Decoding the chromatin proteome of a single genomic locus by DNA sequencing

Abstract

Transcription, replication, and repair involve interactions of specific genomic loci with many different proteins. How these interactions are orchestrated at any given location and under changing cellular conditions is largely unknown because systematically measuring protein-DNA interactions at a specific locus in the genome is challenging. To address this problem, we developed Epi-Decoder, a Tag-chromatin immunoprecipitation-Barcode-Sequencing (TAG-ChIP-Barcode-Seq) technology in budding yeast. Epi-Decoder is orthogonal to proteomics approaches because it does not rely on mass spectrometry (MS) but instead takes advantage of DNA sequencing. Analysis of the proteome of a transcribed locus proximal to an origin of replication revealed more than 400 interacting proteins. Moreover, replication stress induced changes in local chromatin proteome composition prior to local origin firing, affecting replication proteins as well as transcription proteins. Finally, we show that native genomic loci can be decoded by efficient construction of barcode libraries assisted by clustered regularly interspaced short palindromic repeats and CRISPR-associated protein 9 (CRISPR/Cas9). Thus, Epi-Decoder is an effective strategy to identify and quantify in an unbiased and systematic manner the proteome of an individual genomic locus by DNA sequencing.

Fig 1. Outline of Epi-Decoder.

(A) Construction of the Epi-Decoder library: a library of strains encoding TAP-tagged proteins was crossed to a Barcoder library by using standard SGA procedures. (B) Experimental set-up of Epi-Decoder: colonies of an arrayed library are pooled in a flask, expanded, subjected to crosslinking, and used for a TAP-ChIP. The coimmunoprecipitated barcodes are amplified and counted by massive parallel sequencing. Barcodes of input DNA are counted to control for the presence of each barcode in the pool. Indexed input and ChIP samples of multiple pools can be combined into 1 master pool before sequencing. (C) Protein binding is measured by barcode counting. The Barcoder locus consists of a KanMX resistance marker transcribed by the AgTEF1 promoter flanked by 2 unique barcodes and replacing the HO gene. The BC_UP is promoter-proximal and the BC_DN terminator-proximal and close to (50 bp) an origin of replication (ARS404). BC_DN, downstream barcode; BC_UP, upstream barcode; ChIP, chromatin immunoprecipitation; SGA, synthetic genetic array; TAP, Tandem Affinity Purification.

Fig 2. Identifying chromatin binders with DNA barcode counting.

(A) Volcano plots showing the average log2 ChIP/input versus significance (p-value) calculated for 3,994 factors for BC_UP and 3,995 factors for BC_DN across 6 independent biological experiments. The red colour indicates factors with a Benjamini-Hochberg–adjusted p-value < 0.01. Green triangles represent the barcodes associated with the TAP-tagged histone proteins in the library. (B) Venn diagram indicating the number of factors that are significant (FC > 0; FDR < 0.01) at BC_UP, BC_DN, or both. (C) Bar plot showing the fraction of factors in 4 different categories that were determined by using GO annotations (see Methods). The first 2 bars show the fractions in the group of binders (at either BC_UP or BC_DN). The right bar shows the fraction in the background set (the entire library). (D) Density plot of GFP abundance using data from the Cyclops database (http://cyclops.ccbr.utoronto.ca/). The binders were defined as before (FC > 0; FDR < 0.01), and the categories are similar to those in panel C. The ribosome category was defined based on the CYC2008 database (http://wodaklab.org/cyc2008/). Histones are shown separately; they bind at the DNA and are highly abundant. The numbers indicate the size of each group. (E) ChIP-qPCR of selected TAP-tagged strains from the barcoded KanMX Epi-Decoder library, with specific primers in proximity to BC_UP and BC_DN. To compare the barcode counts with ChIP-qPCR signal, the samples were normalized to the Bar1-TAP signal (Bar1 is not expressed in these cells) before calculating ChIP/input. The average of 3 biological replicates is shown; the error bars indicate the SD. Rpl31A—a highly expressed ribosomal subunit frequently used in anchor away studies to deplete proteins from the nucleus—was used as a negative control and as a reference to determine binding specificity in the qPCR analysis. Rpo21 is the largest subunit of RNA polymerase II, Smc3 is a cohesin subunit, Ssa1 and Ssa2 are HSP70 chaperones, Top2 is topoisomerase II, and Cdc19/Pyk1 is pyruvate kinase. The p-values from a two-sided t test compared to Rpl13A are indicated by asterisks. Underlying data for Fig 2C and Fig 2E in S1 Data. BC_DN, downstream barcode; BC_UP, upstream barcode; ChIP, chromatin immunoprecipitation; FC, fold change; FDR, false discovery rate; GFP, green fluorescent protein; GO, gene ontology; IP, immunoprecipitation; qPCR, quantitative PCR; TAP, Tandem Affinity Purification.

