Mechanism and therapeutic implications of pomalidomide-induced immune surface marker upregulation in EBV-positive lymphomas

Epstein-Barr virus (EBV) downregulates immune surface markers to avoid immune recognition. Pomalidomide (Pom) was previously shown to increase immune surface marker expression in EBV-infected tumor cells. We explored the mechanism by which Pom leads to these effects in EBV-infected cells. Pom increased B7-2/CD86 mRNA, protein, and surface expression in EBV-infected cells but this was virtually eliminated in EBV-infected cells made resistant to Pom-induced cytostatic effects. This indicates that Pom initiates the upregulation of these markers by interacting with its target, cereblon. Interestingly, Pom increased the proinflammatory cytokines IP-10 and MIP-1∝/β in EBV infected cells, supporting a possible role for the phosphoinositide 3-kinase (PI3K)/AKT pathway in Pom’s effects. Idelalisib, an inhibitor of the delta subunit of PI3 Kinase, blocked AKT-Ser phosphorylation and Pom-induced B7-2 surface expression. PU.1 is a downstream target for AKT that is expressed in EBV-infected cells. Pom treatment led to an increase in PU.1 binding to the B7-2 promoter based on ChIP analysis. Thus, our data indicates Pom acts through cereblon leading to degradation of Ikaros and activation of the PI3K/AKT/PU.1 pathway resulting in upregulation of B7-2 mRNA and protein expression. The increased immune recognition in addition to the increases in proinflammatory cytokines upon Pom treatment suggests Pom may be useful in the treatment of EBV-positive lymphomas.


Results
Pom-resistant cells have decreased cereblon, decreased surface immune markers, and little or no induction of T-cell activation when exposed to Pom. We previously demonstrated that Pom increases immune surface markers B7-2 and ICAM-1 in certain EBV infected tumor cell lines and that this leads to increased T-cell activation and NK-mediated cell killing in Daudi cells 19 . To determine if Pom-induced upregulation of these markers was correlated with Pom's cytostatic effects, we generated an EBV-infected Daudi cell line resistant to Pom's cytostatic effects by serial passaging with increasing concentrations of Pom up to 8 µM (see 'Material and Methods' section for details). The growth of wild type (WT) Daudi cells over a 72 h period was substantially inhibited by 1 µM Pom (Fig. 1A) even though this concentration of Pom did not induce cell death during this time (Supplemental file 2), suggesting Pom was slowing the growth of Daudi cells but not killing them. The growth of the Pom-resistant line, however, was similar to DMSOtreated cells up to 48 h and only showed a minimal (21%) decrease in growth by 72 h (Fig. 1B). Cereblon is the binding target for Pom and has been linked to its cytostatic and anti-myeloma activity [21][22][23] . Resistance to this anti-myeloma activity has been associated with decreased cereblon expression both in vitro and in clinical trials 24,25 . Consistent with this, our Pom-resistant Daudi cells showed a 67% decrease in the protein levels of cereblon as compared to the respective controls, and the decrease persisted in the presence of 1 µM Pom (Fig. 1C). Although cereblon could still be detected in the resistant line, the 67% decrease was sufficient to prevent most of the Pom-induced cereblon destruction of Ikaros (IKZF1), thus demonstrating a loss of cereblon function (Fig. 1C).
We then investigated the ability of Pom to upregulate B7-2 and ICAM-1 in these two lines. WT Daudi cells showed an average 1.9-fold increase in B7-2 surface expression following Pom treatment, while the resistant line showed only an average 1.12-fold increase when treated with Pom (Fig. 1D). Similarly, surface expression of ICAM-1 increased an average of almost 1.5-fold in WT Daudi cells, while this increase in expression no longer occurred in the resistant line treated with Pom (Fig. 1E). Examples of the surface expression FACS profiles for B7-2 and ICAM-1 and the effect of Pom treatment can be seen in Supplemental file 3. We previously demonstrated that Pom's effects on immune surface markers are associated with an increase in immune recognition, including increases in T-cell activation 48 h after exposure to Pom-treated Daudi cells 19 . Consistent with these previous results, treatment of WT Daudi cells with 1 and 10 µM Pom for 48 h resulted in enhanced T-cell activation by these cells (Fig. 1F). However, there was no significant increase in T-cell activation in the Pom-resistant Scientific Reports | (2023) 13:11596 | https://doi.org/10.1038/s41598-023-38156-z www.nature.com/scientificreports/ Daudi line, consistent with the substantial loss of the upregulation of B7-2 and ICAM-1 (Fig. 1F). Taken together, these data are consistent with Pom acting through cereblon to upregulate the immune surface markers in Daudi cells, resulting in increased immune recognition.

