Mapping Dmef2-binding regulatory modules by using a ChIP-enriched in silico targets approach

Abstract

Mapping the regulatory modules to which transcription factors bind in vivo is a key step toward understanding of global gene expression programs. We have developed a chromatin immunoprecipitation (ChIP)-chip strategy for identifying factor-specific regulatory regions acting in vivo. This method, called the ChIP-enriched in silico targets (ChEST) approach, combines immunoprecipitation of cross-linked protein-DNA complexes (X-ChIP) with in silico prediction of targets and generation of computed DNA microarrays. We report the use of ChEST in Drosophila to identify several previously unknown targets of myocyte enhancer factor 2 (MEF2), a key regulator of myogenic differentiation. Our approach was validated by demonstrating that the identified sequences act as enhancers in vivo and are able to drive reporter gene expression specifically in MEF2-positive muscle cells. Presented here, the ChEST strategy was originally designed to identify regulatory modules in Drosophila, but it can be adapted for any sequenced and annotated genome.

Fig. 1.

Overview of the ChEST strategy. (A) A scheme illustrating the main steps of the ChEST strategy. ChEST consists of the combined in vivo (ChIP) and in silico (computer assisted-CRM prediction) approaches. The in silico approach leads to the selection of a pool of TF-specific CRMs that are spotted to generate a CRM array. In parallel, ChIP is used to purify genomic DNA fragments bound by a given TF in vivo. ChIP-DNA is labeled and used as a probe for hybridization with the generated CRM array, leading to the identification of in silico-predicted CRMs that are enriched in ChIP-DNA. (B) The ChEST strategy used for mapping Drosophila Dmef2 specific-CRMs. The Drosophila genome was scanned by using three combinations of known in vivo binding sites of Dmef2 associated with E-box sites. Asterisk, see Material and Methods for the detailed conditions of genome scanning. This scanning procedure led to the prediction of 1,243 CRMs from which a pool of 99 CRMs were selected, PCR-amplified, and spotted to produce a genomic Dmef2-CRM array. Custom-developed scafi software was used for Dmef2-CRM selection. Filtering was based on the available genome annotations: position of CRMs with respect of adjacent gene and biological properties of the gene (see Fig. 5). DNA fragments bound by Dmef2 in vivo were purified by X-ChIP from Drosophila embryos of stage 11 to stage 15 using anti-Dmef2 antibody. Control X-ChIP was performed with a nonimmune serum. Purified ChIP-DNA was labeled and used to probe the Dmef2-CRM array. The three known Dmef2 target sequences were found to be enriched in ChIP-DNA material (control experiment). Several novel Dmef2-dependent CRMs were identified (example of results). The presented ChIP enriched Dmef2-CRMs are located in the proximity of N-cadherin (CadN), Sulfateless (Sfl), Kettin (Ket), Aconitase (Acon), Inflated (If), and Frizzled2 (Fz2).

Fig. 2.

Genomic positions of ChEST-identified Dmef2-CRMs. Most of the ChIP-enriched CRMs are located upstream of transcription start site of adjacent genes (42%) or within intronic regions (39%). Downstream-located CRMs (19%) are mainly close to the genes (<5 kb from the 3′ end).

Fig. 3.

Dmef2-dependent CRMs act as muscle-specific enhancers in vivo. Nine of 10 analyzed CRMs were found to act as enhancers in vivo.(Left) CRM-driven lacZ reporter gene expression revealed in embryos from the nine different transgenic lines (annotated at the left side). (Middle) Coexpression of lacZ and Dmef2. (Right) lacZ reporter expression with respect to the endogenous expression of the adjacent target gene. In all panels except G-I, which are the ventral views, lateral views of the embryos are shown: A-I, Y and Z′, stage 15 embryos; J-O and V-X, stage 13 embryos; P-R, stage 11 embryos; S-U, stage 12 embryos. (A-C) ket1-lacZ is coexpressed with Dmef2 in a subset of ventral and lateral muscles. (C) ket1-lacZ expression coincides with a high accumulation of endogenous ket protein in ventrolateral region. (D-F) ket2 CRM drives lacZ expression in all Dmef2- and ket-positive somatic muscle fibers. (G-I) ket3-lacZ expression is restricted to only one ventral muscle fiber per hemisegment. Dmef2 and ket are coexpressed in this fiber. (J-L) Ncad-lacZ is detected in dorsally located muscle founder expressing both Dmef2 and Ncad. Activity of Ncad CRM is transient: lacZ expression is seen in stages 12 and 13 only, suggesting it may play a role in myoblast fusion. (M-O) Acon-lacZ is coexpressed with Dmef2 in a discrete subset of lateral muscle precursors located within segmental borders and displaying a low level of Dmef2. (O) Acon transcripts are detected in developing muscles including those expressing Acon-lacZ. (P-R) Fz2-lacZ is expressed in early-stage embryos in one muscle precursor located in the dorsolateral region. (Q and R) Fz2-lacZ coincides with expression of both Dmef2 and Fz2. (S-U) Sfl-lacZ is seen transiently in a subpopulation of dorsal and ventral muscle precursors. (T) Sfl CRM drives lacZ expression in Dmef2-positive cells. (U) Sfl transcripts are distributed ubiquitously implying that Sfl-lacZ is also present in sfl-positive cells. (V and W) dMeso18E CRM drives lacZ expression in the precursor of a Dmef2-positive dorsal muscle. (X) dMeso18E RNA is detected in dMeso18E-lacZ-expressing cells. (Y and Z′) Strong If-lacZ expression is seen in lateral transverse muscle fibers whose nuclei express Dmef2 and extremities accumulate If protein. For immunostaining and in situ hybridization conditions, see Supporting Material and Methods, which is published as supporting information on the PNAS web site.

Fig. 4.

Expression of selected ChEST-identified Dmef2 target genes in wild-type and Dmef2 mutant background. The lateral views of early stage 14 embryos with focal plane on muscle precursors are shown. (A, C, E, G, I, and K) Wild-type expression of six identified Dmef2 target genes. All of them are at least partially coexpressed with Dmef2 (data not shown). In Dmef2 mutant embryos, the expression of Ncad (D), Fz2 (H), Sfl (J), and dMeso18E (I) is dramatically reduced. Dmef2 loss of function leads also to the partial loss of Ket (B) and Acon (F) expression. For immunostaining and in situ hybridization conditions, refer to Supporting Material and Methods.