Table 1.
BnS conditions and enrichment scores.
Figure 1.
BnS identifies Miz-1 preferred DNA binding motifs.
(A) Structure of full length (MBP-Miz-1-FL) and zinc finger domain (MBP-Miz-1-ZF) fusion proteins. MBP-Miz-1-ZF retains the Myc interacting region but not the BTB/POZ domain. (B) Robust expression of purified recombinant MBP tagged proteins was observed at the expected ∼130 kDa size; purification of MBP-Miz-1-FL is shown. Molecular weight standards are labeled in kDa. (C) BnS was performed using MBP tagged proteins, yielding two main motifs, Mizm1 and Mizm2. (D) Ratio of Mizm1-like to Mizm2-like motifs occurring in the list of top 25 BnS hits. (E) Box plot of enrichment scores for Mizm1-like and Mizm2-like motifs identified by BnS.
Figure 2.
EMSA validates binding of labeled P1 and P2 to full-length Miz-1 and its zinc finger domain, with or without MBP tag.
(A) MBP-Miz-1-FL and MBP-Miz-1-ZF bind P1 and P2. MBP alone does not bind either probe, and Miz-1 does not bind the labeled control probe. (B) Untagged Miz-1 (not containing the MBP tag) was produced by IVTT of pCS2-hMiz-1 vector. Molecular weight standards are labeled in kDa. (C) Untagged Miz-1 binds and shifts labeled P1 and P2, but not CP. EV = empty vector (IVTT reaction using pCS2 vector backbone).
Table 2.
Probe sequences used in EMSA experiments.
Figure 3.
Excess unlabeled probes compete with labeled probes to bind MBP-Miz-1-FL (A-B) or MBP-Miz-1-ZF (C-D).
Addition of 200-fold excess unlabeled P1 or P2 abrogates binding of Miz-1-FL or Miz-1-ZF to labeled P1 or P2 (lanes 2 and 5 vs. lane 1; lanes 9 and 12 vs. lane 8). Mutating two to four critical residues in the probe sequences (P1m1, P1m2, P2m1, or P2m2) reduces their ability to compete with labeled probe for binding (lanes 3 and 4 vs. lane 2; lanes 6 and 7 vs. lane 5; lanes 10 and 11 vs. lane 9; lanes 13 and 14 vs. lane 12). Representative images are shown (A, C), along with quantification of three replicate experiments (B, D). * p<0.05, ** p<0.01, *** p<0.001.
Figure 4.
The sequence surrounding the core motif Mizm1 is dispensable for Miz-1 binding.
When the sequence surrounding the Mizm1 motif is mutated (P1m3), the unlabeled probe retains its ability to compete with labeled probe P1 or P2 to bind MBP-Miz-1-FL (A-B) or MBP-Miz-1-ZF (C-D). Representative images are shown (A, C), along with quantification of three replicate experiments (B, D). * p<0.05, ** p<0.01, *** p<0.001.
Figure 5.
Luciferase reporter assays in HeLa cells demonstrate that Miz-1 activates gene expression via Mizm1.
(A) Four luciferase reporter vectors were constructed: pGL3ec containing no putative Miz-1 binding motifs, pGL3e-MizM containing four repeats of both the Mizm1 and Mizm2 motifs upstream of the transcription start site, pGL3e-Mizm1 containing four repeats of the P1 probe sequence, and pGL3e-Mizm2 containing four repeats of the P2 probe sequence. (B) Miz-1 overexpression in HeLa cells produces a statistically significant increase in luciferase reporter activity with all of the three reporter vectors containing putative Miz-1 binding motifs. (C) Three mutant luciferase reporter vectors were constructed, containing two (Mizm1mut2), three (Mizm1mut3), or five (Mizm1mut5) changes in highly conserved bases of the motif. (D) Miz-1 overexpression produces a statistically significant increase in luciferase reporter activation in the presence of Mizm1, but the effect is eliminated by mutating as few as two bases in the motif. (E) Overexpression of c-Myc does not synergize with Miz-1; instead, c-Myc overexpression produces a statistically significant increase in luciferase activity for all conditions: with or without Miz-1 overexpression, and with or without the presence of Miz-1 binding motifs. Luciferase expression was normalized to expression of the Renilla luciferase control reporter vector and to luciferase expression in untreated HeLa cells. * p<0.05; ** p<0.01; *** p<0.001. EV = empty vector control; RC = reverse complement.
