Enhanced and controlled chromatin extraction from FFPE tissues and the application to ChIP-seq

Tissue-level cross-link reversal dramatically increases soluble chromatin from FFPE tissues

The major challenge for chromatin extraction from FFPE tissues is low chromatin yield due to high cross-linking level of the tissue sample during preparation and long storage time of FFPE samples. We hypothesized reducing cross-linking before chromatin preparation would increase chromatin yield from FFPE tissues. To test this hypothesis, we introduced a tissue-level cross-link reversal step by heating before chromatin preparation called Chrom-EXPE (Fig. 1a). As incubation of cross-linked chromatin at 65 °C is routinely utilized in ChIP assay, we first tested tissue-level cross-link reversal with overnight incubation at 65 °C. As a control and for comparison, we tested a commercial kit (Active Motif) for chromatin extraction from FFPE tissues following the manufacturer’s instructions. After tissue-level cross-link reversal, MNase digestion and sonication was utilized to extract chromatin. Notably, the insoluble pellet after centrifugation is much smaller with samples processed with tissue-level cross-link reversal compared with the samples processed with the commercial kit or without overnight incubation. The soluble chromatin yield was calculated from purified DNA from soluble and insoluble pellet fractions after extensive cross-link reversal as described in conventional ChIP protocols. Surprisingly, tissue-level cross-link reversal dramatically increased chromatin yield in FFPE samples from mouse spleen, kidney, and liver (Fig. 1b). The soluble chromatin yields achieved 71.6, 85.4 and 90.2% for liver, kidney, and spleen, respectively, by Chrom-EX PE, while the soluble chromatin yields reached 1.15, 6.48 and 2.54% with the Active Motif commercial kit and 6.37, 15.45, 10.95% without 65 °C incubation for the same tissues. These results demonstrate that Chrom-EX PE increases the effectiveness of chromatin preparation from FFPE samples and can achieve chromatin yields up to 90% of the yields typically obtained using conventional techniques from fresh or frozen tissues.

To further investigate chromatin yields in various incubation temperatures, we performed the experiment at four different temperatures (25 °C, 45 °C, 55 °C, and 65 °C) using FFPE samples from liver tissue. To accurately measure the impacts of incubation temperature on chromatin yield, we isolated the chromatin by sonication but no MNase treatment after tissue-level cross-linking reversal. The yield of soluble chromatin gradually increases with temperature in 45 – 65 °C range compared with the yield from room temperature (Additional file 1), suggesting different level of cross-linking reversal contributes to the increase of soluble chromatin.

Chrom-EX PE enables chromatin to be prepared in a controlled manner

We checked chromatin patterns generated in different incubation temperatures for tissue-level cross-link reversal. Chromatin was prepared from mouse liver FFPE tissues by Chrom-EX PE in the range of 25 to 75 °C, and DNA was purified from the resulting soluble chromatin and analyzed by the Fragment Analyzer (Fig. 1c). In general, low temperatures generated larger DNA fragments over 1 kb and DNA sizes gradually decreased to around mononucleosomal size with the increasing incubation temperatures. In the range of 25–37 °C, a majority of soluble chromatin is larger than 1 kb and a very small fraction of DNA fragments around 50–100 bp. The temperature range of 45–55 °C produced a nucleosomal DNA profile that is typically observed when frozen tissues and cell lines are treated with MNase. The majority of DNA was mononucleosomal size in the range of 60–65 °C, and DNA size is smaller than mononucleosomal DNA at 75 °C. These observations indicated that Chrom-EX PE is able to prepare chromatin in a controlled manner from FFPE tissues. It is noteworthy that the sizes of most soluble chromatin in the range of 45–65 °C are about 100–500 bp, further increasing usable chromatin in the assays and well suited for the requirement of next-generation sequencing library preparation for genome-wide chromatin-based epigenetic assays.

