Histone H3k4me1 and H3k27ac Play Roles in Nucleosome Depletion and Erna Transcription

bioRxiv preprint doi: https://doi.org/10.1101/2021.01.05.425373; this version posted January 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1 Histone H3K4me1 and H3K27ac play roles in nucleosome depletion and eRNA transcription,

2 respectively, at enhancers

5 Yujin Kang, Yea Woon Kim, Jin Kang and AeRi Kim*

7 Department of Molecular Biology, College of Natural Sciences,

8 Pusan National University, Busan 46241, Korea

12 Keywords: H3K4me1, H3K27ac, enhancers, nucleosome depletion, eRNA transcription

16 * To whom correspondence should be addressed. Tel: 82-51-510-3683; Fax: 82-51-513-9258; Email:

17 [email protected]

18 bioRxiv preprint doi: https://doi.org/10.1101/2021.01.05.425373; this version posted January 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

19 ABSTRACT

20 Histone H3K4me1 and H3K27ac are enhancer specific modifications and are required for enhancers to

21 activate transcription of target genes. However the reciprocal effects of these histone modifications on

22 each other and their roles in enhancers are not clear. Here to comparatively analyze the role of these

23 modifications, we inhibited H3K4me1 and H3K27ac by deleting SET domain of histone

24 methyltransferases MLL3 and MLL4 and HAT domain of histone acetyltransferase p300, respectively,

25 in erythroid K562 cells. The loss of H3K4me1 reduced H3K27ac at the β-globin enhancer LCR HSs,

26 but H3K27ac reduction did not affect H3K4me1. This unequal relationship between two modifications

27 was revealed in putative enhancers by genome-wide analysis using ChIP-seq. Histone H3 depletion at

28 putative enhancers was weakened by the loss of H3K4me1 but not by the loss of H3K27ac. Chromatin

29 remodeling complexes were recruited into the β-globin LCR HSs in a H3K4me1-dependent manner. In

30 contrast, H3K27ac was required for enhancer RNA (eRNA) transcription, and H3K4me1 was not

31 enough for it. Forced H3K27ac induced eRNA transcription without affecting H3K4me1 at the β-globin

32 LCR HSs. These results indicate that H3K4me1 and H3K27ac affect each other in different ways and

33 play more direct roles in nucleosome depletion and eRNA transcription, respectively, at enhancers.

35 INTRODUCTION

36 Enhancers are cis-regulatory elements that highly induce the transcription of target genes (Plank and

37 Dean 2014; Smith and Shilatifard 2014). They have unique histone modifications, H3K4me1 and

38 H3K27ac, in the chromatin environment. H3K4me1 and H3K27ac are catalyzed by histone

39 methyltransferases MLL3 and MLL4 (Herz et al. 2012; Hu et al. 2013) and histone acetyltransferases

40 CBP and p300 (Jin et al. 2011; Hilton et al. 2015), respectively, in mammalian cells. Depletion or

41 inactivation of these histone modifying enzymes disturbs cell-type-specific gene expression in addition

42 to the severe decrease of H3K4me1 or H3K27ac at enhancers (Kim and Kim 2013; Lee et al. 2013;

43 Wang et al. 2016; Ebrahimi et al. 2019). However, the reciprocal effects of these histone modifications

44 on each other at enhancers are not clear. Most studies have focused on one of the histone modifications

45 by targeting histone modifying enzymes, and controversial results have been obtained from different 2

46 experimental approaches such as knockout of the modifying enzyme genes, mutation in the catalytic

47 domains and chemical inhibition of the activity domains (Lee et al. 2013; Dorighi et al. 2017; Rickels

48 et al. 2017; Raisner et al. 2018).

49 Enhancers have a special chromatin structure that is hypersensitive to attack by endonucleases such as

50 DNase I and MNase, which allows access of transcription factors (TFs) to their binding motifs (Guertin

51 and Lis 2013; Inukai et al. 2017). Histones are depleted at the hypersensitive enhancers as revealed by

52 ChIP assay (Fang et al. 2009). Studies using MNase-treatment have shown the lack of nucleosome

53 structure at DNase I hypersensitive sites (HSs) (Kim et al. 2007; Grossman et al. 2018). When DNase

54 I sensitivity is increased at enhancers by transcriptional activation, it accompanies more severe

55 depletion of nucleosomes (Johnson et al. 2018). Enhancers defined by H3K4me1 are hypersensitive to

56 DNase I attack (Zentner et al. 2011). These findings suggest a correlation of enhancer specific histone

57 modifications with nucleosome or histone depletion.

58 Long non-coding RNAs are transcribed from enhancers by RNA polymerase II (pol II), which are

59 referred to as enhancer RNAs (eRNAs) (Kim et al. 2010). eRNA transcription is activated by

60 transcription factors such as p53 and nuclear hormone receptor, and by Mediator (Lai et al. 2013; Li et

61 al. 2013; Melo et al. 2013). Transcribed eRNAs have been reported to contribute to many transcriptional

62 activation steps including the recruitment of RNA pol II to promoters (Mousavi et al. 2013; Maruyama

63 et al. 2014), chromatin interaction between enhancers and promoters (Li et al. 2013; Hsieh et al. 2014)

64 and the elongation transition of paused RNA pol II (Schaukowitch et al. 2014). eRNAs-producing

65 enhancers have specific histone modifications including H3K27ac (Zhu et al. 2013). Depletion of

66 MLL3/4 or treatment with a CBP/p300 bromodomain inhibitor reduce eRNA levels, suggesting the

67 correlation of H3K4me1 and/or H3K27ac with eRNAs transcription (Rahnamoun et al. 2018; Raisner

68 et al. 2018).

69 Here, to analyze how enhancer-specific histone modifications H3K4me1 and H3K27ac affect each

70 other and what is the distinct roles of these modifications in enhancer activity, we inhibited H3K4me1

71 and H3K27ac by deleting the SET domain of MLL3 and MLL4 ( MLL3/4) and HAT domain of p300

72 ( p300), respectively, using the CRISPR/Cas9 system in erythroid△ K562 cells. H3K4me1 and

△ 3

73 H3K27ac were comparatively and reciprocally examined at the β-globin enhancer locus control region

74 (LCR) HSs in MLL3/4 and p300 K562 cells (Fig. 1A). In order to expand results from the β-globin

75 locus into the whole△ genome, △we carried out ChIP-sequencing. Histone H3 occupancy was analyzed at

76 the regulatory regions including enhancers in MLL3/4 cells and p300 cells. Recruitment of

77 chromatin remodeling complexes was examined at△ cell specific enhancers△ including the β-globin LCR

78 HSs. The roles of H3K4me1 and H3K27ac in eRNA transcription were explored by analyzing total

79 RNA-seq data. These studies indicate that H3K4me1 and H3K27ac affect each other in different ways

80 and play distinct roles in activating enhancers.

82 RESULTS

83 Loss of H3K4me1 decreases H3K27ac at the β-globin enhancer LCR HSs, but loss of H3K27ac

84 does not affect H3K4me1.

85 To study reciprocal effects of enhancer specific histone modifications H3K4me1 and H3K27ac on each

86 other, we deleted DNA sequences encoding catalytic domains of MLL3 (KMT2C) and MLL4 (KMT2D)

87 or p300 using CRISPR/Cas9 technique in human erythroid cell line K562 (Supplemental Fig. S1).

88 Deletions of SET domains in MLL3/4 ( MLL3/4) and deletion of HAT domain in p300 ( p300) did

89 not affect protein stability of MLL3, MLL4△ and p300 when it was analyzed by western blot△ ting with

90 antibodies against N-terminal regions (Fig. 1B). Amounts of histone H3K4me1 and H3K27ac were

91 decreased by MLL3/4 and p300, respectively, at the total protein levels (Fig. 1C). While p300 did

92 not affect amount△ of H3K4me1,△ H3K27ac was reduced in two MLL3/4 clonal cells in△ spite of

93 unaffected p300 expression. The loss of catalytic domains reduced transcription△ of the erythroid specific

94 γ-globin genes in K562 cells (Fig. 1D). To analyze enhancer histone organization, we carried out ChIP

95 assay and examined the β-globin enhancer LCR using qPCR. Histone H3K4me1 and H3K4me2 were

96 decreased by MLL3/4 at the LCR HSs, but not by p300 (Fig. 1E, F). H3K27ac was decreased by

97 MLL3/4 in △addition to p300 (Fig. 1G). These results△ imply that H3K4me1 affects H3K27ac in

98 enhancer△ s, but not the reverse.△ In addition, ChIP analysis for histone H3 indicated that H3K4me1

99 contributes to histone depletion at the β-globin LCR HSs (Fig. 1H). 4

100

101 Genome wide analysis shows H3K4me1 is required for H3K27ac in putative enhancers but

102 H3K27ac is not for H3K4me1.

103 In order to expand observations from the β-globin LCR HSs into the whole genome, we carried out

104 ChIP-seq analysis for H3K4me1 and H3K27ac. Our ChIP-seq data in control cells were consistent with

105 public ENCODE ChIP-seq data in K562 cells (Supplemental Fig. S2). 108,210 regions were sorted by

106 H3K4me1 enrichment in control cells (Fig. 2A). H3K4me1 was almost completely eliminated by

107 MLL3/4 in 90% (n = 96,928) of the regions, where H3K27ac was also severely reduced. In the other

108 hand,△ p300 decreased H3K27ac in 79% (n = 89,199) of this modification peaks (n = 112,261) (Fig.

