Identification of Proteins Associated with an IFNγ-Responsive Promoter by a Retroviral Expression System for enChIP Using CRISPR

Isolation of specific genomic regions retaining molecular interactions is essential for comprehensive identification of molecules associated with the genomic regions. Recently, we developed the engineered DNA-binding molecule-mediated chromatin immunoprecipitation (enChIP) technology for purification of specific genomic regions. Here, we developed a retroviral expression system for enChIP using CRISPR. We showed that the target genomic locus can be purified with high efficiency by using this system. We also showed that contamination of potential off-target sites is negligible by using this system if the guide RNA (gRNA) for the target site has a sufficiently long unique sequence in its seed sequence. enChIP combined with stable isotope labeling using amino acids in cell culture (SILAC) analysis identified proteins whose association with the interferon (IFN) regulatory factor-1 (IRF-1) promoter region increases in response to IFNγ stimulation. The list of the associated proteins contained many novel proteins in the context of IFNγ-induced gene expression as well as proteins related to histone deacetylase complexes whose involvement has been suggested in IFNγ-mediated gene expression. Finally, we confirmed IFNγ-induced increased association of the identified proteins with the IRF-1 promoter by ChIP. Thus, our results showed that the retroviral enChIP system using CRISPR would be useful for biochemical analysis of genome functions including transcription and epigenetic regulation.


Introduction
A comprehensive understanding of the mechanisms behind genome functions such as transcription and epigenetic regulation requires the identification of the molecules that bind to the genomic regions of interest in vivo.We previously developed the locus-specific chromatin immunoprecipitation (ChIP) technologies consisting of insertional ChIP (iChIP) [1][2][3][4][5] and engineered DNAbinding molecule-mediated ChIP (enChIP) [6,7] for purification of specific genomic regions to identify their associated molecules.In enChIP, a tagged engineered DNA-binding molecule is expressed into the cell to be analyzed so that it recogizes an endogenous target DNA sequence.Subsequently, the target genomic region is subjected to affinity-purification such as immunoprecipitation with an antibody (Ab) against the tag(s).We showed that the clustered regularly interspaced short palindromic repeats (CRISPR) system [8][9][10][11][12][13][14][15][16][17][18][19][20][21] combined with enChIP efficiently isolates specific genomic regions for identification of their associated proteins [6].In this form of enChIP, specific genomic regions are immunoprecipitated with an Ab against a tag(s), which is fused to a catalytically inactive form of Cas9 (dCas9) plus guide RNA (gRNA) interacting with an endogenous DNA sequence in the genomic regions (Figure S1).Because it is easy to generate gRNA targeting specific genomic regions, enChIP using the CRISPR system is a convenient way to perform enChIP analysis.In our previous paper, we used a transient transfection approach to express the CRISPR components [6].The system would be applicable to those cell lines with high transfection efficiency.However, the transient transfection approach might not work for cells with low transfection efficiency.In those cases, stable expression of the CRISPR components would be necessary.
Here, we developed a retroviral expression system for enChIP using CRISPR.Stable isotope labeling using amino acids in cell culture (SILAC) [22] combined with enChIP (enChIP-SILAC) identified proteins induced to interact with the interferon (IFN) regulatory factor-1 (IRF-1) promoter region in response to IFNc stimulation.The retroviral expression system for enChIP using CRISPR would be useful for biochemical analysis of genome functions such as transcription, epigenetic regulation, genomic imprinting, and X chromosome inactivation.

Generation of a retroviral expression system for enChIP using CRISPR
To generate cells stably expressing the components of enChIP using CRISPR more easily and quickly, we developed a retroviral system to express 3xFLAG-dCas9 (dCas9 tagged with the 3xFLAG tag and fused with a nuclear localization signal (NLS)) [6] and gRNA.The coding sequence of 3xFLAG-dCas9 was inserted into pMXs [23]-derived retroviral expression vectors retaining various selection markers (Table 1).In addition, pSIR [24][25][26]-derived self-inactivating retroviral vectors with various selection markers were developed to express gRNA (Table 2).gBlock, an expression unit of gRNA, can be inserted into the multiple cloning sites of these vectors.To target the promoter region of human IRF-1 gene [27], the gBlock of gRNA-hIRF-1 #12 [6] was inserted into pSIR to generate gRNA-hIRF-1 #12/ pSIR.
To examine if the system works, 3xFLAG-dCas9/pMXs-puro was transduced into a human fibrosarcoma cell line, HT1080.After puromycin selection, expression of 3xFLAG-dCas9 was confirmed by immunoblot analysis with anti-FLAG Ab (Figure 1A) (the full-length images with size markers are shown in Figure S2).Subsequently, gRNA-hIRF-1 #12/pSIR was transduced into the HT1080 cells expressing 3xFLAG-dCas9.Cells expressing the gRNA were selected with G418.

