Characterization of Two Distinct Nucleosome Remodeling and Deacetylase (NuRD) Complex Assemblies in Embryonic Stem Cells

Abstract

Pluripotency and self-renewal, the defining properties of embryonic stem cells, are brought about by transcriptional programs involving an intricate network of transcription factors and chromatin remodeling complexes. The Nucleosome Remodeling and Deacetylase (NuRD) complex plays a crucial and dynamic role in the regulation of stemness and differentiation. Several NuRD-associated factors have been reported but how they are organized has not been investigated in detail. Here, we have combined affinity purification and blue native polyacrylamide gel electrophoresis followed by protein identification by mass spectrometry and protein correlation profiling to characterize the topology of the NuRD complex. Our data show that in mouse embryonic stem cells the NuRD complex is present as two distinct assemblies of differing topology with different binding partners. Cell cycle regulator Cdk2ap1 and transcription factor Sall4 associate only with the higher mass NuRD assembly. We further establish that only isoform Sall4a, and not Sall4b, associates with NuRD. By contrast, Suz12, a component of the PRC2 Polycomb repressor complex, associates with the lower mass entity. In addition, we identify and validate a novel NuRD-associated protein, Wdr5, a regulatory subunit of the MLL histone methyltransferase complex, which associates with both NuRD entities. Bioinformatic analyses of published target gene sets of these chromatin binding proteins are in agreement with these structural observations. In summary, this study provides an interesting insight into mechanistic aspects of NuRD function in stem cell biology. The relevance of our work has broader implications because of the ubiquitous nature of the NuRD complex. The strategy described here can be more broadly applicable to investigate the topology of the multiple complexes an individual protein can participate in.

Fig. 1.

Reproducible separation of Mta2-containing complexes by Blue Native PAGE. A, FLAG Western blotting of FLAG immunoprecipitates from Mta2-FTAP cells after blue native PAGE separation. The immunoprecipitate was divided in two aliquots, one was denatured prior to BN-PAGE (D) and the other one was left native (N). Molecular weight markers are indicated in kDa. B, Schematic workflow of the experimental set-up used in this study. Affinity-purified protein complexes were separated by BN-PAGE. The gel lane containing the complexes was excised into 48 equal fractions from bottom to top, which were then analyzed individually by LC-MS/MS. The quantitation data from each fraction was used to generate a migration profile for each protein across the BN-PAGE separation length. C, Migration profile of Mta2 across the separation length of the gel.

Fig. 2.

Hierarchical clustering of all identified Mta2-associated proteins. All identified binding proteins were clustered according to the similarity of their migration profiles. The resulting dendrogram and the associated intensity peak distribution are visualized in a heat-map. The NuRD complex cluster is highlighted and the resulting region of interest is indicated. Nonspecific binding proteins, based on five replicate control affinity purification experiments, are labeled with an asterisk.

Fig. 3.

The core NuRD complex subunits associate in two distinct entities. A, Migration profiles for core NuRD subunits Mta1/2/3, Hdac1/2, and Mbd3 across the BN-PAGE separation length. B, Migration profiles for core NuRD subunits Chd4, Gatad2a/b, and Rbbp4/7 across the BN-PAGE separation length. Only fractions from the range of interest (–46) in a representative experiment are displayed. C, Linear regression plot of normalized iBAQ values relative to Mta2 for Mta1/2/3, Hdac1/2, and Mbd3 outlining their similarity to the migration profile of Mta2. Intensity-based total quantification (iBAQ) values were computed for each identified protein in each fraction. Values were normalized against the bait Mta2 iBAQ value. D, Linear regression plot of normalized iBAQ values relative to Mta2 for Chd4, Gatad2a/b, and Rbbp4/7. Chd4 and Gatad2a/b show steep ascending trends indicative of their different contribution to the two NuRD entities. The linear regressions in C and D were calculated only from fractions within the peak limits from replicates performed with benzonase (fractions 26–36).

Fig. 4.

NuRD associates with the MLL methyltransferase complex regulatory subunit Wdr5. A, BN-PAGE migration profile of Wdr5 showing correlation to the NuRD reference profiles of Mta2 and Hdac1. B, Linear regression plot of normalized iBAQ relative to Mta2 for Wdr5, illustrating the comigration trend. C, Western blotting for NuRD subunits Mta2 (FLAG), Hdac2, Chd4, and Rbbp4 following Wdr5 IP. D, An ESC whole cell lysate was subject to size exclusion chromatography. Twenty-eight fractions were collected (3B5–3D8) and analyzed by Western blotting for Mta2-FLAG, Hdac2, Sall4, and Wdr5.

Fig. 5.

Sall4 associates with the high-mass, but not with the low-mass NuRD entity. A, Migration profiles for Mta2, Hdac1, (NuRD reference profiles) and Cdk2ap1. B, Migration profiles for Mta2, Hdac1, and Sall4, which outline comigration of Sall4 with the high-mass NuRD entity. Mta2 and Hdac1 provide a reference profile of the NuRD complex. C, Sall4 Western blotting of FLAG immunoprecipitates from Mta2-FTAP cells after BN-PAGE separation. The immunoprecipiate was divided in two aliquots of which one was denatured (D) and the other left native (N) prior to BN-PAGE. Molecular weight markers are indicated in kDa. D, Suz12 migration profile in the NuRD region. Mta2 and Hdac1 are references for the NuRD migration profile. E, Schematic diagram of the serial immunoprecipitation procedure. FLAG was isolated from the whole cell lysate. The immunoprecipitate provided the input for the subsequent Sall4 purification. The lysate, both immunoprecipitates (IP) and both flow-throughs (FT) were subject to Western blotting. F, Successive FLAG (Mta2) and Sall4 IPs were performed, first on a whole cell lysate from Mta2-FTAP cells and subsequently on the eluate from the first IP. The input lysate, immunoprecipitates (IP) and flow-throughs (FT) were probed for FLAG, Sall4, Hdac2, and Rbbp4.

Fig. 6.

Co-occupancy of target genes by NuRD, Wdr5, Sall4, and Suz12. A, Comparison of Wdr5 and NuRD target genes in mESCs. The proportional overlap of both gene data sets was visualized. The NuRD data set was derived from an overlap of Chd4, Hdac1, and Hdac2 target gene sets in mESCs. B, Comparison of NuRD (same as in A), Sall4 and Suz12 target gene sets. C, The shared genes in B were combined into a new data set (105-NuRD/Suz12; 34-NuRD/Suz12/Sall4; 342-NuRD/Sall4). This data set was reduced to genes colocalized by Wdr5. The full Sall4 and Suz12 data sets (same as in C) were compared with the resulting gene data set.