109 2B), even△ though the decreases were not severe as the decreases of H3K4me1 by MLL3/4. H3K4me1

110 was maintained in regions where H3K27ac is decreased. These results indicate△ that MLL3/4 and

111 p300 inhibit H3K4me1 and H3K27ac, respectively, genome-wide. H3K4me1 appears△ to affect

112 H3K27ac△ , but it is not likely to be the reverse case.

113 To focus on the relationship between H3K4me1 and H3K27ac at enhancers, we have screened HSs by

114 overlapping DNase-seq data and FAIRE-seq data (Fig. 2C). The HSs (n = 65,377) were classified into

115 putative enhancers by H3K4me1 in extragenic regions, promoters by TSS ± 1 Kb, insulators by CTCF

116 occupancy and the ‘others’ for the remaining regions. H3K4me1 was most notable in putative enhancers,

117 where H3K4me1 was not dependent on H3K27ac, but H3K27ac was dependent on H3K4me1 (Fig. 2D).

118 Unexpectedly, H3K4me1 was increased in promoter by MLL3/4 and this might relate with ratio to

119 H3K4me3. Modification patterns in the ‘others’ were rel△atively similar with the patterns in putative

120 enhancers. This might be because the ‘others’ include enhancers located in genes/introns. Thus, this

121 genome-wide analysis shows that H3K4me1 is required for H3K27ac in enhancers, but it is not likely

122 to be in the opposite direction. It is consistent with the results obtained from the β-globin enhancer LCR

123 HSs.

124

125 Histone depletion at enhancers is dependent on H3K4me1 but not on H3K27ac.

126 Our results from the β-globin LCR HSs suggested that H3K4me1 might be required for histone

127 depletion in enhancers (Fig. 1G). To study this possibility, we analyzed histone H3 occupancy genome-

128 widely in MLL3/4 cells and p300 cells and then compared it at the HSs screened by DNase-seq

129 data and FAIRE△ -seq data. Histone△ H3 was depleted at the HSs as shown by heatmap (Fig. 3A). The

130 depletion was reduced by the loss of H3K4me1 at putative enhancers and ‘others’ that may include

131 enhancers (Fig. 3A, B). Reduction of H3K27ac did not significantly affect H3 depletion at the HSs

132 including putative enhancers. To study the correlation between H3K4me1 and histone depletion at

133 enhancers, the putative enhancers were sorted by H3K4me1 level and divided into three groups

134 according to the levels (Fig. 3C). Comparison between the groups showed that H3K4me1 level parallels

135 histone H3 depletion. It also correlates with chromatin accessibility by ATAC-seq data and with

136 nucleosome depletion by MNase-seq data (Fig. 3D), supporting H3K4me1-dependent histone depletion

137 at enhancers.

138 To more clarify relationship of H3K4me1 and H3K27ac with histone depletion at enhancers, we

139 divided putative enhancers into two groups depending on the changes of H3K4me1 by MLL3/4 (Fig.

140 3E). Histone H3 depletion was more weakened in H3K4me1-reduced enhancers (n =△ 8,184), even it

141 was not in the rest enhancers where H3K4me1 was not reduced. In contrast, histone depletion was

142 maintained in enhancers where H3K27ac was apparently reduced by p300 (n = 5,551) (Fig. 3F).

143 H3K4me1 was also maintained in the H3K27ac-reduced enhancers. Similar△ histone depletion patterns

144 were observed in ‘Others’ regions by reduction of H3K4me1 or H3K27ac (Supplemental Fig. S3).

145 Taken together, these results indicate that H3K4me1 is required for histone depletion at enhancers.

146 H3K27ac does not appear to directly contribute to the histone depletion.

147

148 Chromatin remodeling complexes are recruited into enhancer in a H3K4me1-dependent manner.

149 Because histones/nucleosomes can be depleted by the AT P -dependent chromatin remodeling process,

150 we aligned public ChIP-seq data for chromatin remodeling complexes to putative enhancers. BRG1

151 (SMARCA4), INI1 (SMARCB1) and BAF170 (SMARCC2), subunits of SWI/SNF (BAF) chromatin

152 remodeling complex, and SNF2h (SMARCA5), a subunit of ACF chromatin remodeling complex, were 6

153 robustly detected at the enhancers (Fig. 4A) and their levels positively correlated with histone

154 H3K4me1 level (Supplemental Fig. S4). Views of genome browser showed these subunits bind to the

155 β-globin LCR HSs (Fig. 4B). In order to explore the role of H3K4me1 in recruiting chromatin

156 remodeling complexes, we performed ChIP assay using antibodies for BRG1, SNF2h and ACF1

157 (BAZ1A, a subunit of ACF complex). These subunits were less detected at the β-globin LCR HSs in

158 MLL3/4 cells compared to control cells (Fig. 4C-E). Other erythroid specific enhancers (Kang et al.

159 2015△ ), where the subunits bind to (Supplemental Fig. S5), showed similar binding patterns with the

160 LCR HSs. The results suggest a positive role of H3K4me1 in recruitment of chromatin remodeling

161 complexes. However, no significant changes were observed in p300 cells. Genes coding the

162 remodeling complex subunits were similarly expressed in control cells,△ MLL3/4 cells and p300

163 cells (Fig. 4F). Thus, these results indicate that H3K4me1 plays a role△ in recruiting chromatin△

164 remodeling complexes into enhancers, but H3K27ac is not necessary for it.

165

166 eRNA transcription requires histone H3K27ac in addition to H3K4me1.

167 To study the roles of enhancer specific histone modifications in eRNA transcription, we divided putative

168 enhancers into active and poised enhancers according to the levels of H3K27ac as shown by heatmap

169 (Fig. 5A) and measured transcription in the enhancer regions. eRNAs were more highly transcribed in

170 active enhancers than poised enhancers. eRNA transcription from active enhancers was reduced in

171 MLL3/4 cells and p300 cells (Fig. 5B). Transcription of nearest genes (n = 500) from eRNAs-

172 decreased△ enhancers was△ also reduced as revealed by Gene Set Enrichment Analysis (GSEA) (Fig. 5C).

173 These results indicate that H3K27ac is required for eRNA transcription in addition to H3K4me1.

174 To further analyze relationship of H3K27ac and H3K4me1 with eRNA transcription, active enhancers

175 (n = 4,600) were divided into three groups according to H3K27ac levels or H3K4me1 levels in control

176 cells and transcription levels were measured in each groups (Fig. 5D, E). eRNA levels positively

177 correlate with H3K27ac, but there was no significant correlation between eRNA levels and H3K4me1

178 at active enhancers. When putative enhancers were divided depending on eRNA transcription, H3K27ac

179 was higher in enhancers transcribing eRNAs (n = 7,548) than not transcribing them, but there were no 7

180 great differences in H3K4me1 between the enhancer groups (Fig. 5F). Analysis according to eRNA

181 levels showed that eRNA transcription correlates with H3K27ac strongly but does so with H3K4me1

182 at a very weak level (Fig. 5G). Thus this analysis supports that H3K27ac plays a role in eRNA synthesis

183 at enhancers prepared by H3K4me1.

184

185 Increase of H3K27ac by TSA treatment induces eRNA transcription at the β-globin LCR HSs

186 without increase of H3K4me1.