Chromatin yields from mouse FFPE tissues by Chrom-EX PE

The overall experimental quality of chromatin-based epigenetic assays is typically dependent on the total cellularity but not the weight of tissue sample. As tissues show different cellularity, we next sought to address chromatin yield from different FFPE tissues by Chrom-EX PE in the optimized chromatin stabilization buffer. The tissue volume is measured in the FFPE block, and two 20 μm sections were processed by Chrom-EX PE at 65 °C condition. The DNA amount purified from soluble fraction was measured and the yield was calculated (Fig. 1d).

Chromatin generated by Chrom-EX PE is compatible with ChIP assays

To investigate whether the chromatin generated by Chrom-EX PE is compatible with chromatin-based epigenetic assays, we elected to use chromatin immunoprecipitation assay, which is the most often used methodology in chromatin-based epigenetic analyses. The chromatin prepared by Chrom-EX PE at 65 °C from four different mouse liver FFPE tissue was incubated with anti-H3K4me3 or anti-H3K27me3 antibodies and further processed following published methods [13, 14]. The ChIP products were analyzed by real-time PCR in a transcriptionally active region (GAPDH-TSS, positive for H3K4me3), in a developmentally repressed region (T-TSS, positive for H3K27me3), and in an intergenic region (negative for both H3K4me3 and H3K27me3) (Fig. 2a and Additional file 2). The enrichment of H3K4me3 mark is observed only in transcriptionally active TSS, and H3K27me3 mark is observed only in developmentally repressed locus in all four FFPE tissues. The DNA size of chromatin input from 4 FFPE samples is in the range of 100–200 bp without DNAs larger than 1 kb (Additional file 3), consistent with the previous observation (Fig. 1c) and showing consistency of the Chrom-EX PE approach. These results demonstrate that the chromatin prepared from FFPE tissues using Chrom-EXPE is compatible with a targeted ChIP-qPCR approach. Next, we further investigated the compatibility of Chrom-EX PE with next-generation sequencing library preparation for genome-wide ChIP-seq. The chromatin input isolated by Chrom-EX PE from mouse liver FFPE tissue was subjected to ChIP for histone marks (H3K27Ac and H3K4me3, H3K27me3) and a chromatin-associated protein (RNA Polymerase II, Pol II). To compare the performance of Chrom-EX PE, we also generated ChIP-seq data from frozen tissues originating from the same mouse using published ChIP-seq protocol [13, 14]. To evaluate the reproducibility of Chrom-EX PE, we performed the experiment in two replicates. The purified ChIP DNA was subjected to library preparation as published [13] and the library DNAs were analyzed by the Fragment Analyzer (Additional file 4). The libraries were sequenced on Illumina HiSeq 2000 platform. Peaks were called using the MACS2 algorithm [15] for H3K4me3 and H3K27Ac and SICER [16] for H3K27me3 and RNA Pol II at FDR < =1%. The difference of overall enrichment is variable in sample types per mark (Additional file 5). However, peak patterns visualized in the Integrative Genome Viewer [17, 18] were highly similar among the datasets from FFPE liver tissue by Chrom-EX PE and those from frozen liver tissues by published ChIP-seq protocol (Fig. 2b). As expected [19, 20], H3K4me3 and H3K27Ac peaks are primarily located at TSS of active genes, whereas H3K27me3 peaks are distributed over PRC2-repressed genes such as MYT1. These results strongly suggest that the chromatin generated by Chrom-EX PE technology is compatible with ChIP-seq approach.

Highly correlated and consistent ChIP-seq data are generated from FFPE tissue by Chrom-EX PE

The ChIP-seq data for each mark was generated from two 20-μm FFPE-tissue sections from mouse liver and spleen using Chrom-EX PE technology. To compare the performance of Chrom-EX PE, we also generated ChIP-seq data from frozen tissues originating from the same mouse. The mapping details of all ChIP-seq libraries are summarized (Additional file 6) [21]. In general, we did not observe clear differences in % uniquely mapped pairs, library complexity, and % duplicates between libraries generated from FFPE and frozen tissues. Some libraries generated by Chrom-EX PE showed higher % duplicate, low complexity, and low % unique pairs but other libraries generated in the independent experiment showed the expected range of QC results. We believe higher % duplicate, low complexity, and low % unique pairs in some of ChIP-seq data observed in FFPE samples is caused by over-amplification of the library [22] but not by Chrom-EX PE technology. Importantly, the total peak numbers for these libraries still achieve the requirement for the data analysis.

