Distribution of expression measurements after transient and site-directed stable transfection. (A) Partitioning DNA segments into classes of preCRMs or preNeutrals, based on RP score and predicted GATA-1 binding sites. (B) Transient transfection assay. (Left) Maps of the expression vectors show a firefly luciferase reporter gene expressed from the human A gamma globin gene promoter (HBG1pr) and a multiple cloning site (MCS) for inserting the preCRMs (pCRM). A co-transfection control plasmid has the Renilla luciferase gene expressed from the thymidine kinase promoter (TKpr). (Right) The distributions of luciferase activity measurements for three categories of tested DNA (preCRMcc, preCRMcnc and preNeutral) in transfected K562 cells. (C) Site-directed stable integration assay. The donor plasmid has an EGFP gene expressed from the human HBB promoter, preceded by a MCS into which preCRMs are inserted for test constructs. Cre-mediated exchange via loxP sites can replace the HyTK marker by the test expression cassette. The graphs show the distribution of EGFP fluorescence in pools of cells carrying expression constructs from the three major groups of tested DNAs, represented as log₂ fold change, for each day during HMBA induction period. The distributions are depicted as box-plots, in which the box encompasses the values from the 25th to 75th percentiles, and the median is shown as a line across the box. The whiskers extend to the closer of two values: either the limit of the values in the distribution or 1.5 times the interquartile distance (the difference between the 75th and 25th percentiles). Values beyond 1.5 times the interquartile distance are plotted as circles.

Examples of validated preCRMs in three loci. Selected validated preCRMs from Zfpm1 (A), Gata2 single preCRMs (B), a combination of preCRMs from Gata2 (C), and Hipk2 (D) are shown; features and activities of other validated preCRMs are shown in the Supplemental figures. The conventions for displaying features and activities are the same as in Figure 3.

Correlation of enhancer activity with genomic features of preCRMs and distribution of segments by RP score and validation rates. (A) Correlation of enhancer activity with RP score. The negative log₁₀ P-value for the difference in activity of each preCRM being different from that of preNeutrals is plotted against the mean RP score for the preCRM. The more significant activity from either the transient transfection or the site-directed stable integration assay was chosen for each preCRM. For stable transfection data, the smallest P-value during induction was used. (B) Correlation of enhancer activity with the number of ccGATA-1BSs in each preCRM. The values on the vertical axis are the same as in panel A, and the horizontal axis gives the number of matches to conserved consensus GATA-1 binding sites in each preCRM. (C) The distribution of DNA segments in the eight target loci after segmentation by RP score. Runs of at least 100 nucleotides of mouse DNA with mean RP scores within a designated range (bin in the histogram) were identified, and the number of segments in each bin of RP score is plotted. (Dark gray bars) Validation rate for DNA fragments containing at least one ccGATA1BS in each RP bin; the number of fragments tested in the enhancer assays is given on the top of each bin.

Maps of three target loci with preCRMs and genomic features. The tracks in each graph show chromosomal coordinates (mm7 assembly), positions and abbreviated names of preCRMs, positions of both ccGATA1BSs (matches to conserved consensus GATA-1 binding sites) and cncGATA1BSs (matches to position specific weight matrix of GATA-1 binding sites), a graph of the RP score based on five-species TBA alignments, and the gene exon–intron structure. Similar maps for Btg2, Hebp1, Hipk2, Hist1h1c, and Vav2 are in Supplemental Figure S2.

Features, validation, and mutagenesis of a preCRM in the Alas2 locus. (A) Features and activities of preCRM Alas2R1. The graph on the left shows important features of the preCRM: the chromosomal position and size, segments that match the consensus GATA-1 binding site (black rectangles) and matches to the weight matrix for the GATA-1 binding site that are not matches to the consensus (gray rectangles) in mouse and in non-rodent (non-rod) sequences, regulatory potential scores (RP), and phastCons scores (PhastCons) across the genomic interval. The two graphs on the right summarize the activities of the preCRMs (connected dark filled circles), represented as log₂ fold change and compared with that of the preNeutrals (light gray box-plot). The graph labeled “Ttfx” plots the values of the transient transfection experiments in K562 cells. The seven columns in the rightmost graph show the results for the induction time course (Stbl Days HMBA) of pools of cells after site-directed integration of the test expression cassette in MEL_RL5 cells. Significant differences of preCRM activity compared with the distribution of activity measurements for preNeutrals are denoted by a black bar along the top of the graphs; otherwise the bar is gray. (B) Mutagenesis and tests of GATA motifs in Alas2R1. The alignment of a subregion of mouse Alas2R1 with dog (canFam2) and human (hg17) shows two conserved consensus GATA-1 binding sites (outlined). Alas2R1 was mutated in individual (BSmut1 and BSmut2) and both (BSmut1 + 2) GATA-1 binding sites, with the block substitutions shown beneath the alignment. The activity levels of expression plasmids carrying the wild-type and mutated Alas2R1 preCRM are shown in the bottom panel, using the same conventions as in A, except that the numerical P-values for a difference of the activity of the test construct compared with the activities of preNeutrals is given at the top of each column.

Relative GATA-1 occupancy at a subset of preCRMs. Using antibodies directed against the estrogen-receptor moiety of the GATA-1-ER fusion protein (E) and the N terminus of GATA-1 (G), chromatin fragments were immunoprecipitated from untreated G1E-ER4 cells (uninduced) and G1E-ER4 cells treated with estradiol for 24 h (induced). Normal mouse (m) and rat (r) IgG controls were performed in parallel. Real-time polymerase chain reactions of each amplicon were performed in duplicate and quantified using SYBR green dye. Signals were referenced to a dilution series of the relevant input sample. Amplicons from the β-major globin promoter (Hbb-b1 prom) and an upstream region of the Zfpm promoter (Zfpm1UP) are shown as positive and negative controls.