187 To determine whether H3K27ac has a direct role in eRNA transcription, we treated mouse erythroid

188 MEL/ch11 cells with histone deacetylase inhibitors TSA. A human chromosome 11 containing the β-

189 globin locus is present in MEL/ch11 cells. TSA treatment for 6 h and 24 h increased H3K27ac at the

190 total protein level, without notable changes on H3K4me1 (Fig. 6A). It also elevated H3K27ac at the

191 chromatin level as shown by ChIP assay in the β-globin enhancer LCR HSs (Fig. 6B). Surprisingly,

192 eRNA levels were increased at the LCR HSs with 6 h and 24h TSA treatment (Fig. 6C), and it brought

193 about transcription of β-like globin genes following eRNA transcription (Fig. 6D). However, H3K4me1

194 was largely maintained through the β-globin locus (Fig. 6E). Thus, these results indicate that histone

195 H3K27ac plays a more direct role in eRNA transcription from enhancers rather than H3K4me1.

196

197 DISCUSSION

198 Our comparative and reciprocal studies show that histone H3K4me1 and H3K27ac affect each other at

199 enhancers in a different way; H3K4me1 is required for H3K27ac but this is not true in the opposite

200 direction (Fig. 7). H3K4me1 appears to play a role in histone depletion at enhancers by recruiting

201 chromatin remodeling complexes. H3K27ac is not likely to be necessary for the histone depletion.

202 Instead, eRNA transcription seems to require H3K27ac in addition to H3K4me1. Thus our results

203 suggest that enhancer specific histone modifications H3K4me1 and H3K27ac play distinct roles in

204 activating enhancers.

205 Results from the β-globin locus and genome-wide analysis show that H3K4me1 is required for

206 H3K27ac at enhancers, but H3K27ac is not for H3K4me1. This might be because H3K4me1 precedes 8

207 H3K27ac at poised enhancers but H3K27ac is deposited at active enhancers. There are reports that

208 MLL3/4, rather than H3K4me1, are important for p300 recruitment and H3K27ac (Wang et al. 2016;

209 Dorighi et al. 2017) and MLL3 and MLL4 are destabilized by deletion of the catalytic domains (Jang

210 et al. 2019). However, our catalytic domain deletion did not affect the stability of MLL3 and MLL4 and

211 eliminated H3K4me1 at enhancers, resulting in reduction of H3K27ac. Similar effects were reported in

212 mouse ES cells where MLL3/4 catalytic domain deletion resulted in reduction of H3K4me1 and

213 H3K27ac at active enhancers without affecting stability and occupancy of MLL3 and MLL4 (Rickels

214 et al. 2017). These findings support the contribution of H3K4me1 to H3K27ac at enhancers. The effects

215 of H3K27ac on H3K4me1 are controversial in many studies using chemical inhibitors to CBP/p300

216 activity (Wang et al. 2017; Raisner et al. 2018; Zhu et al. 2018; Ebrahimi et al. 2019). However, our

217 study using p300 catalytic domain mutation indicates that H3K27ac is not necessary for H3K4me1 at

218 enhancers, even though depletion of H3K27ac was not severe as the depletion of H3K4me1 by

219 MLL3/4. Similar results were observed in a study using CBP/p300 bromodomain inhibitor, where

220 H3K27ac△ was reduced at enhancers but H3K4me1 was maintained (Raisner et al. 2018). Taken together,

221 we conclude that the effects of H3K4me1 and H3K27ac on each other at enhancers are not reciprocal.

222 Histone H3K4me1 appears to play a role in recruiting chromatin remodeling complexes into enhancers,

223 resulting in histone depletion. Our analysis showed that several subunits of the complexes co-localize

224 with enhancers selected by H3K4me1 and are recruited in a H3K4me1-dependent manner. BRG1, a

225 subunit of SWI/SNF (BAF) chromatin remodeling complex, has been reported to regulate DNase I

226 sensitivity and histone H3 occupancy at enhancers of the α-globin locus (Kim et al. 2009). Depletion of

227 BRG1 impaires the extension of nucleosome linker regions by nucleosome shifting at tissue specific

228 enhancers (Hu et al. 2011). Consistent with our finding, binding of BRG1 is decreased by loss of

229 H3K4me1 in the Sox2 enhancer (Local et al. 2018). H3K4me1-containing mononucleosomes are more

230 efficiently bound and remodeled by BAF complex than H3K4me3-containing mononucleosomes in in

231 vitro assay (Local et al. 2018). In addition, DPF3, a subunit of BAF complex, has been reported to

232 recognize H3K4me1 using the PHD1-PHD2 domain (Lange et al. 2008; Local et al. 2018). The PHD

233 domain is also present in ACF1 subunit (Eberharter et al. 2004) that interacts with SNF2h in ACF 9

234 chromatin remodeling complex (He et al. 2006). Thus H3K4me1 may contribute to histone depletion

235 by providing recognition sites for subunits of chromatin remodeling complexes.

236 When it is compared to H3K4me1, H3K27ac appears to play a more direct role in eRNA synthesis,

237 even though these two modifications are required for it. eRNA transcription is reduced by inhibition of

238 H3K27ac via CBP/p300 bromodomain inhibitor (Raisner et al. 2018) or by loss of H3K4me1 via

239 depletion of MLL3/4, even though it can be due to the loss of MLL3/4 rather than the loss of H3K4me1

240 (Dorighi et al. 2017; Rahnamoun et al. 2018). However, in our studies, H3K4me1 was maintained at

241 enhancers when H3K27ac is inhibited by p300, implying that H3K27ac is more critical for eRNA

242 transcription and H3K4me1 is not sufficient.△ This proposal is supported by analysis using ChIP-seq

243 data and RNA-seq data that show a positive correlation of eRNA transcription with H3K27ac but no

244 significant correlation with H3K4me1 (Fig. 5). H3K27ac can be recognized by BRD4 (Filippakopoulos

245 et al. 2012; Roe et al. 2015). BRD4 has been reported to be required for recruiting RNA pol II into

246 enhancers (Nagarajan et al. 2014; Lee et al. 2017). Thus H3K27ac is thought to contribute to eRNA

247 transcription by recruiting RNA pol II into enhancers, where H3K4me1 might be already deposited.

248

249 METHODS

250 Cell culture and Trichostatin A treatment

251 K562 cells were grown in RPMI 1640 medium, and 293FT and MEL/ch11 cells were cultured in

252 DMEM medium as previously described (Kang et al. 2017). MEL/ch11 cells (5 × 105 cells/ml) were

253 treated with 25 ng/ml Trichostatin A (TSA, Sigma) for 6 hours (h) and 24 h in complete DMEM medium.

254

255 Deletion of catalytic domain of MLL3/4 or p300 using CRISPR/Cas9 system

256 Single guide RNA (sgRNA) sequences were designed for SET domains of MLL3 and MLL4 and HAT

257 domain of p300 with online tools (http://crispr.mit.edu). Two complementary oligos containing gRNA

258 sequences and flanking sequences for cloning were phosphorylated and annealed as previously

259 described (Kim and Kim 2017). Sequences of oligos for sgRNAs are presented in Supplemental Table

260 S1. Annealed oligos for MLL4 SET or p300 HAT were inserted into lentiCRISPRv2 vector (Addgene 10

261 #52961) (Sanjana et al. 2014) and annealed oligos for MLL3 SET were inserted into pLH-spsgRNA2

262 vector (Addgene #64114) (Ma et al. 2015). The vectors were cloned in Stbl3 bacteria, prepared using

263 the Plasmid Midi Kit (Qiagen), and transfected into 293FT cells using the Virapower packaging mix

264 (Invitrogen) and Lipofectamine 2000 (Invitrogen). Lentiviruses for MLL4 SET and p300 HAT sgRNA

265 were harvested after 3 days and transduced into K562 cells in the presence of 6 µg/ml polybrene. The

266 transduced cells were selected by 2 µg/ml puromycin and seeded in 96 well plates to obtain clones.

267 Lentivirus for MLL3 SET sgRNA was transduced into MLL4 SET deletion K562 cells and was selected

268 by 500 µg/ml hygromycin. Genomic DNA flanking the MLL3/4 SET domain or p300 HAT domain was

269 amplified by PCR (Supplemental Fig. S1). Primer sequences are listed in Supplemental Table S2.