Yield of enChIP for the target site and potential off-target sites
Next, we examined yield of enChIP for the target IRF-1 promoter locus.The cells expressing 3xFLAG-dCas9 and gRNA-hIRF-1 #12 were crosslinked with formaldehyde, and crosslinked chromatin was fragmented by sonication.Complexes containing 3xFLAG-dCas9 and gRNA-hIRF-1 #12 were immunoprecipitated with anti-FLAG Ab.Real-time PCR showed that around 10% of input genomic DNA was immunoprecipitated for the target IRF-1 promoter locus (Figure 1B).This yield was comparable with that observed in 293T cells transiently transfected with plasmids expressing 3xFLAG-dCas9 and gRNA-hIRF-1 #12 [6].We also tested the retroviral enChIP system using CRISPR for a human leukemia cell line, K562.The IRF-1 promoter region was also specifically isolated from K562-derived cells (Figure S3).These results indicated that efficient purification of target genomic regions is feasible by using the retroviral expression system for enChIP using CRISPR.
Next, we examined yield for potential off-target sites.CRISPR tolerates mismatches in the 59 region of target sites but not in the Protospacer Adjacent Motif (PAM) sequence and the seed sequence 59 proximal to PAM [10].No other site in the human genome contains sequences identical to 16-base of the seed sequence of gRNA-hIRF-1 #12 including the PAM sequence.Sites in chromosomes 11, 14, and 17 have the identical 15-base sequences in their seed sequences including PAM as well as mismatches in the 59 side of the identical 15-base sequences (Figure 1B).Sites in chromosomes 1 and 22, which are the most similar to the target site in the human genome, have two-base mismatches, one of which is present in the seed sequences near PAM.As shown in Figure 1B, yield for the potential off-target sites was marginal and comparable to that for the irrelevant Sox2 locus, suggesting that contamination of potential off-target sites can be minimal when target sites have more than a 16-base long unique sequence in the seed sequence including PAM.

Induction of expression of the IRF-1 gene in the presence of 3xFLAG-dCas9 and gRNA
Binding of the CRISPR complexes may interfere with gene expression (CRISPRi) [21].To examine whether IRF-1 gene expression might be abrogated by the binding of 3xFLAG-dCas9 and the gRNA targeting the IRF-1 promoter, we performed immunoblot analysis of the IRF-1 protein after stimulation with IFNc in cells derived from HT1080, in which IFNc stimulation is known to induce IRF-1 expression [28].As shown in Figure 2 (the full-length images with size markers are shown in Figure S4), IFNc induced expression of IRF-1 even in the presence of 3xFLAG-dCas9 and gRNA targeting the IRF-1 promoter.This result suggested that binding of 3xFLAG-dCas9 and gRNA-hIRF-1 #12 does not abolish IRF-1 transcription.
enChIP-SILAC analysis to quantitatively detect changes in the amounts of proteins associated with the IRF-1 promoter in response to IFNc stimulation Next, we performed enChIP-SILAC of the HT1080-derived cells expressing 3xFLAG-dCas9 and gRNA-hIRF-1 #12 to identify proteins whose interaction with the IRF-1 promoter changes in response to IFNc stimulation (Figure S5).In SILAC, cells are differentially labeled by culturing them in Light medium containing normal amino acids or Heavy medium with amino acids such as Lysine and Arginine conatining heavy isotopes.The mass shift of proteins resulted by metabolic incorporation of the Light and Heavy amino acids is detected by mass spectrometry.For quantification of detected proteins, Heavy to Light protein ratios (Heavy/Light values) are calculated within samples by summing average intensity values for all heavy peptides for each protein and dividing by the corresponding light values [22].In our experimental settings, the Heavy/Light values more than 1 indicated that the identified proteins were detected more abundantly from cells stimulated for 30 min with IFNc than from the mock-stimulated cells.We detected a list of proteins whose interactions with the IRF-1 promoter change by IFNc stimulation (Table 3 and Table S1).

IFNc-induced increased association of the candidate proteins with the IRF-1 promoter
To confirm increased association of candidate proteins with the IRF-1 promoter by IFNc stimulation, we attempted to perform ChIP analysis using Abs against endogenous proteins.In this regard, involvement of histone deacetylation has been implicated in IFNc-induced gene expression [48,49].However, molecules involved in IFNc-induced histone deacetylation were not fully revealed.Therefore, we chose proteins involved in histone deacetylation and/or corepressor function for ChIP analysis.As shown in Figure 3, IFNc stimulation increased binding of RBBP4 and PA2G4 to the IRF-1 promoter, whereas their binding to the control Sox2 locus showed a marginal change.IFNc induced increase in binding of RBBP4 at the distal region of the IRF-1 promoter (between 21 kb and 20.3 kb from the transcription start site (TSS)) (Figure 3B), whereas association of PA2G4 increased in the proximal region (between 20.3 kb and +0.1 kb from TSS) by IFNc stimulation (Figure 3C).In this experimental setting, we confirmed IFNc-induced association of Stat1 with the IRF-1 promoter (Figure S6) to ensure validity of stimulation conditions.This result clearly showed that enChIP-SILAC is able to detect relevant proteins whose association with the target genomic regions changes in response to extracellular stimuli.

