TurboCas: A method for locus-specific labeling of genomic regions and isolating their associated protein interactome [ChIPseq]

Regulation of gene expression during development and stress response requires the concerted action of transcription factors and chromatin-binding proteins. Because this process is cell-type specific and varies with cellular conditions, mapping of chromatin factors at individual regulatory loci is crucial for understanding cis-regulatory control. Previous methods only characterize static protein binding. We present “TurboCas,” a method combining a proximity-labeling (PL) enzyme, miniTurbo, with CRISPR-dCas9 that allows for efficient and site-specific labeling of chromatin factors in mammalian cells. Validating TurboCas at the FOS promoter, we identify proteins recruited upon heat shock, cross-validated via RNA polymerase II and P-TEFb immunoprecipitation. These methodologies reveal canonical and uncharacterized factors that function to activate expression of heat-shock-responsive genes. Applying TurboCas to the MYC promoter, we identify two P-TEFb coactivators, the super elongation complex (SEC) and BRD4, as MYC co-regulators. TurboCas provides a genome-specific targeting PL, with the potential to deepen our molecular understanding of transcriptional regulatory pathways in development and stress response. Overall design: The miniTurbo proximity labeling enzyme was fused to the N terminus of a catalytically dead Cas9 and cloned into a lentiviral expression vector with an in-frame blasticidin selection cassette, separated by a 2A autocatalytic peptide. HEK293T human embryonic kidney cells and HCT116 human colon carcinoma cells were used to stably express TurboCas or MiniTurbo. To target TurboCas to the FOS locus, sgRNA were designed against regions ~600bp and ~300bp upstream of the transcription start site (TSS) and the first FOS exon (hereafter referred to as FOS-600, FOS-300, FOX-EX1). TurboCas or MiniTurbo cell lines were then transduced with LentiGuide-Puro virus containing the sgRNA and selected with Puromycin to generate stable sgRNA-expressing cell lines. To confirm that TurboCas was specifically binding at the intended sgRNA target sites, we assessed TurboCas chromatin binding by performing ChIP-seq with a commercial polyclonal dCas9 antibody. In addition to the three sgRNA cell lines (TurboCas-Fos600, TurboCas-Fos300, TurboCas-FosEX1), the experiments were also performed in the MiniTurbo and TurboCas-empty controls. To functionally validate hits identified through TurboCas mass spectrometry, we selected the top hits as targets and performed shRNA knockdown of these genes in HCT116 cells. Cells were selected with puromycin for 10 days and then heat shocked for 3 hours, followed by RNA extraction and RNA-seq both to evaluate shRNA efficiency and to assess gene expression patterns. For dCas9 ChIP, 5 x 107 cells were used. For Pol2 ChIP, 1 x 107 cells were used. In brief, HCT116 or HEK293T cells were fixed in 1% formaldehyde for 15 minutes, followed by quenching, cell lysis and chromatin shearing with an E220 focused ultrasonicator (Covaris). Sheared chromatin was subjected to immunoprecipitation with a Cas9 antibody (Diagenode, #C15310258-100) or a Pol2 antibody (Cell Signaling Technologies, 14958). Immunoprecipitated DNA was then washed, eluted, revers-crosslinked and purified using a ChIP DNA Clean&Concentrator kit (Zymo Research, D5205) prior to library preparation using the KAPA HTP library preparation kit (KAPA Biosystems). RNA from HCT116 cells was isolated using the RNeasy mini kit (Qiagen, #74106) as per manufacturer's instructions. RNA-seq libraries were prepared using the NEBNext® Ultra™ II Directional RNA Library Prep Kit for Illumina® (New England Biolabs, # E7760L). ChIP-seq and RNA-seq libraries were paired-end read sequenced on the NovaSeq 6000 (Illumina). Raw BCL output files were processed using bcl2fastq (Illumina, version 2.20) prior to quality trimming using CutAdapt1.14. Trimmed ChIP-seq reads were aligned to the human genome (University of California at Santa Cruz [UCSC] hg38 build) using Bowtie v2.2.6. Only uniquely mapped reads satisfying the two-mismatch maximum threshold within the entire length of the gene were considered for ensuing analyses. Mapped ChIP-seq reads with MAPQ quality score >=30 were extended to 150bp to represent sequenced fragments. Raw read counts from both ChIP-seq were normalized to total reads per million (RPM). Output BAM files were converted into bigwig coverage plots to generate UCSC genome browser tracks. Peak calling was performed using MACS2 with the default (narrow) option. Called peaks were annotated with HOMER program annotatePeaks.pl. Heatmaps and metaplots from bigwig files were generated using deepTools v3.1.1. For region-based categorization of dCas9 signal, bed files were split by chromosome and coordinate (in the case of chromosome 14), these split files were then used as inputs for profile plotting with deepTools. RNA-seq reads were aligned to the hg38 human genome using STAR aligner v2.5.2 and outputted bam files were sorted using samtools v1.6. For RNA-seq, exonic reads were assigned to specific genes from Ensembl release 72 and genome assembly GRCh38.p13 using the python package HTSeq-0.6.1.