Figure 1.
HMG and Helper site configurations in W-CRMs.
(A) Previously characterized W-CRMs [44], [57] with location of predicted HMG (red arrows) and Helper (blue arrows) indicated (see Figure S3 for PWMs of these motifs). Cutoffs for HMG and Helper sites were 5.35 and 6.5, respectively. Direction of arrow indicates orientation of motif (see inset for consensus sequences in both orientations). Number of nucleotides between each motif is indicated. Black asterisks indicate sites that contributed to W-CRM activation by Wnt signaling in cell culture when mutated individually [44], [57]. Red and blue asterisks denote function when all indicated HMG or Helper sites were mutated simultaneously. (B) Systematic mutagenesis of second Helper site in the nkdIntE W-CRM reporter indicates all seven positions contribute to W-CRM activation. Letters indicate mutated nucleotides, with dashes denoting the wild-type sequence. Reporter constructs were transfected into Drosophila Kc cells with or without a plasmid expressing Arm*. Fold activation represents the ratio of Arm*/control, SD equals standard deviation of three biological replicates. (C) HMG and Helper sites work in closely spaced pairs. The nkdIntE reporter contains three functional HMG and two functional Helper sites [44], which were mutated (striped arrows) in different combinations, and tested for Arm* activation in Kc cells. Values represent the mean of three biological replicates ± SD. A Student's T-test was employed to determine significance for data in panels B and C. (D) Nomenclature and symbology for the four possible HMG-Helper site pair orientations. A right pointing arrow indicates the consensus sequence is read 5′ to 3′ on the “top” strand, a left arrow indicates the consensus is read 5′ to 3′ on the “bottom” strand. The sequences shown for HMG and Helper sites are identical to the ones used for DNA binding experiments in Figure 2 and synthetic Wnt reporter constructs (Figure 3A, 4 and 5).
Figure 2.
Binding preferences of TCF/Pan for various HMG-Helper site configurations in vitro.
(A) Competition electromobility shift assay (EMSA) experiments performed with a recombinant TCF fragment containing the HMG and C-Clamp domains, an AK6 IR-labeled probe, and competitor oligonucleotides containing one of the four orientations at 0 or 6 spaces. Images were taken on the Licor Odyssey, and binding intensity quantified with Image Studio 2.0. AK6 and FF0 were the strongest competitors, but Helper sites in all positions improve binding affinity when compared to binding of the HMG sites alone (HMG only 1 and 2). (B) The IC50 value (the measure of DNA concentration required to reduce binding of the labeled probe to 50% of uncompeted levels) for each competitor was calculated using Prism 6.1 (Graphpad). (C) Semilog graphs depicting competition results from three independent experiments. Error bars indicate SD. Sequences of the HMG and Helper motifs the same as shown in Figure 1D (see Table S1 for full sequence of each competitor).
Figure 3.
HMG-Helper pair configuration preferences in cell culture reporter assays.
Kc cells were transfected with the indicated reporters with or without a plasmid expressing Arm* and assayed for luciferase expression. (A) Synthetic W-CRMs containing two HMG-Helper site pairs in all four orientations with 0, 3, 6, 9 or 12 bp spacing between the HMG and Helper site (see Table S1 for sequences of each synthetic W-CRM). In every construct, the two pairs were separated by 6 bp. Two constructs containing only HMG sites (HMG1 and HMG2) were included in the analysis and the reporter vector without an insert is referred to as empty vector (EV). The data shown represents the average of three independent experiments with three biological replicates each. Error bars represent SD and asterisks indicate a significant difference in activation compared to HMG site constructs. (B) The cartoon at the top of the panel depicts the nkdIntE2P2 W-CRM, all HMG and Helpers sites were mutated (striped arrowheads) except one AK6 pair (green box). The spacing and orientation of this pair were altered as indicated. The endogenous AK6 configuration displayed the highest level of activation by Arm*, while AK0, AK12 and FF0 constructs also exhibited higher activation than the HMG site only control. The data are the means of two independent experiments with three biological replicates each, ± SD. Asterisks indicate a significant increase in activation compared to the HMG site only construct. (C) In the nkdUPE2 W-CRM, the HMG and Helper site pair (green box) which contributes most potently to activation [57] was altered as indicated. AK1 and AK6 configurations responded as well as the endogenous FF1 pair, while the FF6 configuration was less active. The data are the means of three biological replicates ± SD, and are representative of several independent experiments. Asterisks indicate reduced activation by Arm* compared to the wild-type construct. In all cases, Student's T-tests were employed to determine significance.
Figure 4.
HMG-Helper pair configuration preferences in imaginal discs.
Brightfield images of imaginal discs from 3rd instar larva containing the indicated lacZ reporter constructs stained for lacZ activity. (A) Wing imaginal discs. (B) Eye/antennal discs. The FF0 and AK6 reporters display the highest expression in these tissues. Neither the promoter alone (HSP70) nor the HMG site only (HMG1 shown) constructs have detectable expression. At least 20 discs for each reporter were analyzed, with representative images shown. The same base pair sequences used in Figure 3 were utilized for the transgenic reporters (see Table S1 for sequences).
Figure 5.
Tissue-specific activity of HMG-Helper pair reporters.