Pom treatment increases B7-2 mRNA and B7-2 cellular protein levels in Daudi cells.
To determine if the increases in B7-2 and ICAM-1 were due to increases in their transcription, we assessed mRNA expression levels after treatment of WT Daudi and Pom-resistant Daudi cells with Pom for 24 h. Treatment of WT Daudi cells with 1 μM Pom led to a significant 2.5-fold increase in B7-2 mRNA over its basal expression but interestingly did not induce a significant increase in ICAM-1 mRNA ( Fig. 2A). By contrast, B7-2 mRNA expression was not significantly increased in Pom-resistant cells and ICAM-1 mRNA also remained unchanged (Fig. 2B). For B7-2, the change in mRNA was mirrored in changes in the cytoplasmic protein levels of B7-2 with a 2.2-fold increase following Pom treatment of WT Daudi cells but not in Pom-resistant cells (Fig. 2C). We were unable to detect ICAM-1 protein in cellular extracts by immunoblot after trying multiple antibodies. Therefore, we could not assess differences in cellular protein levels for ICAM-1. Overall, these data are consistent with the increase in B7-2 in Daudi cells being caused, at least in part, by an increase in its mRNA although this appears not to be the case for the increase in ICAM-1 surface expression.  (Fig. 3). However, Pom treatment consistently increased both IP-10 and MIP-1∝/β over the control (average 4.5-fold, P < 0.05 for IP-10 and average 10.3-fold, P < 0.05 for MIP-1∝/β) (Fig. 3). Of note, the increases in MIP-1∝/β and IP-10 While there is evidence that MIP-1∝ activates phosphatidylinositol 3-kinase (PI3K) [28][29][30] , and this pathway has been implicated in activating PU.1 which is a known to bind and activate the B7-2 promoter, we did not observe an increase in B7-2 or ICAM-1 upon the addition of recombinant MIP-1∝ (Supplemental file 5). As discussed more later, the upregulation of MIP-1∝/β and IP-10 may contribute to tumor responsiveness through additional mechanisms.
A role for PU.1 and the PI3K/AKT pathway in upregulation of B7-2 surface expression. We next explored whether PU.1 and the PI3K/AKT pathway might play a role in Pom-induced upregulation of B7-2 mRNA. It is known that treating certain cells with Pom induces cereblon-dependent degradation of Ikaros, which is a repressor of the PI3K pathway 31 . Also, the PI3K/AKT pathway has been shown to be involved in activating PU.1, an E26 transformation-specific transcription factor (ETS) 32 that directly binds to the B7-2 promoter and can upregulate its mRNA expression in murine and human dendritic cells 29,30 . Pom has also been shown to upregulate PU.1 in myeloma cells 33  ChIP was performed with 4 antibodies (Ab1, Ab2, Ab3 and Ab4) specific to PU.1 and enrichment was determined by qPCR. We also tested an antibody to IRF8 in the ChIP (Ab5), as PU.1 is known to associate with this factor in certain cases 41 . We examined the primary putative B7-2 promoter region of isoform 1 and the putative promoter of isoform 2 (Fig. 6A), both of which contain PU.1 binding sequences (Fig. 6A). Four sets of primers were designed specifically to the regions that showed the highest accumulation of putative PU.1 binding sites in these respective promoters. Three sets of primers cover the primary promoter of isoform 1 of the B7-2 gene whereas one pair covers the internal promoter region of isoform 2 (designated pD) (Fig. 6A). As expected, and previously observed by other groups 29,30 , there was a small but consistent enrichment of PU.1 on the promoters of both B7-2 isoforms in the DMSO-treated cells as compared to IgG controls. This enrichment of PU.1 in the control samples supports a role of PU.1 in basal transcription (Fig. 6B,C,D,E). This basal binding of PU.1 on both promoters was very evident in case of Ab1 and Ab2 in regions pB and pC. In the primary promoter region pB, there was a 7.5 and 6.5-fold enrichment of PU.1 with Ab1 and Ab2, respectively (Fig. 6C). Ab4 failed to show significant PU.1 binding for pA,pB, and pC and the IRF8 antibody (Ab5) only showed modest increases over basal levels for the pA region (Fig. 6B). Importantly, upon Pomtreatment there