To investigate the experimental consistency of Chrom-EX PE technology, we performed Pearson correlation analysis between the datasets generated by Chrom-EX PE from experimental replicates. We observed uniformly high correlation for three histone marks with Pearson correlation coefficients ranging between 0.898 and 0.991 (p < 0.001) (Fig. 3a). The detailed peak numbers and overlapping rate in experimental replicates are shown for FFPE samples (Table 1) and for frozen tissues (Additional file 7). We calculated the overlap rate based on the dataset with fewer peaks to investigate how many peaks are observed in the same locations in two replicates. For H3K27Ac mark, about 92,000 peaks were detected with 91% overlap rate in repeats. About 48,000 peaks were detected with 84% overlap rate for H3K4me3 mark and about 20,000 peaks were detected with 64% overlap for H3K27me3 mark. These results indicate that ChIP-seq data generated by Chrom-EX PE is highly reproducible and consistent.

Mark	Total Peaks 1	Total Peaks 2	% Overlap
H3K27ac	128,624	92,165	90.91%
H3K4me3	56,688	48,129	83.90%
H3K27me3	19,765	23,234	64.14%

To determine whether Chrom-EX PE technology is applicable to large and non-histone protein or chromatin complexes, we performed ChIP-seq for RNA polymerase II (RNAP II) in two pairs of FFPE and frozen mouse liver tissues. As expected, RNA Pol II peaks shown in the IGV browser overlap with H3K4me3 marks in TSSs of actively transcribed genes (Fig. 2b), and peak patterns are similar among the datasets generated from FFPE and frozen liver tissues. Pearson correlation coefficients range from 0.930 to 0.945 (p < 0.001) in matched sample pairs (Fig. 3d). The detailed peak numbers and overlapping rate in match pairs are shown in Table 2. About 12,000 and 16,000 RNA Pol II peaks were detected in two different FFPE tissues. These results indicate that Chrom-EX PE is applicable to at least some non-histone protein or chromatin complex in archived FFPE tissues.

Animals and tissues

Frozen mouse tissues were collected from healthy mice under the Mayo Clinic IRCUC A00001233–16. The tissues were excised and divided. One piece from each pair was flash-frozen in liquid nitrogen and stored at − 80 °C and the other was fixed in 20-fold excess volume of 10% neutral phosphate-buffered formalin (Leica Biosystems, Winnipeg, MB) for 48 h. Fixed tissue was dehydrated and embedded in paraffin using an automated tissue processor. FFPE blocks were stored at RT for at least 2 years. These samples were used to establish and optimize Chrom-EX PE protocol.