Brightfield images of tissues from 3rd instar larva containing the indicated lacZ reporter constructs stained for lacZ activity. (A–C) LacZ expression in larval epidermis. Expression is seen in cells underlying the naked cuticle, located between denticle belts (arrows). The HMG1 reporter has no detectable expression; FF0 has weak expression and RW0 drives robust expression. (D–F) Ring glands with expression in the Corpora Allata (dotted lines). HMG1 has no detectable expression with AK6 and KK6 displaying weak and strong expression, respectively. (G–G′) Expression of TCFDN in the CA (via the Aug21Gal4 driver) abolishes expression of the KK6 reporter. (H) Summary of expression data from all tissues examined, with number of plus signs and blue hue indicative of the relative level of reporter gene expression. At least 12 samples of each reporter line were analyzed for each tissue, with similar results observed between individual samples.
Figure 6.
TCF/Pan-bound chromatin is enriched for HMG-Helper site pairs.
(A) Distribution of HMG-Helper pairs in genomic sequences bound by TCF/Pan (blue bars) versus random DNA (yellow bars). TCF/Pan-bound sequences were obtained from a ChIP-seq data set from germband extended Drosophila embryos [28]. Equivalent amounts of random DNA from intergenic, intronic and 5′/3′UTR regions were analyzed, with the average of ten such runs displayed. Nearly all the HMG-Helper site configurations were enriched in TCF/Pan bound regions, with the highest degree of enrichment in FF0-2 and AK0-6. Due to the semi-palindromic nature of the Helper site, many sites with nucleotide mismatches were called as both AK/RW or FF/KK. The overlap is indicated in the Venn diagram to the right of each pair of graphs. (B) AK and FF motifs are highly enriched in peaks which are uniquely bound by TCF/Pan (see text for further explanation). Enrichment is less dramatic in TCF/Pan peaks that are shared by Pannier or pMAD.
Figure 7.
Optimizing motif architecture increases transcriptional output.
(A) Cartoon of the nkdUPE2 W-CRM. A RW4 motif (green box) was altered to an AK6, point mutations indicated in lowercase. (B–D) Brightfield images of 3rd instar imaginal discs stained with X-gal. The “optimized” AK6 containing (OPT) W-CRM (B′, C′, D′) drives higher expression levels of the reporter transgene than the wildtype (WT) W-CRM (B,C,D), and expands the region which responds to Wg signaling (red arrows). Representative wing (B), haltere (C), and eye/antennal (D) imaginal discs shown. (E,F) Confocal images of lacZ immunostained nkdUPE2 W-CRM reporters at embryonic stage 12 (E, E′) and stage 13 (F, F′). The “optimized” reporter has a dramatic increase in expression.
Figure 8.
Computationally identified W-CRM in the fd96C locus.
(A) Cartoon of the fd96C region, displaying the location of the W-CRM between fd96Ca and b genes (UTRs in gray, coding regions in black). The position of HMG (red arrows) and Helper (blue arrows) sites and their spacing in the W-CRM is indicated. Asterisks indicate the high scoring FF1 motif identified in the initial computational search. (B) The fd96c W-CRM drives expression in both the dorsal (white arrowhead) and ventral (white arrow) stripes overlapping the Wg expression domain. (B′) RNAi depletion of Arm results in loss of reporter gene activation (yellow arrow head and arrow).
Figure 9.
Identification of a W-CRM at 3R:24.4M that is active in the ring gland.
(A) Cartoon of the W-CRM in relation to the fkh gene, with the location of HMG (red arrows) and Helper (blue arrows) sites indicated, along with their spacing. Asterisks indicate the high scoring FF1 motif identified in the initial computational search. (B) Confocal images of a 3rd instar larval ring gland from a 3R:24.4M fly immunostained for lacZ and Wg. The large lateral cells of the prothoracic gland (PG) display strong nuclear LacZ staining. Wg protein can be seen at the cortical surface of these cells. (C, C′) Confocal images of ring glands from phmGal4 (C) or PhmGal4>TCFDN (C′) ring glands containing the 3R:24.4M transgene immunostained for lacZ and DAPI. TCFDN expression results in a dramatic reduction in lacZ expression in The PG. (D) Normalized pixel intensity of the LacZ reporter was calculated for 120 nuclei representing each condition, and the data were summarized using a Tukey box plot, with outliers represented by circles.
Figure 10.
DNA bending by the HMG domain could explain preferential binding for AK6 and FF0 configurations.
A cartoon based on the NMR-deduced structure of the HMG domain and basic tail (BT) of murine LEF-1 [80] bending the HMG site. The HMG domain is composed of three alpha-helices (red barrels), the first of which binds the minor groove of the HMG site (shown in red), while the BT (red crescent) wraps around to make contact with the major groove. This binding induces a sharp bend in the DNA, most pronounced between positions T3 and T4 (CCTTTGATCTT). In TCF/Pan, which shares 92% identity with the HMG domain of LEF1, the BT is followed by a 10 residue linker, and then by the C-Clamp (blue oval), recently shown to chelate a zinc ion [48]. The C-clamp can bind to Helper sites (blue sequences) either upstream (AK) or downstream (FF) of the HMG site. We postulate that the linker places a constraint on the optimal spacing for the FF and AK orientation of 0 and 6 bp, respectively.