DNA sequence features in the establishing of H3K27ac

The presence of H3K27me3 has been demonstrated to correlate with the CpG content. In this work, we tested whether H3K27ac has similar sequence preferences. We performed a translocation of DNA sequences with various properties into a beta-globin locus to control for the local chromatin environment. Our results suggest that in contrast to H3K27me3, H3K27ac gain is unlikely affected by the CpG content of the underlying DNA sequence, while extremely high GC-content might contribute to the gain of the H3K27ac.

Modification of histone proteins is a key mechanism of epigenetic regulation. Histone modifications vary between cells in some genomic locations but not others. This observation raises the question to what extent histone modifications depend on the underlying nucleotide sequences. It has been reported that the attraction of PRC2 complex and consequent H3K27me3 is positively correlated with the local density of CpG dinucleotides 1 . More complex sequence patterns, such as transcription factor (TF) binding sites (TFBS), also affect the presence of histone modifications. For example, SUZ12, a member of the PRC2 complex, binds DNA in a sequence-specific manner and NRCF and ZBTB33 recruit histone deacetylase. The ENCODE project demonstrated a specific histone modification profile around binding sites of many TFs 2 .
Since the same lysine residue cannot be both methylated and acetylated, the presence of H3K27ac is negatively correlated with the presence of H3K27me3 3 . Although there are several pieces of evidence showing that H3K27me3 is established at least partially in a sequence-specific manner, it is unclear if H3K27ac -an antagonistic activator to a repressive H3K27me3 -shares any sequence specific patterns. A computational approach 4 predicted some TFBS to be linked to H3K27ac, yet the results were inconsistent and dependent on the a background set. In this work, we perform a direct experiment to test whether a specific genomic sequence is capable of recovering H3K27ac.

Cas9 target selection and plasmid design
We selected a site for Cas9 in the intergenic region of human beta-globin locus using services CCTop 5 (hg38) and Off-Spotter 6 (hg38). The chosen targeting sequence was CTTGTCCCTGCA GGGTATTA. Then we designed targeting oligonucleotides (Table 2) for pSpCas9(BB)-2A-Puro (pX459 plasmid (Addgene), a gift from Egor Prokhorchouck, E.P.). Oligonucleotides Bgl1Cas_F and Bgl1Cas_R were diluted to 10 µM each with annealing buffer (10 mM Tris, pH 7.5, 100 mM NaCl, 1 mM EDTA), heated for 5 minutes to 95°C and then slowly cooled down to room temperature. 500ng of pX459 was digested with 5U of BpiI (Thermo Scientific) for 1h at 37°C, heat inactivated, run on 1.5% agarose, linearized plasmid was cut out and purified with QIAquick Gel extraction kit (Qiagen, cat. no. 28706) and ligated with Bgl1Cas_F/Bgl1Cas_R duplex. The ligation product was transformed into E. coli Top-10 cells, one of the clones was chosen and the insert was confirmed by Sanger sequencing.
Below we refer to this plasmid as pX459-b1.

