increases expression levels (50%, P < 0.01, 24 h; 250%, P < 0.0001, 72 h). Mutational analysis indicates that the A-C haplotype activator function requires both associated A and C alleles. A minimal 202-bp element is sufficient for function and also specifically binds a protein complex. Mass spectrometry identified these proteins as the transcription factors, Cut-like homeobox 1 (Cux1) and nuclear factor I/B (Nfib). Subsequent antibody supershifts and chromatin immunoprecipitations demonstrated that human CUX1 and NFIB bind the A-C haplotype. Co-transfection and knock-down experiments determined that both CUX1 and NFIB are required for the A-C haplotype activator function. These data demonstrate that the ASD-associated A-C haplotype is a transcriptional activator, and both CUX1 and NFIB mediate this activity. These results provide biochemical evidence that the ASD-associated A-C haplotype is functional, further supporting EN2 as an ASD susceptibility gene.

Figure 1.

ASD-associated rs1861972–rs1861973 A–C haplotype increases gene expression. (A) Luciferase (luc) constructs used for transfections are diagramed: TATA—pGL3pro vector driven by SV40 minimal promoter; A–C and G–T—pGL3pro vector containing full-length human EN2 intron with ASD-associated A–C haplotype (A–C) or unassociated G–T haplotype (G–T). The intron was cloned 3′ of luc gene and 5′ of poly A signal, so it is transcribed and spliced as the endogenous gene. (B, C) Equimolar amount of the three constructs were transiently transfected into P6 mouse cerebellar granule neurons and cultured for 24 h (B) or 72 h (C). Luciferase activities were then measured and normalized to the levels of Renilla reniformis. Relative luc units are expressed as percent of TATA control. Note the A–C haplotype function is more pronounced at 72 h. n = 7 (B), n = 10 (C). *P < 0.05, **P < 0.01, ***P < 0.0001; two-tailed paired Student's t-test.

Figure 2.

The A–C haplotype function requires both the ASD-associated A and C alleles. (A) Luc constructs used for transfections are diagramed: TATA—pGL3pro vector driven by SV40 minimal promoter; A–C and G–T—pGL3pro vector containing full-length human EN2 intron with ASD-associated A–C haplotype (A–C) or unassociated G–T haplotype (G–T) as described in Fig. 1; A–T and G–C—pGL3pro vector containing full-length human EN2 intron with one ASD-associated (A or C) and one unassociated allele (T or G). (B) Constructs were transfected and analyzed as previously described. Luc activities were measured after 24 h in culture. Only the ASD-associated A–C haplotype increases gene expression above the TATA control and is significantly different from all the other haplotypes. n= 6, *P < 0.05, ** P < 0.01, *** P < 0.0001; two-tailed paired Student's t-test. (C, D) EN2 intron encompassing rs1861972 and rs1861973 was sequenced and compared from five primate species: P. pithecia, A. nancymai, T. gelada, P. troglodytes and M. mulatta. Sequence alignment was performed for 10-bp 5′ and 10-bp 3′ of rs1861972 (C) and rs1861973 (D). Arrows indicate position of rs1861972 and rs1861973 in humans and primate species. Perfect matches between all the species are marked with asterisks at the bottom. For heterozygous loci, both alleles are shown. Note that the ASD-associated A–C haplotype is conserved in primate species and the G–T haplotype is only observed in humans. Neither A–T nor G–C haplotype pairs are observed in our analysis. These data suggest that the A–C haplotype evolved together.

Figure 3.

202 bp encompassing rs1861972–rs1861973 is sufficient for the A–C haplotype function. (A) Luc constructs used for transfections in (B) are diagramed: TATA—pGL3pro vector driven by SV40 minimal promoter; A–C 202 and G–T 202—pGL3pro vector containing 202 bp encompassing rs1861972–rs1861973 with A–C (A–C 202) or G–T (G–T 202) haplotype cloned 5′ of SV40 promoter. (B) Constructs were transfected and analyzed as previously described. Luc activities were measured after 24 h in culture. Two-hundred and two base pairs are sufficient to recapitulate the full-length A–C haplotype activator function. n= 9, *P < 0.05, ***P < 0.0001; two-tailed paired Student's t-test. (C) Luc constructs used for transfections in (D) are diagramed: TATA—pGL3pro vector driven by SV40 minimal promoter; AC-40 and GT-40—pGL3pro vector containing 20-bp oligos for rs1861972-A and rs1861973-C or G and T, which are adjoined 5′ to 3′ and cloned 5′ of SV40 promoter. (D) Constructs were transfected and analyzed as previously described. Luc activities were measured after 24 h in culture. Both oligomers result in increased gene expression but failed to show significant differences between ASD-associated and unassociated haplotypes. n= 4, **P < 0.01; two-tailed paired Student's t-test.

Figure 4.