Discussion
In this study, we developed a retroviral enChIP system using CRISPR.In our previous paper of enChIP using CRISPR, we used a transient transfection approach [6].The approach worked well in those cell lines with high transfection efficiency such as 293T cells.However, the transient transfection approach might be difficult to use for cells with low transfection efficiency.In those cases, stable expression of components of the CRISPR system would be necessary.Retroviral transduction of enChIP compo-nents enables us to establish stable cell lines in a shorter period of time and with less effort than conventional transfection strategies.We showed that the retroviral enChIP system is able to purify the targeted genomic regions efficiently and specifically from cell lines including HT1080 (Figure 1) and K562 (Figure S3).Other endogenous loci were also specifically purified from K562-derived cells by enChIP (T.F. and H.F., unpublished data).The yield of the retroviral enChIP system using HT1080-derived cells was comparable with that observed in 293T cells transiently transfected with 3xFLAG-dCas9 and gRNA [6].
It has been suggested that one of potential drawbacks of the CRISPR system is off-target effects [50][51][52][53].Therefore, we examined if contamination of potential off-target sites has a big impact on the results of enChIP using CRISPR.As shown in Figure 1B, yield for potential off-target sites was negligible compared with that of the target site when the gRNAs for target sites have more than 16-base long unique sequences in their seed sequences.It would be likely that contamination of potential offtarget sites would become larger as the target site has a shorter unique sequence.Our results also indicated importance of the seed sequence near PAM because a single base mismatch near PAM significantly decreased the yield (potential off-target sites in chromosomes 1 and 22).These results are consistent with the data showing that cleavage of target DNA by wild-type Cas9 is governed by the seed sequence of gRNA [10,15] and a recent report that DNA strand separation and RNA-DNA heteroduplex formation initiate at PAM and proceed directionally towards the distal end of the target sequence [54].Our data might be informative to design gRNA for enChIP using CRISPR.
Next, by using the retroviral enChIP system, we attempted to identify proteins associated with the IRF-1 promoter in an IFNcdependent manner.We detected a list of proteins whose association increases upon IFNc stimulation (Table 3 and Table S1).The list contained proteins involved in transcriptional regulation and other genome functions.Especially, the fact that the enChIP-SILAC analysis identified proteins involved in histone deacetylation and/or corepressor function is consistent with the previous reports that histone deacetylases positively regulate IFNcinduced gene expression [48,49].For example, HDAC1 has been implicated in IFNc-induced gene expression [48].
Finally, we confirmed IFNc-induced association of two of the identified proteins, RBBP4 and PA2G4, with the IRF-1 promoter by ChIP (Figure 3).It has been shown that RBBP4 and PA2G4 are involved in histone deacetylation and/or corepressor function [40][41][42], but their involvement in IFNc-induced gene expression has not been documented.Considering that we also detected TBL3 [43,44], another component of histone deacetylase/corepressor complexes, in enChIP-SILAC analysis, it is likely that the histone deacetylase/corepressor complexes containing these proteins play important roles in IFNc-induced transcription together with HDACs such as HDAC1.Elucidation of the functions of the proteins identified in this study in IFNc-induced gene expression would be an interesting future issue.Since we detected many proteins in our enChIP-SILAC analysis, it would be a challenge to make a comprehensive view on their roles in IFNc-induced gene expression at this stage.Systems biology approaches in the analysis of the detected proteins would be an attractive option in the future analysis.

Conclusions
In this study, we developed a retroviral enChIP system using CRISPR.We showed that the target genomic locus could be purified with high efficiency by using this system (Figure 1).We also showed that contamination of potential off-target sites is negligible by using this system if gRNAs for the target sites have sufficiently long unique sequences in their seed sequences (Figure 1).enChIP-SILAC analysis identified proteins induced to bind to the IRF-1 promoter region in response to IFNc stimulation (Table 3).The list of the associated proteins contained many novel proteins in the context of IFNc-induced gene expression as well as proteins related to histone deacetylase complexes whose involvement has been suggested in IFNcinduced gene expression (Table 3).We confirmed IFNc-induced association of identified proteins with the IRF-1 promoter (Figure 3).Thus, our results showed that the retroviral enChIP system using CRISPR would be useful for biochemical analysis of genome functions.