Plasmids design
Selection cassette design (HSV thymidine kinase, T2A peptide, NeoR in one frame) was performed with Ugene 7 . HSVtk and NeoR sequences were PCR amplified with Phusion polymerase (Thermo Scientific) from pBS246-neo/Tk (a gift from E.P., construction of this plasmid is described elsewhere 8 ) with primers HSVtk_F, HSVtk_R, and G418_F, G418_R, correspondingly. T2A peptide coding sequence, corresponding to amino acid sequence GSGEGRGSLLTCGDVEENPGP, was synthesized by hybridization of oligonucleotides T2A(+) and T2A(-) in annealing buffer and consequent treatment of hybridized duplex with T4 polymerase (Thermo Scientific) and gel purified. The NeoR fragment was digested with XbaI (Thermo Scientific), the T2A fragment was digested with NheI (Thermo Scientific), then these fragments were ligated and the fragment of expected size (881bp) was gel purified. This T2A-NeoR fragment was digested with XhoI (Thermo Scientific), HSVtk fragment was digested with SalI (Thermo Scientific), fragments were ligated and the fragment of predicted length was gel purified again. This HSVtk-T2A-NeoR fragment was double digested with EcoRI (Thermo Scientific) and BshTI (Thermo Scientific) and ligated with pX459-b1 double-digested with EcoRI and BshTI. This step gave the plasmid with full-length selection cassette. Below we refer to this plasmid as pHSVtk-T2A-NeoR. Then the pHSVtk-T2A-NeoR plasmid was double-digested with XbaI and Acc65I (Thermo Scientific), gel purified and ligated with hybridized AdaptUp(+)-AdaptUp(-) duplex. A successful insert was verified by digestion of newly introduced restriction sites: SalI and BamHI. We refer to this plasmid as pAdaptUp-HSVtk-T2A-NeoR. This plasmid was digested with NotI (Thermo Scientific), treated with FastAP (Thermo Scientific), gel purified and ligated with AdaptDown(+)-AdaptDown(-) duplex to introduce a 2xBpiI site that generates half-sites for BclI and XhoI after cleavage. The resulting plasmid was transformed into E. coli Top-10 cells. We refer to this plasmid as pAdaptUp-HSVtk-T2A-NeoR-AdaptDown. To obtain LoxPflanked sequences, we used a pBK-CMV-derived plasmid with a modified multiple cloning site containing restriction sites for BclI, NheI, and XhoI. This plasmid was transformed into E. coli JM110 to eradicate Dam methylation, double digested with BclI and NheI, dephosphorylated, gel purified and ligated with LoxP(+)-LoxP(-) duplex, treated with T4 PNK. This plasmid was transformed into E. coli Top10 cells and referred to as pLoxP. To obtain homology regions that flank Cas9 cleavage site we PCR amplified them from Caki1 gDNA, using primer pairs Bgl1Up_F and Bgl1Up_R for an upstream fragment and Bgl1 Down_F and Bgl1Down_R for a downstream fragment. These PCR fragments were double digested with NheI and XhoI and ligated with pLoxP, double digested by the same sites and dephosphorylated. Ligation products were transformed into E. coli JM110, and purified plasmids were digested with BclI and XhoI to yield upstream and downstream fragments of DNA bearing LoxP site on its end. The upstream fragment was ligated with pAdaptUp-HSVtk-T2A-NeoR-AdaptDown double digested with SalI and BamHI to give pUp1L-HSVTK-T2A-NeoR-AdaptDown. The downstream fragment was ligated with pUp1L-HSVTK-T2A-NeoR-AdaptDown treated with BpiI and dephosphorylated. We refer to this plasmid as pUp1L-HSVTK-T2A-NeoR-L1Down.

Amendments from Version 1
We carefully checked all the names of the genes and plasmids as well as the rest of the text for typos and modified a few statements.

REVISED
(Thermo Scientific) in 2cm 2 wells following manufacturers instructions. After one week of Puromycin (3 µg/ml) selection, cells were split into 96-well plates and selected with 1mg/ml G418 for two weeks. Cells from successfully growing clones were split into two equal aliquots, one for growth and another for genomic DNA isolation. Clones were checked for the presence of the insert with a primer pair BGL1pcr_F -BGL1pcr_R surrounding the insert. One clone (referred to as Caki1-GcvS-G418R) with a homozygous insert was selected for the further experiments.

Construction of a recombination target
Plasmid pBK-CMV was digested with SacI and HpaI, blunted with T4 polymerase, self-ligated, transformed into E. coli Top10 and called pBCK-CMVdHpaI-SacI. LoxP(+) oligo was PNK treated and annealed with LoxP(-) to form a duplex. Bait(+) and Bait(-) oligos were annealed and ligated with hemi-phosphorylated LoxP duplex. Ligation products were resolved on 3% agarose gel, the longest fragment was excised and purified with QIAquick gel extraction kit, treated with T4 PNK and ligated to NheI treated and dephosphorylated pBCK-CMVdHpaI-SacI. The ligation product was transformed into E. coli Top10, we refer to the resulting plasmid as p1L-bait-L1. Ten sequences chosen to be inserted (Table 1) were PCR amplified from genomic DNA of Caki1 cells with primers (Table 2, rows 25 to 44). Amplicons were gel purified, diluted to a concentration of 100nM, treated with T4 PNK and ligated with a plasmid p1L-bait-L1 treated with Ecl136II and dephosphorylated. Then the library of ligation products was transformed into E. coli Top10, the plasmid library was purified using a plasmid mini kit (Evrogen, cat. no. BC021) and co-transfected with pBS598 EF1alpha-EGFPcre (Addgene) to Caki1-GcvS-G418R cells for Cre-mediated recombination exchange of the insert and the HSVTK-T2A-NeoR cassette. This step was performed in two independent replicates. After 3-day growth Ganciclovir (2 µM) was added to eliminate cells that did not undergo recombination. After selection for 10-14 days survived cells were grown to a subconfluent monolayer in 10cm dishes and then ChIP on H3K27Ac was performed.