270

271 Western blot analysis

272 Proteins were extracted from 2 × 106 cells using RIPA buffer with sonication. Equal amounts of the

273 proteins were electrophoresed in 4-15% SDS-PAGE gradient gel (Bio-rad) and transferred to 0.2 µm

274 NC membrane or 0.45 µm PVDF membrane. After blocking with 5% skim milk in PBST, the membrane

275 was incubated with primary antibodies against H3 (ab1791), H3K4me1 (ab8895), and H3K27ac

276 (ab4729) from Abcam and p300 (sc-8981) and β-tubulin (sc-9104) from Santa Cruz Biotechnology and

277 MLL3 and MLL4 from Jaehoon Kim (KAIST, Korea) at 4 °C overnight and then with secondary

278 antibodies, anti-rabbit HRP (sc-2030) in PBST with 1% skim milk. Rabbit polyclonal anti-MLL3 and

279 anti-MLL4 antibodies were developed against purified histidine-tagged human MLL3 and MLL4

280 fragments (amino acid residues 1-120 and 4442-4561, respectively), and affinity purified (AbClon).

281 Protein signals were exposed using ECL reagent.

282

283 Reverse transcription-PCR (RT-PCR)

284 Total RNA was extracted from 2 × 106 cells using QIAzol Lysis Reagent (Qiagen) and cDNA was

285 generated from 0.5 µg of RNA with random hexamers using the GoScript kit (Promega) as previously

286 described (Kang et al. 2018). Quantitative PCR was performed with TaqMan probes in 10 µl of reaction

287 volume using a 7300 Real-time PCR system (Applied Biosystems). The sequences of the probes and 11

288 primers are listed in Supplemental Table S3.

289

290 Chromatin immunoprecipitation (ChIP)

291 ChIP was performed as previously described (Cho et al. 2008). Cells (1 × 107) were cross-linked in 1%

292 formaldehyde and digested using MNase. Fragmented chromatin was reacted with antibodies and

293 recovered using protein A or G agarose beads. DNA was purified by phenol extraction and ethanol

294 precipitation and then analyzed by quantitative PCR. The sequences of primers for ChIP assay are

295 presented in Supplemental Table S3, S4. Antibodies used for ChIP experiment were H3 (ab1791),

296 H3K4me1 (ab8895), H3K27ac (ab4729), SNF2h (ab3749) from Abcam, H3K4me2 (07-030) from

297 Millipore, ACF1 (A301-318A) from Bethyl Laboratories and BRG1 (sc-17796) from Santa Cruz

298 Biotechnology. Normal rabbit IgG (sc-2027) from Santa Cruz Biotechnology was used as a negative

299 control.

300

301 RNA library preparation and sequencing

302 Total RNA was extracted using RNeasy Plus Mini Kit (Qiagen) according to the manufacturer’s

303 instructions and qualified using Qubit Fluorometer (RNA IQ > 7). After depleting ribosomal RNA

304 (rRNA) using NEBNext rRNA Depletion Kit (New England Biolabs #E6350L), RNA was reversely

305 transcribed using NEBNext Ultra II Directional RNA Library Prep Kit (New England Biolabs #E7765S)

306 as suggested by the manufacturer. This procedure includes fragmentation, priming using random

307 primers, first-strand cDNA synthesis, second-strand cDNA synthesis, cDNA end repair, adaptor ligation,

308 amplification using adaptor primers (8 cycles), and purification using beads. Final libraries were

309 quantified by Qubit dsDNA HS assay (Invitrogen) and 100 bases paired-end reads were sequenced

310 using an Illumina NovaSeq 6000 system.

311

312 RNA-seq analysis

313 Quality scores of sequenced reads were assessed using quality control tool. Average score was Q = 38

314 at each base across reads. Input read ends with poor quality values were removed from raw reads by 12

315 Trimmomatic (quality score 20) (Bolger et al. 2014). Remaining reads were aligned to the human

316 reference genome (hg19) using STAR (Dobin et al. 2013). Exon annotation GTF file was obtained from

317 UCSC database. Aligned reads were counted per gene ID (Entrez) using featureCounts (Liao et al. 2014).

318 Differentially Expressed Genes (DEG) were statistically analyzed using edgeR (Robinson et al. 2010)

319 with q-value under 0.05. CPM values above 1 were considered as meaningful transcription level in

320 genes. Gene set enrichment analysis (GSEA) was performed according to the instructions at GSEA

321 platform (www.gsea-msigdb.org/gsea/index.jsp). Genes near enhancers were sought using GREAT

322 platform (http://great.stanford.edu/public/html) and limited to 500 genes by eRNA reduction.

323

324 ChIP DNA library preparation and sequencing

325 ChIP DNA (10 ng for H3K4me1, H3K27ac, H3 and input) was processed with NEBNext Ultra II DNA

326 Library Prep Kit (New England Biolabs #E7103S) according to manufacturer’s instructions. ChIP DNA

327 were repaired at the ends, ligated with 1.5 µM NEBNext adaptors, selected in 200 bp size with NEBNext

328 sample purification beads, and amplified with the adaptor primers (7 cycles). After purifying, final

329 libraries were quantified with Qubit dsDNA HS assay (Invitrogen) and 100 bases paired-end reads were

330 sequenced on an Illumina NovaSeq 6000 system.

331

332 ChIP-seq analysis

333 ChIP-seq raw reads were qualified by removing input reads with poor quality values (quality score 20)

334 and then aligned to the hg19 canonical genome using Bowtie2 (Langmead and Salzberg 2012). Aligned

335 BAM files were filtered by minimum MAPQ quality score 20 and sorted by chromosomal coordinate.

336 Potential PCR duplicates were removed from BAM files. Peak regions of H3K4me1 and H3K27ac were

337 identified using MACS2 (Feng et al. 2012) providing histone H3 data as control. Thresholds of

338 enrichment q-value were 0.05 for narrow peak and 0.05 for broad peak. Matrix files were generated

339 from BAM files and bigwig files using ComputeMatrix, and then used for plotting heatmaps and

340 average profiles with plotHeatmap and plotProfile tools, respectively (Ramírez et al. 2016). Bin size of

341 matrix files was 50 bp. ChIP-seq signals were visualized using Integrated Genome Browser (IGB)

342 (Freese et al. 2016).

343

344 Public NGS data

345 Public NGS data were obtained from Gene Expression Omnibus (GEO). GEO accession numbers of

346 the data are H3K4me1_1 (GSM788085), H3K4me1_2 (GSM733692), H3K27ac_1 (GSM733656),

347 H3K27ac_2 (GSM646434), p300 (GSM935401), GATA1 (GSM1003608), TAL1 (GSM935496),

348 BRG1 (GSM935633), SNF2h (GSM2424122), INI1 (GSM935634), BAF170 (GSM3634134), DNase-

349 seq (GSM816655), FAIRE-seq (GSM864361), ATAC-seq (GSM3452726) and MNase-seq

350 (GSM920557) in K562 cells and HeLa RNA-seq (GSM765402) in HeLa-S3 cells.

351

352 Statistical analysis

353 All qPCR results are presented as the means ± standard error of the mean (SEM) of 3 to 5 independent

354 experiments. The p-values were calculated using the two tailed Student's t-test (*P < 0.05). Boxplot data

355 is statistically analyzed using two tailed Student's t-test between two groups of classified enhancers. All

356 boxplots indicate log2-transformed (CPM + 1).

357

358 DATA ACCESS

359 All raw and processed sequencing data generated in this study have been submitted to the NCBI Gene

360 Expression Omnibus (GEO; https://www.ncbi.nlm.nih.gov/geo/) under accession number GSE147826.

361

362 ACKNOWLEDGEMENTS

363 We are grateful to Jaehoon Kim for kind gifts of antibodies against MLL3 and MLL4. This research

364 was supported by Basic Science Research Program through the National Research Foundation of Korea

365 (NRF) funded by the Ministry of Science, ICT and Future Planning (NRF-2014R1A2A1A11051702),

366 and by PNU-RENovation (2019-2020).

367 14

368 AUTHOR’S CONTRIBUTIONS

369 Y. Kang and A. Kim designed research; Y. Kang performed research; Y. Kang analyzed data; Y. Kang,

370 Y.W. Kim, J. Kang and A. Kim discussed results; Y. Kang and A. Kim wrote the manuscript.

371

372 COMPETING INTERESTS

373 The authors declare that they have no conflict of interest.

374

375 REFERENCES

376 Bolger AM, Lohse M, Usadel B. 2014. Trimmomatic: a flexible trimmer for Illumina sequence data.

377 Bioinformatics 30: 2114-2120.

378 Cho Y, Song SH, Lee JJ, Choi N, Kim CG, Dean A, Kim A. 2008. The role of transcriptional activator

379 GATA-1 at human β-globin HS2. Nucleic Acids Res. 36: 4521-4528.

380 Dobin A, Davis CA, Schlesinger F, Drenkow J, Zaleski C, Jha S, Batut P, Chaisson M, Gingeras TR.

381 2013. STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29: 15-21.