Establishment of cells stably expressing 3xFLAG-dCas9 and gRNA
For establishment of cells expressing 3xFLAG-dCas9, 5 mg of 3xFLAG-dCas9/pMXs-puro together with 5 mg of an amphotropic helper plasmid, pPAM3 [57], was transfected into 1610 6 of 293T cells using Lipofectamine 2000 (Life Technologies) according to the instructions by the manufacturer to produce retrovirus particles.Two days after transfection, HT1080 cells or K562 cells were infected with the supernatant (5 ml) of the 293T cells containing the virus particles.HT1080-or K562-derived cells expressing 3xFLAG-dCas9 were selected in DMEM or RPMI medium containing 10% FCS and puromycin (0.5 mg/ml), respectively.

Immunoblot analysis
Expression of 3xFLAG-dCas9 was detected by immunoblot analysis with anti-FLAG M2 Ab (F1804, Sigma-Aldrich) as described previously [6].For detection of the IRF-1 protein, HT1080 and its derived cells were stimulated with 100 ng/ml of recombinant human IFNc (rhIFNc).Nuclear extracts (NE) were prepared with NE-PER Nuclear and Cytoplasmic Extraction Reagents (Thermo Fisher Scientific).7.5 mg of NE were subjected to immunoblot analysis with anti-human IRF-1 Ab (C-20, Santa Cruz Biotechnology).

Enchip-real-time PCR
enChIP-real-time PCR was performed as previously described [6].Primers used in the analysis are shown in Table S2.

ChIP
HT1080 cells were stimulated with 100 ng/ml of rhIFNc at 37uC for 30 min.The cells (2610 7 ) were fixed with 1% formaldehyde at 37uC for 5 min.The chromatin fraction was extracted and fragmented by sonication as described previously [58] except for using 800 ml of of Sonication Buffer and Ultrasonic disruptor UD-201.160 ml of the sonicated chromatin in Sonication Buffer with 1% Triton X-100 was pre-cleared with 5 mg of normal rabbit IgG conjugated to 30 ml of Dynabeads-Protein G and subsequently incubated with 5 mg of anti-PA2G4/EBP1 Ab (ab33613, Abcam), anti-RBBP4/RbAp48 Ab (ab38135, Abcam), or normal rabbit IgG conjugated to 30 ml of Dynabeads-Protein G at 4uC for 20 h.The immunoprecipitants were washed once each with 1 ml of Low Salt Wash Buffer, High Salt Wash Buffer, LiCl Wash Buffer, and TE Buffer (10 mM Tris-HCl (pH 8.0), 1 mM EDTA).After reverse-crosslinking at 65uC for at least 4 h, the DNA was purified with ChIP DNA Clean & Concentrator (Zymo Research) and used as template for real-time PCR with SYBR Select PCR system (Applied Biosystems) using the Applied Biosystems 7900HT Fast Real-Time PCR System.PCR cycles were as follows: heating at 50uC for 2 min followed by 95uC for 10 min; 40 cycles of 95uC for 15 sec and 60uC for 1 min.The primers used in this experiment are shown in Table S2.

Statistical analysis
p-values were calculated with the Prism software (Graphpad) usint t test.

Figure 1 .
Figure1.Yield of enChIP analysis for the target site and potential off-target sites.(A) Expression of 3xFLAG-dCas9 in HT1080-derived cells.Expression of 3xFLAG-dCas9 was detected by immunoblot analysis with anti-FLAG Ab.Coomassie Brilliant Blue (CBB) staining is shown as a protein loading control.(B) Upper panel: Yield of enChIP analysis for the target site and potential off-target sites (mean +/-SD, n = 3).Lower panel: Alignment of the target site and potential off-target sites.The PAM sequences and mismatches are shown in blue and red, respectively.doi:10.1371/journal.pone.0103084.g001

Figure 2 .
Figure 2. IFNc-induced expression of IRF-1.HT1080 and its derived cells were stimulated with 100 ng/ml of IFNc for indicated time intervals.Nuclear extracts were subjected to SDS-PAGE and immunoblot analysis with anti-IRF-1 Ab.CBB staining is shown as a protein loading control.doi:10.1371/journal.pone.0103084.g002

Figure S1
Figure S1 The scheme of enChIP using CRISPR.(PDF) Figure S2 The full-length images of Figure 1A including molecular size markers.(PDF) Figure S3 Specific isolation of the IRF-1 promoter region from K562-derived cells by enChIP using CRISPR.(PDF) Figure S4 The full-length images of Figure 2 including molecular size markers.(PDF) Figure S5 The scheme of enChIP-SILAC.(PDF) Figure S6 IFNc-induced association of Stat1 with the IRF-1 promoter.(PDF)

Table 3 .
Examples of proteins identified by enChIP-SILAC.All of identified proteins, identified peptides, and the raw Heavy/Light value are shown in TableS1.doi:10.1371/journal.pone.0103084.t003chip Development Center for Infectious Diseases (RIMD, Osaka University).