ASD-associated A–C haplotype specifically binds a protein complex. EMSAs were performed using 202-bp probes for the ASD-associated A–C haplotype (AC probe) or unassociated G–T haplotype (GT probe) and nuclear extract from differentiated P6 mouse cerebellar granule neurons. (A) Nuclear extract was incubated with biotinylated A–C or G–T probes and run on 5% non-denaturing polyacrylamide gel. Arrow indicates specific binding of the A–C haplotype. For competition analysis, 120 molar excess of unlabeled competitors (120× AC and 120× GT) were added into the reactions. Note that excess competitors not only reduced the probe–protein binding in sequence-specific manner, but also chased the protein–DNA shift to a faster migrating complex (arrowhead). (B) Protein titration was carried out using decreasing amount of nuclear extract. Lower protein/probe ratio results in a faster migrating complex similar to the one observed in A (arrowhead).

Figure 5.

Proteins specifically binding to the A–C haplotype were identified by affinity purification and mass spectrometry. Protein identification procedure is diagramed. Nuclear extract from differentiated P6 cerebellar granule neurons was pre-absorbed with poly d(I-C), whereas the 202-bp biotinylated DNA probes with A–C, G–T or lambda sequences were bound to streptavidin-conjugated magnetic beads. Nuclear extract and probes were then incubated together in EMSA binding buffer and thoroughly washed. Proteins bound to the probes were eluted and briefly separated on an SDS–PAGE gel. Proteins were excised from the gel and digested with trypsin. Extracted peptides were subjected to LC-MS/MS.

Figure 6.

CUX1 and NFIB bind the A–C haplotype together. Supershifts were performed using nuclear extract from HEK293T cells and 202-bp probes encompassing rs1861972–rs1861973 with A–C haplotype (AC). (A, B) Rabbit polyclonal anti-CUX1 antibody (CUX1 in A), rabbit polyclonal anti-NFI antibody (NFI in B) or unrelated rabbit polyclonal antibody (UR) was added to nuclear extract before binding to probes and separated on 4% acrylamide gel. Arrow indicates supershifted band by anti-CUX1 (A) or anti-NFI antibody (B). (C) Effect of anti-CUX1 (CUX), anti-NFI (NFI) antibody or of both antibodies (CUX1, NFI) was compared on the same EMSA gel. Brackets highlight the supershifted bands by anti-CUX1, anti-NFI or both antibodies together. Arrowhead indicates higher shift when both anti-CUX1 and anti-NFI antibodies were applied together.

Figure 7.

NFIB binds the A–C haplotype in vivo. ChIP assays were conducted in human in HEK293T and SH-SY5Y cells. Cells were fixed with 1% formaldehyde and sonicated before immunoprecipitation with rabbit polyclonal anti-NFI antibody (anti-NFI). Immunoprecipitation without antibody (NoAb) was used as negative control. Precipitated DNA was assayed by qPCR to detect the presence of the area encompassing rs1861972 (972), rs1861973 (973) or non-specific region (NC) as negative control. Relative levels of binding were normalized to 0.1% of total input DNA. Error bars represent standard error of mean measured from four independent experiments. (A) When NFIB was over-expressed in HEK293T cells, binding to the A–C haplotype is observed ∼3-fold higher compared with the two negative controls. (B) Endogenous binding of NIFIB was examined in SH-SY5Y cells. Significant enrichment to the rs1816972 and rs1861973 was observed compared with no antibody and non-specific region (NC) controls.

Figure 8.

CUX1 and NFIB are sufficient to regulate the A–C haplotype function. Full-length intronic A–C and G–T luc constructs (Fig. 1A) were co-transfected with empty pCMV-SPORT6 vector (Empty) or CUX1 and NFIB expression constructs (CUX1 + NFIB) into HEK293T cells. Luc activities were measured and normalized to the levels of R. reniformis. (A) Relative luc units are expressed as percent of the promoter control (Pro). Note that CUX1 and NFIB expression is sufficient to convert the A–C haplotype into an activator in HEK293 cells. (B) To investigate whether CUX1 and NFIB affected the difference in luc activity for the A–C and G–T haplotypes, A–C/G–T luc ratios were calculated in the absence (Empty) and presence (CUX1, NFIB and CUX1 + NFIB) of over-expression. The difference between the A–C and G–T haplotype was significantly increased when both CUX1 and NFIB are over-expressed. n= 4, two-tailed paired Student's t-test, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Figure 9.

Endogenous EN2 levels are affected by the CUX1-NFIB double knock-down. To further investigate the regulation of endogenous human EN2 expression by CUX1 and NFIB, stable single (CUX1, NFIB) or double (CUX1 + NFIB) knock-down (k/d) cell lines were established in HEK293T cells. A non-silencing control was also generated. Endogenous EN2 mRNA levels were measured by Taqman qRT-PCR for each cell line. EN2 levels are presented as fold change compared with non-silencing control. n= 3, *P < 0.05, **P < 0.01, t-test, two-tailed, paired.