382 Dorighi KM, Swigut T, Henriques T, Bhanu NV, Scruggs BS, Nady N, Still CD, 2nd, Garcia BA,

383 Adelman K, Wysocka J. 2017. Mll3 and Mll4 Facilitate Enhancer RNA Synthesis and

384 Transcription from Promoters Independently of H3K4 Monomethylation. Mol. Cell 66: 568-

385 576.e564.

386 Eberharter A, Vetter I, Ferreira R, Becker PB. 2004. ACF1 improves the effectiveness of nucleosome

387 mobilization by ISWI through PHD-histone contacts. EMBO J. 23: 4029-4039.

388 Ebrahimi A, Sevinç K, Gürhan Sevinç G, Cribbs AP, Philpott M, Uyulur F, Morova T, Dunford JE,

389 Göklemez S, Arı Ş et al. 2019. Bromodomain inhibition of the coactivators CBP/EP300

390 facilitate cellular reprogramming. Nat. Chem. Biol. 15: 519-528.

391 Fang X, Yin W, Xiang P, Han H, Stamatoyannopoulos G, Li Q. 2009. The higher structure of chromatin

392 in the LCR of the β-globin locus changes during development. J. Mol. Biol. 394: 197-208.

393 Feng J, Liu T, Qin B, Zhang Y, Liu XS. 2012. Identifying ChIP-seq enrichment using MACS. Nat.

394 Protoc. 7: 1728-1740. 15

395 Filippakopoulos P, Picaud S, Mangos M, Keates T, Lambert JP, Barsyte-Lovejoy D, Felletar I, Volkmer

396 R, Müller S, Pawson T et al. 2012. Histone recognition and large-scale structural analysis of

397 the human bromodomain family. Cell 149: 214-231.

398 Freese NH, Norris DC, Loraine AE. 2016. Integrated genome browser: visual analytics platform for

399 genomics. Bioinformatics 32: 2089-2095.

400 Grossman SR, Engreitz J, Ray JP, Nguyen TH, Hacohen N. 2018. Positional specificity of different

401 transcription factor classes within enhancers. Proc. Natl. Acad. Sci. U. S. A. 115: E7222-E7230.

402 Guertin MJ, Lis JT. 2013. Mechanisms by which transcription factors gain access to target sequence

403 elements in chromatin. Curr. Opin. Genet. Dev. 23: 116-123.

404 He X, Fan HY, Narlikar GJ, Kingston RE. 2006. Human ACF1 alters the remodeling strategy of SNF2h.

405 J. Biol. Chem. 281: 28636-28647.

406 Herz HM, Mohan M, Garruss AS, Liang K, Takahashi YH, Mickey K, Voets O, Verrijzer CP, Shilatifard

407 A. 2012. Enhancer-associated H3K4 monomethylation by Trithorax-related, the Drosophila

408 homolog of mammalian Mll3/Mll4. Genes Dev. 26: 2604-2620.

409 Hilton IB, D'Ippolito AM, Vockley CM, Thakore PI, Crawford GE, Reddy TE, Gersbach CA. 2015.

410 Epigenome editing by a CRISPR-Cas9-based acetyltransferase activates genes from promoters

411 and enhancers. Nat. Biotechnol. 33: 510-517.

412 Hsieh CL, Fei T, Chen Y, Li T, Gao Y, Wang X, Sun T, Sweeney CJ, Lee GS, Chen S et al. 2014.

413 Enhancer RNAs participate in androgen receptor-driven looping that selectively enhances gene

414 activation. Proc. Natl. Acad. Sci. U. S. A. 111: 7319-7324.

415 Hu D, Gao X, Morgan MA, Herz HM, Smith ER, Shilatifard A. 2013. The MLL3/MLL4 branches of

416 the COMPASS family function as major histone H3K4 monomethylases at enhancers. Mol.

417 Cell. Biol. 33: 4745-4754.

418 Hu G, Schones DE, Cui K, Ybarra R, Northrup D, Tang Q, Gattinoni L, Restifo NP, Huang S, Zhao K.

419 2011. Regulation of nucleosome landscape and transcription factor targeting at tissue-specific

420 enhancers by BRG1. Genome Res. 21: 1650-1658.

421 Inukai S, Kock KH, Bulyk ML. 2017. Transcription factor-DNA binding: beyond binding site motifs. 16

422 Curr. Opin. Genet. Dev. 43: 110-119.

423 Jang Y, Broun A, Wang C, Park YK, Zhuang L, Lee JE, Froimchuk E, Liu C, Ge K. 2019. H3.3K4M

424 destabilizes enhancer H3K4 methyltransferases MLL3/MLL4 and impairs adipose tissue

425 development. Nucleic Acids Res. 47: 607-620.

426 Jin Q, Yu LR, Wang L, Zhang Z, Kasper LH, Lee JE, Wang C, Brindle PK, Dent SY, Ge K. 2011.

427 Distinct roles of GCN5/PCAF-mediated H3K9ac and CBP/p300-mediated H3K18/27ac in

428 nuclear receptor transactivation. EMBO J. 30: 249-262.

429 Johnson TA, Chereji RV, Stavreva DA, Morris SA, Hager GL, Clark DJ. 2018. Conventional and

430 pioneer modes of glucocorticoid receptor interaction with enhancer chromatin in vivo. Nucleic

431 Acids Res. 46: 203-214.

432 Kang J, Kim YW, Kim A. 2018. Histone variants H3.3 and H2A.Z are incorporated into the β-globin

433 locus during transcription activation via different mechanisms. Biochim. Biophys. Acta. Gene

434 Regul. Mech. 1861: 637-646.

435 Kang Y, Kim YW, Kang J, Yun WJ, Kim A. 2017. Erythroid specific activator GATA-1-dependent

436 interactions between CTCF sites around the β-globin locus. Biochim. Biophys. Acta. Gene

437 Regul. Mech. 1860: 416-426.

438 Kang Y, Kim YW, Yun J, Shin J, Kim A. 2015. KLF1 stabilizes GATA-1 and TAL1 occupancy in the

439 human β-globin locus. Biochim. Biophys. Acta. 1849: 282-289.

440 Kim A, Song SH, Brand M, Dean A. 2007. Nucleosome and transcription activator antagonism at

441 human β-globin locus control region DNase I hypersensitive sites. Nucleic Acids Res. 35: 5831-

442 5838.

443 Kim SI, Bresnick EH, Bultman SJ. 2009. BRG1 directly regulates nucleosome structure and chromatin

444 looping of the α globin locus to activate transcription. Nucleic Acids Res. 37: 6019-6027.

445 Kim TK, Hemberg M, Gray JM, Costa AM, Bear DM, Wu J, Harmin DA, Laptewicz M, Barbara-Haley

446 K, Kuersten S et al. 2010. Widespread transcription at neuronal activity-regulated enhancers.

447 Nature 465: 182-187.

448 Kim YW, Kim A. 2013. Histone acetylation contributes to chromatin looping between the locus control 17

449 region and globin gene by influencing hypersensitive site formation. Biochim. Biophys. Acta.

450 1829: 963-969.

451 Kim YW, Kim A. 2017. Deletion of transcription factor binding motifs using the CRISPR/spCas9

452 system in the β-globin LCR. Biosci. Rep. 37.

453 Lai F, Orom UA, Cesaroni M, Beringer M, Taatjes DJ, Blobel GA, Shiekhattar R. 2013. Activating

454 RNAs associate with Mediator to enhance chromatin architecture and transcription. Nature 494:

455 497-501.

456 Lange M, Kaynak B, Forster UB, Tonjes M, Fischer JJ, Grimm C, Schlesinger J, Just S, Dunkel I,

457 Krueger T et al. 2008. Regulation of muscle development by DPF3, a novel histone acetylation

458 and methylation reader of the BAF chromatin remodeling complex. Genes Dev. 22: 2370-2384.

459 Langmead B, Salzberg SL. 2012. Fast gapped-read alignment with Bowtie 2. Nat. Methods 9: 357-359.

460 Lee JE, Park YK, Park S, Jang Y, Waring N, Dey A, Ozato K, Lai B, Peng W, Ge K. 2017. Brd4 binds

461 to active enhancers to control cell identity gene induction in adipogenesis and myogenesis. Nat.