Fig 3. Capturing the chromatin interactome with Epi-Decoder.

(A) Scatter plot of ChIP/input values, which can be interpreted as binding scores, of BC_UP versus BC_DN. For some factors, only the BC_UP or BC_DN value was available. In these cases, the missing value was set to −1 in order to still visualize the remaining value. Factors are coloured based on known functions or complexes. Triangles represent histone proteins. The average values of 6 replicates are shown. (B) Enrichment plot for 3 categories: initiation (TFIID-E and GRFs), termination (Cleavage and Polyadenylation and THO complex), and replication (ORC and MCM) factors. Binders were ranked based on the BC_UP/BC_DN ratio. The top part shows the running sum enrichment for each category. Initiation factors were significantly enriched at BC_UP (p = 3.18⁻³) and termination and replication at BC_DN (p = 2.03⁻⁷ and p = 1.06⁻⁵). The bottom lines indicate the factors represented in the categories. (C) Illustration of the different protein complexes that Epi-Decoder identified at the barcoded KanMX gene at the HO locus. BC_DN, downstream barcode; BC_UP, upstream barcode; ChIP, chromatin immunoprecipitation; GRF, general regulatory factor; IP, immunoprecipitation; MCM, minichromosome maintenance; ORC, origin recognition complex.

Fig 4. Chromatin rewiring upon HU treatment.

(A) DNA regions replicated in the HU arrest were determined by DNA copy number assessment. The DNA coverage in bins of 200 bp is plotted across the genome for untreated and HU-treated samples. The lower panel shows a zoom-in of chromosome IV, which contains the barcoded locus on the left arm (indicated by the arrow). Replication timing in the absence of HU was obtained from Alvino and colleagues [36]. The percentage of replicated DNA is plotted for each bin at 10, 12.5, 15, 17.5, 25, and 40 minutes after release from a G1 arrest, as indicated by the shades of grey. (B) RT-qPCR shows relative mRNA expression of KanMX/HphMX in untreated and HU-treated samples. The untreated samples were set to 1. The average of 3 biological replicates is shown; error bars indicate SD. (C) Volcano plots showing the ratio of binding in HU-treated/untreated for factors that were significantly enriched in either one or both of the conditions. The coloured dots are factors with significantly different binding scores (FDR < 0.05). BC_DN, downstream barcode; BC_UP, upstream barcode; FDR, false discovery rate; HU, hydroxyurea; ORC, origin recognition complex; RT-qPCR, quantitative reverse transcription PCR.

Fig 5. Application of Epi-Decoder to a native genomic locus.

(A) Procedure to generate a custom Epi-Decoder collection. A selection marker (NatMX) was inserted in proximity to the locus of interest, downstream of the to-be-barcoded ADE2 gene. Subsequently, a Cas9- and gRNA-expressing vector was introduced in this strain together with a repair template that is composed of a 15 bp random barcode-sequence (BC) flanked by 40 bp arms with homology to the gRNA target site, here at the 5’ end of the ADE2 coding sequence (details see S5A Fig). Transformants, the vast majority of which represent barcoded clones, were transferred to an arrayed format. Rows and columns were pooled for identifying barcodes and their location on the array by high-throughput sequencing. The barcoder library was mated and crossed with the TAP-tagged protein library (see Fig 1) to generate heterozygous diploids and then haploids, after which Epi-Decoder screens could be performed. For diploid screens, the insertion of a selectable marker genetically linked to the barcode can be omitted. (B) Epi-Decoder results of BC_5’-ADE2 (n = 2, average) in diploid cells were compared to BC_UP in haploid cells (Fig 2). Binding scores were visualized in a scatterplot with histones shown as triangles and classes of proteins coloured: TFIIH, RNAPII, DSIF. BC, barcode; BC_5’-ADE2, barcode in 5’ region of ADE2; BC_UP, upstream barcode at HO locus; Cas9, CRISPR-associated protein 9; gRNA, guide RNA; IP, immunoprecipitation; TAP, Tandem Affinity Purification.