Chromatin immunoprecipitation (ChIP)
ChIP was performed according to Abcam X-chip protocol with the following modification: we increased the number of washes in High Salt buffer from one to three times. Sonication was performed in PCR tubes (SSIbio, cat. no. 3245-00) on ice using Sonics Vibra-Cell VCX 130 with an eight-element probe (cat. no. 630-0602). Sonication setup was: 10s pulse, 20s pause, 75% power, total sonication time is 30min. We used 2 µg of Anti-H3K27Ac antibody (Abcam, ab4729) per ChIP. Biological replicates of targeted inserts of DNA sequences were processed independently. Control ChIP was performed under the same conditions with Caki1 cells.

Nested PCR
After ChIP, the DNA was pooled and 1ng of DNA was PCR amplified with a primer pair Bait-seq_F -Bait-seq_R. At this step we amplified all DNA sequences inserted between this primer pair. Then 10ng of PCR amplicons from the first step was used as a template for a PCR with a primers (Table 2, rows 25 to 44), each pair in an individual reaction tube.

Results and discussion
None of the GC-and CpG rich promoter regions, that were acetylated in their original genomic loci (rows 1-3 in Table 1 and lanes 1-3 in Figure 1) recovered H3K27ac after relocation to a foreign genomic context in the beta globin locus, suggesting that H3K27ac may not depend directly on such features. An alternative explanation is that some of the CpG dinucleotides became methylated in a foreign genomic environment preventing chromatin from gaining H3K27ac -an active chromantin mark. Surprisingly, two extremely GC-rich but CpG poor (and, therefore, unmethylated) sequences (rows 5, 6 in Table 1 and lanes 5, 7 in

Competing interests
No competing interests were disclosed.

Grant information
This work was supported by the Russian Science Foundation [15-14-30002].
The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.  The study is focused on question of relation of primary genomic sequences and H3K27 acetylation preferences. Original method, based on the translocation of the different CG% and CpG -based sequences to the same genomic locus, followed by H3K27ac analysis was employed. The results suggest that that the context is more important than GC or CpG properties of the genomic sequence itself. The results of certain interest for the wide pool of researcher working with epigenetic aspects. At the same time, I believe that involvement of more techniques, such as Chip-Seq would be suggested to confirm the results. Also, it is not clear from the text how many replicates were taken and how general is the conclusion, if, for example, different genomic loci (or different genome, for example extremely GC or AT-rich one) will be used.

If applicable, is the statistical analysis and its interpretation appropriate? Yes
Are all the source data underlying the results available to ensure full reproducibility? Partly