462 Commun. 8: 2217.

463 Lee JE, Wang C, Xu S, Cho YW, Wang L, Feng X, Baldridge A, Sartorelli V, Zhuang L, Peng W et al.

464 2013. H3K4 mono- and di-methyltransferase MLL4 is required for enhancer activation during

465 cell differentiation. Elife 2: e01503.

466 Li W, Notani D, Ma Q, Tanasa B, Nunez E, Chen AY, Merkurjev D, Zhang J, Ohgi K, Song X et al.

467 2013. Functional roles of enhancer RNAs for oestrogen-dependent transcriptional activation.

468 Nature 498: 516-520.

469 Liao Y, Smyth GK, Shi W. 2014. featureCounts: an efficient general purpose program for assigning

470 sequence reads to genomic features. Bioinformatics 30: 923-930.

471 Local A, Huang H, Albuquerque CP, Singh N, Lee AY, Wang W, Wang C, Hsia JE, Shiau AK, Ge K et

472 al. 2018. Identification of H3K4me1-associated proteins at mammalian enhancers. Nat. Genet.

473 50: 73-82.

474 Ma H, Naseri A, Reyes-Gutierrez P, Wolfe SA, Zhang S, Pederson T. 2015. Multicolor CRISPR labeling

475 of chromosomal loci in human cells. Proc. Natl. Acad. Sci. U. S. A. 112: 3002-3007. 18

476 Maruyama A, Mimura J, Itoh K. 2014. Non-coding RNA derived from the region adjacent to the human

477 HO-1 E2 enhancer selectively regulates HO-1 gene induction by modulating Pol II binding.

478 Nucleic Acids Res. 42: 13599-13614.

479 Melo CA, Drost J, Wijchers PJ, van de Werken H, de Wit E, Oude Vrielink JA, Elkon R, Melo SA,

480 Leveille N, Kalluri R et al. 2013. eRNAs are required for p53-dependent enhancer activity and

481 gene transcription. Mol. Cell 49: 524-535.

482 Mousavi K, Zare H, Dell'orso S, Grontved L, Gutierrez-Cruz G, Derfoul A, Hager GL, Sartorelli V.

483 2013. eRNAs promote transcription by establishing chromatin accessibility at defined genomic

484 loci. Mol. Cell 51: 606-617.

485 Nagarajan S, Hossan T, Alawi M, Najafova Z, Indenbirken D, Bedi U, Taipaleenmäki H, Ben-Batalla I,

486 Scheller M, Loges S et al. 2014. Bromodomain protein BRD4 is required for estrogen receptor-

487 dependent enhancer activation and gene transcription. Cell Rep. 8: 460-469.

488 Plank JL, Dean A. 2014. Enhancer function: mechanistic and genome-wide insights come together. Mol.

489 Cell 55: 5-14.

490 Rahnamoun H, Hong J, Sun Z, Lee J, Lu H, Lauberth SM. 2018. Mutant p53 regulates enhancer-

491 associated H3K4 monomethylation through interactions with the methyltransferase MLL4. J.

492 Biol. Chem. 293: 13234-13246.

493 Raisner R, Kharbanda S, Jin L, Jeng E, Chan E, Merchant M, Haverty PM, Bainer R, Cheung T, Arnott

494 D et al. 2018. Enhancer Activity Requires CBP/P300 Bromodomain-Dependent Histone

495 H3K27 Acetylation. Cell Rep. 24: 1722-1729.

496 Ramírez F, Ryan DP, Grüning B, Bhardwaj V, Kilpert F, Richter AS, Heyne S, Dündar F, Manke T. 2016.

497 deepTools2: a next generation web server for deep-sequencing data analysis. Nucleic Acids Res.

498 44: W160-165.

499 Rickels R, Herz HM, Sze CC, Cao K, Morgan MA, Collings CK, Gause M, Takahashi YH, Wang L,

500 Rendleman EJ et al. 2017. Histone H3K4 monomethylation catalyzed by Trr and mammalian

501 COMPASS-like proteins at enhancers is dispensable for development and viability. Nat. Genet.

502 49: 1647-1653. 19

503 Robinson MD, McCarthy DJ, Smyth GK. 2010. edgeR: a Bioconductor package for differential

504 expression analysis of digital gene expression data. Bioinformatics 26: 139-140.

505 Roe JS, Mercan F, Rivera K, Pappin DJ, Vakoc CR. 2015. BET bromodomain inhibition suppresses the

506 function of hematopoietic transcription factors in acute myeloid leukemia. Mol. Cell 58: 1028-

507 1039.

508 Sanjana NE, Shalem O, Zhang F. 2014. Improved vectors and genome-wide libraries for CRISPR

509 screening. Nat. Methods 11: 783-784.

510 Schaukowitch K, Joo JY, Liu X, Watts JK, Martinez C, Kim TK. 2014. Enhancer RNA facilitates NELF

511 release from immediate early genes. Mol. Cell 56: 29-42.

512 Smith E, Shilatifard A. 2014. Enhancer biology and enhanceropathies. Nat Struct Mol Biol 21: 210-219.

513 Wang C, Lee JE, Lai B, Macfarlan TS, Xu S, Zhuang L, Liu C, Peng W, Ge K. 2016. Enhancer priming

514 by H3K4 methyltransferase MLL4 controls cell fate transition. Proc. Natl. Acad. Sci. U. S. A.

515 113: 11871-11876.

516 Wang SP, Tang Z, Chen CW, Shimada M, Koche RP, Wang LH, Nakadai T, Chramiec A, Krivtsov AV,

517 Armstrong SA et al. 2017. A UTX-MLL4-p300 Transcriptional Regulatory Network

518 Coordinately Shapes Active Enhancer Landscapes for Eliciting Transcription. Mol. Cell 67:

519 308-321.e306.

520 Zentner GE, Tesar PJ, Scacheri PC. 2011. Epigenetic signatures distinguish multiple classes of

521 enhancers with distinct cellular functions. Genome Res. 21: 1273-1283.

522 Zhu Y, Li Y, Lou D, Gao Y, Yu J, Kong D, Zhang Q, Jia Y, Zhang H, Wang Z. 2018. Sodium arsenite

523 exposure inhibits histone acetyltransferase p300 for attenuating H3K27ac at enhancers in

524 mouse embryonic fibroblast cells. Toxicol. Appl. Pharmacol. 357: 70-79.

525 Zhu Y, Sun L, Chen Z, Whitaker JW, Wang T, Wang W. 2013. Predicting enhancer transcription and

526 activity from chromatin modifications. Nucleic Acids Res. 41: 10032-10043.

527

528 FIGURE LEGENDS

529 Figure 1. H3K4me1 and H3K27ac at the β-globin enhancer LCR HSs in MLL3/4 cells and p300

20 △ △

530 K562 cells. (A) The diagram represents the human β-globin locus. The β-globin LCR HSs and β-like

531 globin genes are indicated by yellow and red/brown rectangles, respectively. Black vertical bars

532 represent insulators. Protein levels of MLL3, MLL4 and p300 (B) and histone H3K4me1 and H3K27ac

533 (C) were analyzed by western blot in control cells (Con) and two clones of MLL3/4 cells and p300

534 cells. The β-tubulin, GAPDH and H3 ware used as loading control. (D) Transcription△ of the human△ γ-

535 globin genes were measured by quantitative RT-PCR. The amounts of cDNA for the γ-globin genes

536 were compared with a genomic DNA standard and then normalized to the Actin cDNA. The results of

537 4 to 7 independent experiments were graphed with ± SEM. Enrichments of histone H3K4me1 (E),

538 H3K4me2 (F) and H3K27ac (G) were measured using ChIP assay and quantitative PCR at the β-globin

539 LCR HSs. DNA immunoprecipitated by antibodies to H3K4me1, H3K4me2 and H3K27ac were

540 quantitatively compared with DNA reacted by H3 antibodies. (H) DNA immunoprecipitated by H3

541 antibodies was compared with input DNA. The Actin was used as positive control. Normal rabbit IgG

542 (IgG) served as negative experimental control. Results are presented as the means ± SEM of 3 to 5

543 independent experiments. The p-values were calculated using the two tailed Student's t-test (*P < 0.05).