was a significant enrichment over basal levels of PU.1 in regions of the primary promoter (pA, pB and pC) but not in the internal promoter of isoform 2 (pD) and this was particularly notable for the rabbit polyclonal ab1 (Fig. 6B,C,D). Upon Pom treatment, regions pB and pC of the B7-2 primary promoter showed a 15-and sevenfold enrichment, respectively, with Ab1 ( Fig. 6C,D). Although PU.1 was enriched at the internal promoter of isoform 2 (pD (Fig. 6E), there was no increase in enrichment following Pom treatment suggesting  We had previously found that Pom increases surface expression of B7-2 in PEL cells, but this was not associated with an increase in its mRNA 14 . Consistent with this, while PU.1 was easily detectable on the promoter in Daudi cells it was essentially undetectable in BCBl-1 cells (Supplemental file 8). The lack of PU.1 in these cells may in part explain the failure of Pom to increase B7-2/CD86 expression 20 . We confirmed the lack of expression of PU.1 in BCBL-1 cells by analyzing nuclear and cytoplasmic extracts for the presence of PU.1 in the absence and presence of Pom. Consistent with previous studies, PU.1 was not detected in BCBL-1 cell extracts with or without Pom treatment but was readily detectable in nuclear CA46 cell extracts (a control B cell line) (Supplemental file 9A). Interestingly, CA46 is a EBV-negative Burkitt lymphoma line and like BL41-, Pom did not increase B7-2 expression in these cells (Supplemental file 9B) suggesting PU.1 alone is not sufficient for Pom to upregulate B7-2 but appears to require EBV infection as well.
Pom increases B7-2 and ICAM-1 mRNA and surface expression in other EBV-infected cell lines. We then investigated whether Pom-induced increases in mRNA, protein and surface marker expression for B7-2 and ICAM-1 were similarly observed in other EBV-infected lines or virus-negative B cell lines. To explore this, we used EBV negative BL41 Burkitt lymphoma cells (BL41-), EBV-infected BL41 (BL41 +) cells 43 , and Namalwa EBV-positive Burkitt lymphoma cells that contain two integrated copies of the EBV genome 44 . Pom did not increase B7-2 or ICAM-1 mRNA in the uninfected BL41-cells but led to a 1.4-fold average increase in B7-2 mRNA expression in the EBV-infected BL41 + cells (Table 1). Unlike the case with Daudi cells, Pom also  (Table 1). We next assessed whether the PI3K pathway is activated in these other EBV infected lines by measuring AKT-ser phosphorylation by immunoblot. AKT-ser phosphorylation was increased 1.6-fold by Pom treatment www.nature.com/scientificreports/ in BL4 + cells and in Namalwa cells, and basal and Pom-induced AKT phosphorylation was eliminated when pretreated with idelalisib (Fig. 7A,B). Similarly, B7-2 surface marker expression was increased by Pom 1.6 and 1.2-fold for BL41 + and Namalwa, respectively (Fig. 7C,D). Importantly, idelalisib inhibited the basal and Pominduced expression of B7-2 (Fig. 7C,D). Cellular B7-2 expression was also increased by Pom in these two lines (Supplemental file 10). Idelalisib, however, was not able to suppress the Pom-induced increases in ICAM-1 suggesting again that the upregulation of ICAM-1 by Pom goes through a different pathway (Fig. 7E,F)   www.nature.com/scientificreports/ not substantially increase B7-2 or ICAM-1 surface expression in the uninfected BL41-cells even at 5 or 10 µM Pom, or the cellular B7-2, suggesting that infection of the cells is required for the effects of Pom in BL lines (Supplemental file 10A,B). We also investigated the effects of Pom on the B-cell lymphoma line MC116. While Pom decreased the levels of B7-2 in uninfected MC116 cells, it increased its expression in the KSHV-infected line up to 1.7-fold (Supplemental file 10C). Taken together, these data suggest that Pom's effects on B7-2 and ICAM-1 upregulation in certain lymphoma lines requires the presence of virus infection.

Discussion
Our group has shown that treating KSHV-infected PEL or certain EBV-infected lymphoma B-cell lines with Pom leads to the upregulation of important immune surface markers including ICAM-1, B7-2, MICA, and MHC-I 19 .