Are the conclusions drawn adequately supported by the results? Partly
No competing interests were disclosed. The study is focused on question of relation of primary genomic sequences and H3K27 acetylation preferences. Original method, based on the translocation of the different CG% and CpG -based sequences to the same genomic locus, followed by H3K27ac analysis was employed. The results suggest that that the context is more important than GC or CpG properties of the genomic sequence itself. The results of certain interest for the wide pool of researcher working with epigenetic aspects. A: We thank the reviewer for a careful reading of the manuscript and useful comments. Although we have to disagree that our results show the context to be more important than GC or CpG content of the sequence. In fact, our results suggest that GC content may contribute to the establishing of H3K27ac. To avoid further confusions we modified the abstract and the results accordingly to make this statement more clear. "Our results suggest that in contrast to H2K27me3, H3K27ac gain is unlikely affected by the CpG content of the underlying DNA sequence, while extremely high GC-content might contribute to the gain of the H3K27ac" Q: At the same time, I believe that involvement of more techniques, such as Chip-Seq would be suggested to confirm the results. A: We agree with the reviewer that a ChIP-seq may provide a more unbiased approach. Yet, a regular ChIP-seq experiment after the insertion seems to produce a lot of waste since only one region is expected to be changed. On the other hand a modified version of ChIP-seq experiment with multiple insertions into the same location may indeed be more unbiased and will hopefully increase the number of tested sequences by the order of magnitude making the conclusions more solid. We are working on this protocol now but we believe that this much more complicated experimental setup is beyond the scope of the current pilot paper. Q: Also, it is not clear from the text how many replicates were taken and how general is the conclusion, if, for example, different genomic loci (or different genome, for example extremely GC or AT-rich one) will be used. A: The insertion of 10 target sequences was performed done in two independent biological replicates, then every replicateate was CHiP'd and resulting DNA was pooled before nested PCR step. To avoid any further confusions the corresponding text was added to the Methods. Section "Construction of a recombination target" "This step was performed in two independent replicates." was added to the Section "Chromatin Immunoprecipitation (ChIP)" "Biological replicates of targeted insertion of DNA sequences were processed independently"" Section "Nested PCR" "the DNA was pooled" was added to section "Nested PCR" The beta-globin locus was chosen based on its lack of the chromatin modifications in the most of the adult cell types so the surrounding chromatin does not interfere with the inserted sequence and the gain of the modifications in the inserted fragment is determined by its sequence. We agree that additional loci lacking chromatin marks might be also tested, yet for now we focus on the modified ChIP-seq protocol to test many more inserted sequences. At this stage we are not generalizing our conclusions to other genomes with the extreme AT-or GC-contents of the genome since apart of these changes these organisms may also have a quite different H3K27ac-attracting machinery.
No competing interests were disclosed. Competing Interests: This study aimed to test whether H3K27 acetylation depends on the properties of the target sequence itself, or rather defined by the surrounding genomic context. To assess this question, authors translocated sequences with different CG% and CpG content, to the same beta globin locus, and then tested their H3K27ac in the new location.
As three tested highGC-highCpG sequences lost their H3K27ac in the new location, authors conclude that the context is more important that GC or CpG properties of the sequence itself. At the same time, two other sequences actually acquire H3K27ac in the same new location, and only one of those can be possibly explained by the antagonistic H3K27me3 mark.
Authors describe their methods with all the details which should be sufficient for reproducibility of their main results. I had, however, some minor technical questions regarding their plasmids construction procedure. For example: 1) From the sequence of T2A oligos they provide, peptide sequence seems to be PSPLPSPCLLTCGDVEENPGP, not GSGEGRGSLLTCGDVEENPGP. The latter one would be present in the original pX459 plasmid, but not in the constructed one. 2) To my understanding, annealing of T2A(+) and T2A(-) oligos will produce a hybrid with only 18bp double-stranded overlap -for further digestion with NheI, its single-stranded ends need to be built up with some polymerase; I don't see this step in the Methods. 3) I don't see XhoI site in the T2A-NeoR fragment, and don't see any other potential site compatible with SalI of HSVtk fragment. In addition, it is recommended that all the source plasmids used should be properly referred -either by their original publication, or by the sequence reference in any available repository (Genbank, Addgene, etc); referring as "pHSVTK-Neo (a gift from E.P.) " is not sufficient if the plasmid's sequence cannot be explicitly found by that name. Finally, authors should double-check their text for typos in the names of all constructs and genes (for ex., GAPGH), be sure to use all the titles exactly as they are originally defined (for ex., PX459 instead of pX-459, etc). All these issues, however, are minor and do not distort the main content of the paper.
From the technical point of view, the only question I had is about the final readout for H3K27ac: authors perform nested PCR followed by visualization of the PCR products on the gel. I am wondering this method was chosen and not qPCR (as more quantitative) or ChIP-seq (as more unbiased).
One of my main concerns is about the reproducibility of the findings: authors do not report any biological or technical replicates which makes the results potentially questionable.
Talking about the interpretation of the results, I am not sure how strong can be the conclusion based on only 10 tested sequences, and even more so, only one genomic context. Even with this limited number of tests, some results cannot be easily explained (for example, acquired H3K27ac for sample #5), so in future it would be necessary to expand the panel of both sequences and contexts.
Besides above-mentioned questions (hopefully authors can address at least some of them in the revised version), I think the paper is clearly of scientific interest and adds important piece of knowledge to its field.

Is the work clearly and accurately presented and does it cite the current literature? Yes
Is the study design appropriate and is the work technically sound? Yes

If applicable, is the statistical analysis and its interpretation appropriate? Not applicable
Are all the source data underlying the results available to ensure full reproducibility? Yes

Are the conclusions drawn adequately supported by the results? Partly
No competing interests were disclosed.