544

545 Figure 2. Genome wide analysis of H3K4me1 and H3K27ac in MLL3/4 cells and p300 cells. (A)

546 Heatmaps show signal enrichment of H3K4me1 and H3K27ac at H3K4me1△ peaks in control△ cells (Con)

547 and MLL3/4 cells. H3K4me1 regions (n = 108,210) were divided into two groups; H3K4me1-reduced

548 regions△ in MLL3/4 cells (0 < Con CPM - MLL3/4 CPM, n = 96,928) and others (0 ≥ Con CPM -

549 MLL3/4 △CPM, n = 11,282). (B) Signal △ enrichment of H3K27ac and H3K4me1 was showed at

550 H3K27ac△ peaks in Con and p300 cells by heatmaps. H3K27ac regions (n = 112,261) were divided

551 into two groups; H3K27ac-reduced△ regions in p300 cells (0 < Con CPM - p300 CPM, n = 89,199)

552 and others (0 ≥ Con CPM - p300 CPM, n△ = 23,062). Five kilobase pairs△ around the center of

553 H3K4me1 or H3K27ac are displayed.△ Color scales indicate the relative signal intensity on heatmaps.

554 (C) DNase I and FAIRE enrichment were presented by heatmaps at DNase I peak (± 5 Kb from the

555 center) in K562 cells. Venn diagram depicts overlapping regions in DNase-seq and FAIRE-seq peaks

556 (n = 65,377, right top panel). The overlapping regions were classified into four groups; enhancers, 21

557 promoters, insulators and others. (D) H3K4me1 and H3K27ac ChIP signal (± 5 Kb from the center)

558 were presented by heatmaps at the four groups.

559

560 Figure 3. Histone depletion at enhancers in MLL3/4 cells and p300 cells. Histone H3 ChIP signal

561 (± 2 Kb from the center) was presented by heatmaps△ for four groups△ classified in Fig. 2C (A) and by

562 average profiles (B). (C) Putative enhancers (PEs, n = 9,201) were sorted by H3K4me1 signal in Con

563 cells and divided into three groups according to the signal (left). Average profiles of H3 signal (± 2 Kb

564 from the PEs center) were drawn for the groups (right). (D) Average profiles of chromatin accessibility

565 signal (± 2 Kb from the PEs center, left) and nucleosome depletion signal (± 1 Kb from the PEs center,

566 right) were drawn using public ATAC-seq and MNase-seq data. (E) Heatmaps present signal enrichment

567 of H3K4me1 and H3K27ac (± 5 Kb from the PEs center) in two PEs groups; H3K4me1-reduced in

568 MLL3/4 cells (0 < Con CPM - MLL3/4 CPM, n = 8,184) and H3K4me1-unreduced (0 ≥ Con CPM

569 -△ MLL3/4 CPM, n = 1,017). H3△ signal (± 2 Kb from the center) in two PEs groups was presented by

570 average△ profiles (lower). (F) Heatmaps present signal enrichment of H3K27ac and H3K4me1 (± 5 Kb

571 from the PEs center) in two PEs groups; H3K27ac-reduced in p300 cells (0 < Con CPM - p300

572 CPM, n = 5,551) and H3K27ac-unreduced (0 ≥ Con CPM - p300△ CPM, n = 3,650). H3 signal△ (± 2

573 Kb from the center) in two PEs groups was presented by average△ profiles (lower).

574

575 Figure 4. Occupancy of subunits of chromatin remodeling complexes at enhancers in MLL3/4 cells

576 and p300 cells. (A) BRG1, SNF2h, INI1 and BAF170 signal (± 5 Kb from the center)△ were presented

577 by heatmaps△ at PEs in K562 cells (WT). (B) IGB genome browser tracks show distribution of p300,

578 GATA1, TAL1, BRG1, SNF2h, INI1 and BAF170 at the β-globin locus. Enhancers are highlighted with

579 yellow shadow. The occupancy of BRG1 (C), SNF2h (D) and ACF1 (E) was analyzed by ChIP assay at

580 the β-globin enhancers and other erythroid specific enhancers in Con cells, MLL3/4 cells and p300

581 cells. The amount of immunoprecipitated DNA was quantitatively compared△ with input DNA.△ The

582 results of 4 independent experiments were graphed with ± SEM. The p-values were calculated using

583 the two tailed Student's t-test (*P < 0.05). (F) Expression levels of GATA1, TAL1, BRG1, SNF2h, ACF1, 22

584 INI1 and BAF170 were determined using total RNA-seq data. The y axis indicates log2-transformed

585 (CPM + 1).

586

587 Figure 5. eRNA transcription in MLL3/4 cells and p300 cells. (A) PEs were divided into active (n

588 = 4,600) and poised (n = 4,601△) according to H3K27ac△ signal in Con cells. Signal enrichment of

589 H3K27ac and H3K4me1 (± 5 Kb from the center), and eRNA transcription were presented using

590 heatmaps and boxplot, respectively, for both enhancer groups. (B) eRNA expression (± 2 Kb from the

591 center) from active enhancers (AEs) was depicted by average profiles in Con cells, MLL3/4 cells and

592 p300 cells. (C) Expression for the nearest genes of eRNA-decreased AEs in MLL3/4△ cells (upper)

593 and△ in p300 cells (lower) was analyzed by gene set enrichment analysis (GSEA).△ ‘Up’ and ‘down’

594 indicate△ the relative transcription level of genes in MLL3/4 cells and p300 cells compared to Con

595 cells. AEs were divided into three groups according △to H3K27ac signal (D)△ and H3K4me1 signal (E) in

596 Con cells. Each histone modification signal and eRNA transcription of the groups were presented in

597 average profile and boxplot, respectively. (F) PEs (n = 9,201) were divided into two groups according

598 to eRNA transcription; eRNA+ (0 < Con eRNAs CPM, n = 7,548) and eRNA- (0 = Con eRNAs CPM,

599 n = 1,653). Average profiles represent H3K27ac and H3K4me1 signals in two enhancer groups. (G)

600 eRNA-transcribing enhancers (n = 7,548) were divided into three groups according to the eRNA signal

601 in Con cells. Boxplot depicts expression level of eRNAs at the groups. Average signals of H3K27ac

602 and H3K4me1 (± 5 Kb from the center) were profiled on each groups. The y axis of all boxplots

603 indicates log2-transformed (CPM + 1).

604

605 Figure 6. eRNA transcription from the β-globin LCR HSs in TSA-treated MEL/ch11 cells. (A) Protein

606 levels of H3K27ac, H3K4me1 and H3 were detected by Western blot in control and TSA-treated cells.

607 The H3 was used as experimental control. H3K27ac (B) and H3K4me1 (E) at chromatin level were

608 analyzed by ChIP assay in the β-globin LCR HSs and the promoter and exons of the β-globin gene.

609 eRNA from the LCR HSs (C) and mRNA from the β-like globin genes (D) were measured by

610 quantitative RT-PCR. The results of 4 to 5 independent experiments were graphed with ± SEM. The p- 23

611 values were calculated using the two tailed Student's t-test (*P < 0.05).

612

613 Figure 7. Effects by depletion of H3K4me1 and H3K27ac on enhancers. Results from all Figures were

614 combined and represented in diagrams. Depletion of H3K4me1 at enhancers impairs H3K27ac, histone

615 depletion, recruitment of chromatin remodeling complexes, eRNA transcription and mRNA

616 transcription of target gene. Depletion of H3K27ac at enhancers decreases eRNA transcription and its

617 target gene transcription without affecting H3K4me1, histone depletion and recruitment of chromatin

618 remodeling complexes.

Fig. Fig. B E G A 250 250 H3K27ac / H3 H3K4me1 / H3 bioRxiv preprint kDa kDa 1 10.0 0.0 3.0 6.0 9.0 0.0 2.0 4.0 6.0 8.0 → → . (which wasnotcertifiedbypeerreview)istheauthor/funder.Allrightsreserved.Noreuseallowedwithoutpermission. * * β - MLL4 MLL3 * tubulin doi: * * * https://doi.org/10.1101/2021.01.05.425373 * * * * GAPDH * p300 * ∆p300 K4me1 ∆p300 IgG ∆p300 K4me1 ∆MLL3/4 IgG ∆MLL3/4 ConK4me1 ConIgG ∆p300 K27ac ∆p300 IgG ∆p300 K27ac ∆MLL3/4 IgG ∆MLL3/4 ConK27ac ConIgG C H3K4me1 H3K27ac H3 F H ; this versionpostedJanuary5,2021. H3 / Input H3K4me2 / H3 0.0 2.0 4.0 6.0 8.0 0.0 2.0 4.0 6.0 * * D cDNA / Actin * * 0.0 0.5 1.0 1.5 * * * ∆MLL3/4 ∆MLL3/4 #2 ∆MLL3/4 #1 Con * * * * The copyrightholderforthispreprint * 0.0 0.5 1.0 1.5 ∆p300 K4me2 ∆p300 IgG ∆p300 K4me2 ∆MLL3/4 IgG ∆MLL3/4 ConK4me2 ConIgG * ∆p300 H3 ∆p300 IgG ∆p300 H3 ∆MLL3/4 IgG ∆MLL3/4 ConH3 ConIgG * ∆p300 #2 ∆p300 #1 Con * * bioRxiv preprint doi: https://doi.org/10.1101/2021.01.05.425373; this version posted January 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Fig. 2.