We have also shown that this treatment results in increased T-cell activation as well as increased natural killer cell activity toward these cell lines 15 Figure 7. Idelalisib inhibits AKT-Ser 473 phosphorylation and B7-2 surface expression induced by Pom in BL41 + and Namalwa cells. BL41 + and Namalwa cells were pretreated with DMSO or the PI3K δ-subunit inhibitor idelalisib for approximately 2 h and then treated with DMSO or 1 µM Pom for an additional 48 h. Representative immunoblots of cytoplasmic extracts show AKT phosphorylation levels with Pom treatment and/or idelalisib for (A) BL41 + cells or (B) Namalwa cells. Fold changes of AKT phosphorylation and total AKT are shown below the blot and calculated by first normalizing to β-actin. AKT phosphorylation was normalized to total AKT present after normalizing to actin. BL41 + and Namalwa cells treated as indicated in above were analyzed for surface expression levels of (C, D) B7-2 and (E, F) ICAM-1 by flow cytometry using PerCP/Cy5.5-conjugated anti-ICAM-1 or anti-B7-2 antibodies. Graphs show fold change in median fluorescent intensity (MFI) of surface expression levels measured via flow cytometry relative to DMSO-treated cells. The data are means + / − SD from 3 or more experiments and the statistically significant differences (****p ≤ 0.0001, ***p ≤ 0.001, **p ≤ 0.01, *p ≤ 0.05, ns not significant) between various treatments are indicated. www.nature.com/scientificreports/ tumors, since an effective T-cell immune response against virus infection requires expression of MHC-I and/ or co-stimulatory molecules such as B7-2 and ICAM-1 on target cells 45 . To explore the mechanism(s) for these effects, we have now developed cell lines resistant to Pom's cytostatic effects by serially passaging these lines with increasing concentrations of Pom; these include a BCBL-1 Pom-resistant cell line 15 , and in this report, a Pomresistant Daudi cell line. In both cases, resistance to Pom's cytostatic effect is associated with a decrease in the expression of cereblon, the primary target of pomalidomide and related derivatives. Although Pom treatment of KSHV and EBV infected lines appear to require the presence of cereblon and ultimately lead to similar outcomes regarding increased immune surface markers and immune recognition, here we have found distinct differences in the mechanism leading to these effects in PEL lines versus EBV-associated tumor lines. When we initiated these experiments, we hypothesized that Pom would upregulate immune surface markers in EBV-infected lines in the same way as in KSHV infected lines. Although Pom appears to require cereblon to activate the PI3K pathway in both KSHV and EBV-infected cell lines, the mechanism diverges from there. Most notable among the differences was that Pom induced increases in B7-2 mRNA in EBV infected cells while this was not observed in KSHV infected PEL cells. An underlying reason for this difference may be that KSHV and EBV-infected tumor lines differ in the expression of the important ETS transcription factor PU.1 which is important in myeloid and B-cell development. Here, we confirmed that BCBL-1 cells do not express PU.1, as reported by others 20 and this is the case without or with Pom treatment. However, we found that Daudi cells and other non-Hodgkin EBV + lymphoma lines, including Namalwa, express significant levels of PU.1 consistent with previous reports 46 and this may explain the higher basal expression of B7-2 compared to PEL lines. Since PU.1 has been shown to upregulate B7-2 in murine dendritic cells and more recently in human dendritic cells by binding directly to the promoter of B7-2 29,30 , we explored its potential role in Pom induced upregulation of B7-2 mRNA. Using ChIP assays, we demonstrated the presence of PU.1 at the promoter of B7-2 in WT Daudi cells and its absence at the promoter in control BCBL-1 cells. More importantly, we find that Pom treatment increased the levels of PU.1 in the nucleus and increased the levels of PU.1 associated with the B7-2 promoter in Daudi cells but not in BCBL-1 cells. It is important to note that PU.1 expression alone is not enough to lead to increased B7-2 following Pom treatment. CA46 cells express PU.1 but Pom does not increase the levels of PU.1 nor raise B7-2 in these cells. Thus, while PU.1 expression is required, it alone is not sufficient to induce B7-2 upregulation in EBV-negative BL lines.