Figure 6.
Miz-1 overexpression produces a dose-dependent increase in luciferase reporter expression at high-range (A-B) and low-range (C-D) Miz-1 dosages, while luciferase expression from pGL3ec vector is unaffected by Miz-1.
Miz-1 relative protein expression (x-axis in A and C) was determined by quantification of Western blots (representative images shown in B and D) using Image J, and was normalized to beta-actin and to expression in control untransfected HeLa cells. * p<0.05; ** p<0.01; *** p<0.001.
Figure 7.
Cux1 binds a similar, but not identical, motif to that identified for Miz-1.
(A) Tomtom was used to identify motifs similar to Mizm1 and Mizm2. Alignments are shown between known Cut homeodomain binding motifs and Mizm1 or Mizm2. (B) EMSA reveals that Miz-1 preferentially binds Mizm1 (lane 2 vs. lane 5), while Cux1 preferentially binds Mizm2 (lane 3 vs. lane 6). One representative image is shown (B), along with quantification of three replicate experiments (C). Relative binding intensity of Miz-1 and Cux1 to P1 versus P2 in (C) is defined as the intensity of bound probe in lane 2/lane 5 for Miz-1, and lane 3/lane 6 for Cux1. (D) Cux1-overexpressing HeLa cells show a decrease in luciferase activity when the core Cux1 binding motif (ATCGAT) is present in the reporter vector, but not when the reporter vector contains Mizm1, which does not contain ATCGAT. * p<0.05; ** p<0.01; *** p<0.001.
Table 3.
Motifs with significant similarity to Mizm1 and Mizm2 identified using Tomtom.
Figure 8.
Motifs enriched in Miz-1 ChIP-seq peaks match Mizm1.
(A) The top-scoring peak in NPCs, as reported by Wolf et al., has little similarity to Mizm1. (B) Tomtom alignment of the top-scoring motif in MDA peaks (MDAm) with the second-top scoring motif in NPC peaks (NPCm2), demonstrating that they are nearly identical to each other (p = 8.8×10−10). (C) Tomtom alignment of Mizm1 with the central portion of MDAm, showing statistically significant similarity (p = 0.009). (D) FIMO was used to identify ChIP-seq peaks containing instances of MDAm in the NPC (left) and MDA (right) Miz-1 ChIP-seq peak sets, using a cutoff of p<0.0001. (E) FIMO was used to identify ChIP-seq peaks containing instances of Mizm1 in the NPC (left) and MDA (right) Miz-1 ChIP-seq peak sets, using a cutoff of p<0.0001. (F) FIMO was used to search the ChIP-seq peaks that Wolf et al. validated by ChIP-qPCR for matches to the MDAm and Mizm1 motifs. Examples of statistically significant matches are shown (p = 1.7×10−7 for MDAm in RORC-TSS; p = 1.22×10−5 for Mizm1 in Vps28-TSS).
Figure 9.
Analysis of ChIP-seq peak locations with respect to genes.
(A-B) Peaks lacking MDAm are highly concentrated within 1 kb of the TSS in the ChIP-seq data sets from MDA cells (A) and NPCs (B), while in both cases, peaks containing MDAm are less likely to be localized near the TSS. Density plots were generated in R using the ggplot2 package; peaks occurring more than 50 kb from the nearest TSS were plotted at +/−50 kb. (C-D) Homer annotations of peak locations for ChIP-seq peaks from MDA cells (C) and NPCs (D). The promoter is defined as −1 kb to +100 bp surrounding the TSS; TTS (transcription termination site) is defined as −100 bp to +1 kb surrounding the TTS. Peaks containing MDAm were identified using FIMO, and the distance to nearest TSS and gene-centered annotations were determined using Homer with all RefSeq human (A, C) or mouse (B, D) genes. *** p<0.0001 (Chi-squared test).