Development of Chrom-EX PE, chromatin isolation from FFPE tissues

Two 20-μm sections from the FFPE blocks derived from mouse tissues were deparaffinized with xylene substitute and were rehydrated by progressively increasing percentage of water (95/5, 70/30, 50/50, 20/80). During the progressive dehydration with different ethanol concentrations, the tissues were resuspended in 0.5 ml of chromatin stabilization buffer (10 mM Tris-HCl, pH 7.5, 10 mM NaCl, 10 mM EDTA, 0.5% Triton X-100, 0.1% Sodium deoxycholate, 20% EtOH, proteinase inhibitor cocktails) and incubated in different temperatures ranging 25 °C to 75 °C overnight. As a control, the tissues were processed without overnight incubation. After the centrifugation (21,130 x g for 5 min), the supernatant was removed. And 0.25 ml of cell lysis buffer (10 mM Tris HCl, pH 7.5, 10 mM NaCl, 0.5% IGEPAL) was added and incubated on ice for 10 min. The lysates were washed with MNase digestion buffer (20 mM Tris-HCl, pH 7.5, 15 mM NaCl, 60 mM KCl, 1 mM CaCl₂) and were incubated in the fresh 150 μL MNase digestion buffer containing proteinase inhibitor cocktails in the presence of 10 gel units of MNase (NEB, Cat.# M0247S) at 37 °C for 20 min with continuous mixing in thermal mixer. After adding the same volume of sonication buffer (100 mM Tris-HCl, pH 8.1, 20 mM EDTA, 200 mM NaCl, 2% Triton X-100, 0.2% sodium deoxycholate), the lysates were sonicated for 15 min (30 s on / 30 s off) using Bioruptor Twin (UCD-400) (Diagenode, Inc., Denville, NJ) and centrifuged at 21,130 x g for 10 min. The supernatant was transferred in a new tube. The pellet was resuspended in 100 ul of 1X FFPE Stop buffer 1 (50 mM Tris-HCl, pH 8.1, 10 mM EDTA, 100 mM NaCl, 1% Triton X-100, 0.1% Sodium deoxycholate, 0.05% SDS) and was subjected to sonication for 15 min (30 s on / 30 s off), and centrifuged at 21,130 x g for 10 min. The supernatant was collected into the previously saved fraction. The pellet was resuspended in 100 ul of 1X FFPE Stop buffer 2 (50 mM Tris-HCl, pH 8.1, 10 mM EDTA, 100 mM NaCl, 1% Triton X-100, 0.1% Sodium deoxycholate, 0.1% SDS) and was subjected to sonication for 15 min (30 s on / 30 s off), and centrifuged at 21,130 x g for 10 min. The supernatant was collected into the previously saved fraction and the combined soluble fraction was served as soluble chromatin. The equal volume of soluble chromatin was mixed with 2X ChIP elution buffer (20 mM Tris-HCl, pH 8.0, 20 mM EDTA, 300 mM NaCl, 10 mM DTT, 2% SDS) and incubated at 65 °C overnight. The pellet was resuspended in 1X ChIP elution buffer and incubated at 65 °C overnight. DNAs were purified using Min-Elute PCR purification kit after the treatment of RNase A and proteinase K. DNA amount was measured by the Qubit dsDNA High Sensitivity assay (Invitrogen, Q32851), and DNA size was analyzed by the Fragment Analyzer (Advanced Analytical Technologies; AATI; Ankeny, IA) using the High Sensitivity NGS Fragment Analysis Kit (Cat. #DNF-486). For the comparison of soluble chromatin yield, the ChIP-IT® FFPE Chromatin Preparation Kit (Active Motif, Cat # 53030) was utilized according to the manufacture’s instructions.

Determination of chromatin yield from mouse FFPE tissues

The tissue image in FFPE block was captured every 5 cut, and the surface area of tissue was calculated by the NIH Image J program to determine the tissue size in the cuts. Two 20-μm sections were deparaffinized with xylene substitute and were rehydrated by progressively increasing percentage of water in ethanol up to 20%. The tissues were resuspended in 0.5 ml of chromatin stabilization buffer and incubated at 65 °C overnight. The tissues were processed as described above. Purified DNA was measured by the Qubit assay. The chromatin amount from each FFPE tissue is calculated by DNA amount per tissue size.