We pursued the mechanism by which Pom might induce PU.1 activation in Pom treated Daudi cells. Our cytokine/chemokine data did not show increases in IFN-γ, IL-21, or IL-4, which have been shown to increase B7-2 in myeloma cell lines. However, we did consistently see increases in MIP-1∝/β and IP-10 after treatment of Daudi cells with Pom. We explored whether MIP-1∝ might be involved in the upregulation of ICAM-1 and B7-2 as MIP-1∝ has been shown to upregulate the PI3K pathway 28 , but we found no evidence for its involvement. However, when we further explored the PI3K pathway using an inhibitor of the delta subunit of PI3K, it was able to block the upregulation of B7-2 surface expression. Idelalisib blocked serine 473 phosphorylation of AKT in Daudi cells, which is required for phosphorylation of PU.1 by AKT 35 , and blocked the upregulation of B7-2 surface expression, but not that of ICAM-1. The PI3K/AKT pathway was only weakly activated by Pom treatment in the resistant Daudi cell line, suggesting that activation of that pathway is likely dependent on cereblon. Unlike the Daudi wild-type line, our Daudi cell line resistant to Pom showed an impaired ability to downregulate ikaros in response to Pom. Thus, Pom-induced degradation of ikaros by cereblon may lead to the activation of the PI3K pathway, as has been reported in leukemia cells 31 , and in the case of EBV-infected cells, this leads to an increase in PU.1 activation and upregulation of B7-2 mRNA.
A distinct difference between the mechanism of B7-2 upregulation in PEL and Daudi cells is the role played by the delta subunit of PI3K. Idelalisib, a specific inhibitor of the delta subunit of PI3K, was able to block AKT phosphorylation in Daudi cells and completely block the upregulation of B7-2 surface expression. By contrast, idelalisib used at the same and higher concentrations only minimally lowered the surface expression of B7-2 in PEL cells. Also, while ICAM-1 and B7-2 in PEL lines both seem to share a similar pathway that goes, in part, through PI3K 15 , there are differences in the mechanism for these two markers in EBV-infected cells. For example, inhibitors of the PI3K/AKT pathway had no inhibitory effect on the Pom-induced upregulation of ICAM-1 in Daudi cells. ICAM-1 expression has been shown to either be variably upregulated or downregulated by PI3K in multiple cell types indicating different mechanisms are present for this surface marker 47,48 .
We also explored the effects of Pom on ICAM-1 and B7-2 surface expression in two other EBV-infected BL lines (Namalwa and BL41 +) as well as the EBV negative BL counterpart, BL41-. Interestingly, although both BL41 − and BL41 + cells have downregulated CD54/ICAM-1 and CD86/B7-2 expression 49,50 , Pom only increased immune surface markers in the EBV positive BL41 line. Both Namalwa and BL41 + showed increases of B7-2 and ICAM-1 with Pom treatment although there were differences in the extent of upregulation among these EBV positive lines. We also studied the effects of Pom on the B-cell lymphoma line MC116 either uninfected or infected with the EBV-related virus, KSHV. Like that observed for the BL41 cells, only the infected MC116 cells showed responses to Pom. While our data demonstrate a clear difference between the effects of Pom on matched uninfected and infected lines (BL41 and MC116) it remains possible that certain non-virus infected lymphoma lines might also respond significantly to Pom with surface marker upregulation. For Daudi cells, we have previously demonstrated that the increases in B7-2, ICAM-1 and MHC class I polypeptide related sequence A (MICA) translate to increased immune recognition by natural killer cells and T-cells ( Fig. 1 and 19 ). The data presented here and reported previously suggests that Pom may be useful in the treatment of those BL associated with EBV infection.
It is also noteworthy that Pom treatment of Daudi cells significantly increased MIP-1∝/β and IP-10 levels in the supernatant of Pom-treated EBV-infected cells. MIP-1∝ and IP-10 levels have been associated with regressing BL tumors 51 . In fact, IP-10 was shown to mediate tumor necrosis of Burkitt lymphoma tumors 51  www.nature.com/scientificreports/ due to the upregulation of IP-10. Additionally, therapy of tumors with individual chemokines including MIP-1∝ and IP-10 can induce tumor regression and immunity to subsequent tumor challenge ( 53 and references therein). Together these findings suggest that Pom may be useful in the treatment of EBV-infected lymphomas by two mechanisms. One, by enhancing immune recognition and the other by eliciting antitumor effects through the production of proinflammatory chemokines.