Competing Interests:
Referee Expertise: Cancer biomarkers, oncogenesis, signal transduction, miRNA regulation I have read this submission. I believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.
GAAT GCTAGC This sequence contains two restriction sites (XhoI and NheI). The sequence between restriction sites is translated as GSGEGRGSLLTCGDVEENPGP in 5'->3' direction, frame 1 ( ). https://web.expasy.org/cgi-bin/translate/dna2aa.cgi Q: To my understanding, annealing of T2A(+) and T2A(-) oligos will produce a hybrid with only 18bp double-stranded overlap -for further digestion with NheI, its single-stranded ends need to be built up with some polymerase; I don't see this step in the Methods. A: We thank the reviewer for pointing out this mistake. The typo in Methods section is corrected and enzyme name is changed from T4 PNK to T4 Polymerase. We modified the Methods section and added the following text: "T2A peptide coding sequence corresponding to amino acid sequence GSGEGRGSLLTCGDVEENPGP was synthesized by hybridization of oligonucleotides T2A(+) and T2A(-) in annealing buffer and consequent treatment of hybridized duplex with T4 polymerase (Thermo Scientific) followed by gel purification." Q: In addition, it is recommended that all the source plasmids used should be properly referredeither by their original publication, or by the sequence reference in any available repository (Genbank, Addgene, etc); referring as "pHSVTK-Neo (a gift from E.P.) " is not sufficient if the plasmid's sequence cannot be explicitly found by that name. Finally, authors should double-check their text for typos in the names of all constructs and genes (for ex., GAPGH), be sure to use all the titles exactly as they are originally defined (for ex., PX459 instead of pX-459, etc). All these issues, however, are minor and do not distort the main content of the paper. A: We carefully checked all the names of the genes and plasmids as well as the rest of the text for typos. The plasmid name pHSVTK-Neo is internal; this plasmid is refered as pBS246-neo/tk in the folowing paper: . The reference https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1317619/ describing a the plasmid pBS246-neo/tk is added to the main text: "HSVtk and NeoR sequences were PCR amplified with Phusion polymerase (Thermo Scientific) from plasmid pBS246-neo/Tk (a gift from E.P.), construction of this plasmid is described elsewhere: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1317619/" Q: From the technical point of view, the only question I had is about the final readout for H3K27ac: authors perform nested PCR followed by visualization of the PCR products on the gel. I am wondering this method was chosen and not qPCR (as more quantitative) or ChIP-seq (as more unbiased). A: For a pilot study we decided to focus on making sure that the insertion of the selected fragments into a foreign environment (beta-globin locus) did take place as a proof of principle therefore we have chosen to perform nested PCR. In this regards qPCR will not make the conclusions more reliable. Performing a regular ChIP-seq experiment after the insertion seems like a bit of overkill since histone mark only in one region is expected to be changed. On the other hand a modified version of ChIP-seq experiment with multiple insertions may indeed be more unbiased. Yet, we believe that this much more complicated experimental setup is beyond the scope of the current believe that this much more complicated experimental setup is beyond the scope of the current paper.
Q: One of my main concerns is about the reproducibility of the findings: authors do not report any biological or technical replicates which makes the results potentially questionable. A: The insertion of 10 target sequences was performed in two independent biological replicates, then every replicate was ChIP'd independently and resulting DNA was pooled before nested PCR step. To avoid any further confusions the corresponding text was added to the Methods: Section "Construction of a recombination target" "This step was performed in two independent replicates." was added to the Section "Chromatin Immuniprecipitation (ChIP)" "Biological replicates of targeted insertion of DNA sequences were processed independently" Section "Nested PCR" "the DNA was pooled" Q: Talking about the interpretation of the results, I am not sure how strong can be the conclusion based on only 10 tested sequences, and even more so, only one genomic context. Even with this limited number of tests, some results cannot be easily explained (for example, acquired H3K27ac for sample #5), so in future it would be necessary to expand the panel of both sequences and contexts. A: We agree with the reviewer that some of the results (seq #5) are puzzling although reproducible since we performed all the tests in two biological replicates. At this stage we decided to report the results of the pilot project and focus on the performing a more complex experiment based on ChIP-seq as mentioned above which hopefully will increase the number of tested sequences by the order of magnitude. Q: Besides above-mentioned questions (hopefully authors can address at least some of them in the revised version), I think the paper is clearly of scientific interest and adds important piece of knowledge to its field. A: Again, we are grateful to the reviewer for the high evaluation of our work, careful reading of the manuscript and useful comments.
No competing interests were disclosed. Competing Interests: