Streamlined analysis schema for high-throughput identification of endogenous protein complexes

Abstract

Immunoprecipitation followed by mass spectrometry (IP/MS) has recently emerged as a preferred method in the analysis of protein complex components and cellular protein networks. Targeting endogenous protein complexes of higher eukaryotes, particularly in large-scale efforts, has been challenging due to cellular heterogeneity, high proteome complexity, and, compared to lower organisms, lack of efficient in-locus epitope-tagging techniques. It is further complicated by variability in nonspecific identifications and cross-reactivity of primary antibodies. Still, the study of endogenous human protein networks is highly desired despite its challenges. Here we describe a streamlined IP/MS protocol for the purification and identification of extended endogenous protein complexes. We investigate the sources of nonspecific protein binding and develop semiquantitative specificity filters that are based on peptide spectral count measurements. We also outline logical constraints for the derivation of accurate complex composition from IP/MS data and demonstrate the effectiveness of this approach by presenting our analyses of different transcriptional coregulator complexes. We show consistent purification of novel components for the Integrator complex, analyze the composition of the Mediator complex solely from our data to demonstrate the wide usability of spectral counts, and deconvolute heterogeneous HDAC1/2 networks into core complex modules and several novel subcomplex interactions.

Fig. 1.

IP/MS optimization for deep interactome coverage. (A) Immunoprecipitation procedure for purification of extended endogenous complexes. (B) Proteins in IP/MS result can be separated into the specific and nonspecific categories. Specific proteins constitute antibody affinities, including targeted (intended) and nontargeted (secondary, cross-reacting) complexes.

Fig. 2.

Extended Integrator interactome. (A) Reciprocal IPs against Integrator subunits retrieve previously known core module and interacting polymerase subunits. (B) Multiple new interactors are discovered consistently with the Integrator: a phosphatase module, OBFC2A/B, four uncharacterized predicted proteins, and a unique Z3 complex consisting of ZMYND8, ZNF687, and ZNF592. (C) Reproducible antibody-specific identifications contain potential antibody cross-reactivity.

Fig. 3.

Core complex subunits of Mediator are defined by 3N analysis. (A) Top IPs where MED12 is present at highest levels (> 5 peptides) were clustered with 3N constraints (see text). BL#, antibody IDs; ∗ identifies primary antibodies where Mediator is a secondary interacting or cross-reacting complex. (B) 3N analysis was performed for all Mediator subunits with sufficient number of identification in our dataset. Protein neighbors that are copresent in multiple reciprocal 3Ns (•) define potential core complex clusters for Mediator. Mediator-interacting polymerase is effectively stratified from the Mediator core by this analysis.

Fig. 4.

De novo IP/MS deconvolution of human HDAC1/2 corepressor complex network. (A) HDAC1-containing CHD4, SIN3A, and RCOR1 complexes were defined by comparison of reciprocal 3Ns. Heterogeneity of HDAC1/2 complexes is revealed as these modules break apart from each other in 3N analysis. Proteins that were directly targeted as antigens are shaded in blue; unique core complex associations are highlighted in orange. (B) Subunit assignment for the HDAC1/2 network intercomplex interactor PBRM1/BRD7 complex.