Methods
Cells and cell culture. The  T-cell activation. T-cell activation assays were performed using IL2-Jurkat T-cells (Promega, cat# J1651) as the effector cells according to the manufacturer's recommended protocol and as described previously 15 . Briefly, 3 × 10 5 Daudi cells per mL were treated with DMSO control or indicated concentrations of Pom for 2 days. They were then co-incubated with 10 5 IL2-Jurkat T-cells at a 1:5 target to effector ratios in a 37 °C incubator and stimulated with 10 µg/mL of anti-human CD3 monoclonal antibody (OKT3 from ThermoFisher Scientific, cat# 16-0037-81). Relative light units (RLU) were measured after 6 h using Victor X3 multilabel plate reader (PerkinElmer), and background luminescence from media control was subtracted from all wells. Fold change in activation by Pom was determined after subtracting baseline RLU obtained from Jurkat cells incubated alone from that obtained from Jurkat cells co-incubated with Daudi cells.
Flow cytometry analysis. Analysis of live (unfixed) cells for surface marker expression was carried out as described previously 14 . PerCP-Cy5.5-labeled isotype or antibodies toward CD86 (cat # 374,215), and CD54 (cat # 353,119) were purchased from Bio-legend, San Diego CA. Analysis was conducted using a flow cytometry Calibur™ Flow Cytometry system (BD Biosciences, San Jose, CA) followed by FlowJo flow cytometry analysis software (flowjo.com).
Immunoblotting. Nuclear and cytoplasmic extracts were prepared from 2 × 10 6 -4 × 10 6 live cells using the www.nature.com/scientificreports/ from the exposed membranes, they were then incubated with a fluorescent labeled streptavidin antibody at 1:2000 (Li-Cor, #926-32,230), washed and scanned. Image-Studio was used to quantify the intensities of the spots on the membrane, in duplicate, and the average was taken and used as the level of intensity.
Real-time quantitative reverse transcription PCR. After treatment, mRNA was extracted using Qia-gen™ kit (ThermoFisher Scientific, Waltham MA). cDNA synthesis was performed using High-Capacity cDNA Reverse Transcription kit (ThermoFisher Scientific) on a T100 Thermal Cycler (Bio-Rad). q-PCR reaction setup using SYBR-green included enzyme activation at 95 °C for 10 min and 40 cycles of amplification at 95 °C for 15 s and 60 °C for 1 min followed by melting curve analysis. Expression was normalized to 18S endogenous control RNA and quantification of relative mRNA expression was performed using ∆∆Cт method. The following primers (5' to 3') were used in for q-PCR: 18S: GCC CGA AGC GTT TAC TTT GA and TCC ATT ATT CCT AGC TGC GGT ATC , Primers for B7-2 and ICAM-1 were from Bio-Rad (B7-2: 10,025,636, qHsaCED0043530) and (ICAM-1: 10,025,636, qHsaCED0004281) (Hercules, CA).
Chromatin immunoprecipitation (ChIP) assay. Chromatin immunoprecipitation (ChIP) assays were performed using an Abcam ChIP kit (Ab500) according to the manufacturer's instructions. Briefly, Daudi cells were treated with 1 µm Pomalidomide or equivalent amount of carrier DMSO. Treatment was performed 2 h post seeding of 0.2 million per mL Daudi cells. Twenty hours post incubation, 5 million cells were collected and washed with PBS. Cells were fixed with 1.1% of formaldehyde for 10 min at RT and samples were neutralized using glycine followed by lysis in Buffer C. Lysis reaction was stopped by addition of 100 µl buffer D followed by sonication for 10 min (cycle: 25 s on/off) in a Diagenode Bioruptor. For the ChIP, 10 µg of four different PU.1 antibodies; Ab#1 (PA5-115,807, Invitrogen), Ab#2 (6688-MSM2, Neo Biotechnologies), Ab#3 (PA351558, Invitrogen), Ab#4 (ab227835, Abcam), Ab#5 (PA17505, Invitrogen), an IRF8 antibody (D20D8, Cell Signaling technology) and rabbit IgG (Sigma) were incubated overnight with the sheared chromatin. Antibody pull-down was performed with Protein-A beads (Thermo) and bound beads were washed 5 times in wash buffer to remove non-specifically bound material. Bound DNA was isolated using a DNA purifying bead slurry provided with the kit. The amount of chromosomal DNA immunoprecipitated by each antibody was determined by quantitative PCR using the primers indicated (Supplemental File 1) and qPCR was performed using QuantStudio 3 (Applied Biosystems). Fold enrichment method was used to calculate the abundance of each fragment for different antibodies.

Data availability
The data that support the findings of this study are presented in the main text or the supplementary material. Additional data that support the findings of this study are available from the author, [R.Y.], upon reasonable request and/or will be deposited in the Dryad repository.