ChIP-seq

Frozen tissues (25 mg) were homogenized for 30 s in PBS using tissue grinder (ACTGene, ACT-AG 3080). Homogenized tissues were cross-linked to final 1% formaldehyde, quenched with 125 mM glycine, and washed with TBS. The fixed homogenates were resuspended in cell lysis buffer and incubated on ice for 10 min. The lysates were washed with MNase digestion buffer and were incubated in the fresh 500 μL MNase digestion buffer containing proteinase inhibitor cocktails in the presence of 1000 gel units of MNase at 37 °C for 20 min with continuous mixing in thermal mixer. After adding the same volume of sonication buffer, the lysates were sonicated for 15 min (30 s on / 30 s off) using Bioruptor Twin and centrifuged at 21,130 x g for 10 min. The supernatants were served as chromatin input. For FFPE tissues, chromatin input was prepared from two 20-μm sections as described in the development of Chrom-EX PE. The chromatin input was incubated with anti-H3K4me3 (abcam ab8580, lot GR188707–1), anti- RNA pol II antibodies (Bethyl A300–653, lot 3), anti-H3K27Ac (CST 8173 BC) or anti-H3K27me3 (CST, 9733 s, lot 8) overnight. After adding 10–30 μL of protein G-magnetic beads, the reactions were further incubated for 3 h. The beads were extensively washed with ChIP buffer (50 mM Tris-HCl, pH 8.1, 10 mM EDTA, 100 mM NaCl, 1% Triton X-100, 0.1% sodium deoxycholate), high salt buffer (50 mM Tris-HCl, pH 8.1, 10 mM EDTA, 500 mM NaCl, 1% Triton X-100, 0.1% sodium deoxycholate), LiCl₂ buffer (10 mM Tris-HCl, pH 8.0, 0.25 M LiCl₂, 0.5% NP-40, 0.5% Sodium deoxycholate, 1 mM EDTA), and TE buffer. Bound chromatins were eluted and reverse-crosslinked at 65 °C overnight. DNAs were purified using Min-Elute PCR purification kit after the treatment of RNase A and proteinase K. ChIP enrichment was validated by performing qPCR in the genomic loci targeting the transcription start sites of an active or inactive gene and an intergenic region. ChIP-seq libraries were prepared using the Ovation ultralow DR Multiplex kit (NuGEN, San Carlos, CA) or the ThruPLEX® DNA-seq Kit V2 (Rubicon Genomics, Ann Arbor, MI) according to the manufacture’s instructions. The ChIP-seq libraries were sequenced to 51 base pairs from both ends on an Illumina HiSeq 2000 instrument in the Mayo Clinic Center for Individualized Medicine Medical Genomics Facility.

Real-time PCR analysis

Real-time PCR analysis was performed using SYBR Green universal PCR mixes (Bio-Rad). The following primer sequences were used in the experiments: H3K4me3- and RNA pol II-positive control locus: mGAPDH-F: 5′- CTCATCCCCGCAAAGGCGGA -3′, mGAPDH-R: 5′- TCGGACCTGGCGATGGCTCG-3′. H3K27me3-positive control locus: mT1-F: 5′- GAGACGCCGATCCGCCGAAG -3′, mT1-R: 5′- ACTCTCCACTCCCACGCGCT-3′. H3K4me3-, RNA pol II-, and H3K27me3-negative control locus: mIntergenic-F: 5′-CCTGCTGCCTTGTCTCTCTC -3′, mIntergenic-R: 5′-ATGGCCTAGGGATTCCAGCA -3′.

Mapping and analysis of ChIP-seq data

Raw sequencing reads were processed and analyzed using the HiChIP pipeline [21] to obtain visualization files and a list of peaks. Briefly, paired-end reads were mapped to the mouse reference genome (mm10) by BWA [23] with default settings, and only pairs with at least one of the ends being uniquely mapped were retained for further analysis. Duplicates were removed using the Picard tool set. (https://broadinstitute.github.io/picard/). Peaks were called using the MACS2 algorithm and SICER at FDR < 1%. Fragment size was calculated from properly mapped read pairs. Pearson correlation analysis was performed by our in-house scripts where all datasets were randomly downsized to 25 million pairs of reads. In brief, the whole genome was divided into 5-kb bins for H3K4me3 and H3K27Ac and 100-kb bins for H3K27me3 and RNA Pol II, and the number of mapped reads in each bin was calculated, in which genomic regions that have zero or missing values in all samples are excluded. The read count per bin was normalized to 25 million mapped reads (at least one end uniquely mapped and duplicates removed), or to 25 million mapped reads from non-peak regions (total - reads mapped to peak regions). The normalized counts by logarithm log2 (normalized count + 1) were used for pairwise correlation analysis with Pearson coefficient. Here, 1 is a pseudo-count to avoid an undefined error of logarithm of zero. We also profiled the average signal levels across peak center +/− 2 kb (H3K4me3 and H3K27Ac) and peak region +/− 2 kb (H3K27me3 and RNAPII) using ngs.plot [24]. For each pair of samples, peaks from the two peak lists were first merged if they were within 1 bp (H3K4me3 and H3K27Ac) or 200 bp of each other (H3K27me3 and RNAPII). For each merged peak, a single original peak with the lowest FDR was selected as the representative and used in the plot.