A B H3K4me1 H3K27ac H3K27ac H3K4me1

Con ∆MLL3/4 Con ∆MLL3/4 Con ∆p300 Con ∆p300 n = n 89,199 n = 96,928n (n = 112,261)(n (n = 108,210)(n Sorted by H3K27ac by Sorted Sorted by H3K4me1 by Sorted

-5 0 +/-5 0 +/-5 0 +/-5 0 +5 -5 0 +/-5 0 +/-5 0 +/-5 0 +5

0 8 0 8 0 12 0 12 0 12 0 12 0 8 0 8 Distance from H3K4me1 peak (Kb) Distance from H3K27ac peak (Kb)

C D DNase I H3K4me1 H3K27ac

Con ∆MLL3/4 ∆p300 Con ∆MLL3/4 ∆p300

FAIRE Enhancers (n = 170,026) (n = 9,201) DNase I (n = 112,025) 65,377 (n = 112,025)(n Others Sorted by DNase I DNase by Sorted (n = 20,492)

-5 0 +5

0 1 FAIRE Promoters (n = 12,900) (n = 65,377)(n Sorted by DNase I DNase by Sorted

Insulators (n = 22,784) (n =112,025) (n Sorted by DNase I DNase by Sorted

-5 0 +/-5 0 +/-5 0 +/-5 0 +/-5 0 +/-5 0 +5 -5 0 +5 0 8 0 8 0 8 0 12 0 12 0 12 0 0.1 Distance from center (Kb) Distance from DNase I peak (Kb) bioRxiv preprint doi: https://doi.org/10.1101/2021.01.05.425373; this version posted January 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Fig. 3.

H3 B A Con ∆MLL3/4 ∆p300 Con ∆MLL3/4 ∆p300 H3 H3 Enhancers 1.5 1.4 (n = 9,201) Putative enhancers Others 1.3 1.2 Others 1.1 (n = 20,492) 1.0 0.9 Average signal Average 0.7 0.8 Promoters -2 0 2 -2 0 2 (n = 12,900) H3 H3 (n = 65,377)(n 1.2 Promoters 1.8 Insulators Sorted by DNase I DNase by Sorted 1.5 0.9 Insulators (n = 22,784) 1.2 0.6 0.9 Average signal Average -2 0 +/-2 0 +/-2 0 +2 0.3 0.6 -2 0 2 -2 0 2 0 2 0 2 0 2 Distance from center (Kb) Distance from center (Kb) C D Putative Enhancers (PEs) H3 Chromatin Nucleosome accessibility 1.6 25 1.4 Group1 20 1.3 1.2 15 1 10 Group2 1 0.7 5 Average signal Average Average signal Average (n = 9,201)(n 0.4 0 0.8

Sorted by H3K4me1 by Sorted Group3 -2 0 2 -2 0 2 -1 0 1

Distance from PEs (Kb) Distance from PEs (Kb) Group1 Group2 Group3

F E H3K4me1 H3K27ac H3K27ac H3K4me1

Con ∆MLL3/4 Con ∆MLL3/4 Con ∆p300 Con ∆p300 n = 5,551n n = n 8,184 (n = 9,201)(n (n = 9,201)(n Sorted by H3K27ac by Sorted Sorted by H3K4me1 by Sorted

-5 0 +/-5 0 +/-5 0 +/-5 0 +5 -5 0 +/-5 0 +/-5 0 +/-5 0 +5

0 10 0 10 0 20 0 20 0 20 0 20 0 10 0 10

H3 H3

H3K4me1 - H3K4me1 - H3K27ac - H3K27ac - reduced enhancers unreduced enhancers reduced enhancers unreduced enhancers 1.5 1.5 1.5 1.5

1.3 1.3 1.3 1.3

1.1 1.1 1.1 1.1

0.9 0.9 0.9 0.9 Average signal Average Average signal Average 0.7 0.7 0.7 0.7 -2 0 2 -2 0 2 -2 0 2 -2 0 2 Distance from center (Kb) Distance from center (Kb) Con ∆MLL3/4 ∆p300 Fig. Fig. E C A

ACF1 / Input BRG1 / Input Sorted by H3K4me1 bioRxiv preprint 4 0.0 0.1 0.2 0.3 0.4 0.0 0.1 0.2 0.3 (n = 9,201) . - 0 12 0 12 0 12 0 12 5 BRG1 * (which wasnotcertifiedbypeerreview)istheauthor/funder.Allrightsreserved.Noreuseallowedwithoutpermission. β β HS4 HS4 +/ 0 - - globinLCR HSs globinLCR HSs * Distance from center (Kb) doi:

HS3 HS3 - 5 0 +/ 0 5 SNF2h https://doi.org/10.1101/2021.01.05.425373 * HS2 HS2 WT * * - HS1 HS1 5 INI1 * 0 +/ 0 * erythroid enhancers erythroid enhancers Ppox Ppox - * * 5 0 +5 0 5 ∆p300 ∆p300 BRG1 ∆MLL3/4 BRG1 Con BRG1 ∆p300 ∆p300 ACF1 ∆MLL3/4 ACF1 Con ACF1 Ermap Ermap BAF170 * Slcc2a4 Slcc2a4 * * Gypc Gypc B BAF170 GATA1 SNF2h BRG1 ; TAL1 D F p300 this versionpostedJanuary5,2021. INI1 Gene expression SNF2h / Input 0.10 0.15 0.20 log2(CPM + 1) 0.00 0.05 10 0 2 4 6 8 NS NS β HS4 - globinLCR HSs NS HS3 NS Con * NS HS2 NS ∆MLL3/4 *

NS HS1 The copyrightholderforthispreprint NS * erythroid enhancers Ppox NS ∆p300 NS * Ermap ∆p300 SNF2h ∆MLL3/4 SNF2h Con SNF2h NS NS Slcc2a4 NS * NS Gypc bioRxiv preprint doi: https://doi.org/10.1101/2021.01.05.425373; this version posted January 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Fig. 5.

A B C Con eRNAs eRNAs Nearest genes of eRNAs-decreased enhancers H3K27ac H3K4me1 1.9E-145 Con ∆MLL3/4 ES : -0.33 ∆p300 NES : -1.64 2.5 FDR : 0.00 Active Active

(n = 4,600) (n 2 1.5 (Con CPM+1) (Con 2 1 (n = 9,201)(n

Log 0.5 Poised Poised Sorted by H3K27ac by Sorted Average signal Average 0 ES : -0.24

-2 0 2 (ES) score Enrichment NES : -1.35 -5 0 +/-5 0 +5 FDR : 0.00 Distance from AEs (Kb) 0 20 0 10 Distance from center (Kb)

D E Active H3K27ac eRNAs Active H3K4me1 eRNAs Enhancers (AEs) Enhancers (AEs)

G1 4.07E-21 G1 NS G1 G2 G1 G2 G3 1.47E-17 G3 40 12 30 10 G2 G2 8 20 (Con CPM+1) (Con 6 CPM+1) (Con 2 2 (n = 4,600)(n (n = 4,600)(n 10 4 Log Log Average signal Average Average signal Average G3 Sorted by H3K27ac by Sorted G3 0 H3K4me1 by Sorted 2 -5 0 5 -5 0 5 Distance from AEs (Kb) Distance from AEs (Kb)

F H3K27ac G eRNAs-transcribing eRNAs H3K27ac H3K4me1 15 enhancers

10 G1 G1 G1 G2 G2 5 20 G3 8 G3 15 6 0 eRNAs G2 -5 0 5 10 4 (Con CPM+1) (Con H3K4me1 2 8 = 7,548)(n 5 2 Log Average signal Average Sorted by by Sorted Average signal Average 6 G3 0 0 -5 0 5 -5 0 5 4 Distance from center (Kb) 2 0 -5 0 5

Distance from PEs (Kb) bioRxiv preprint doi: https://doi.org/10.1101/2021.01.05.425373; this version posted January 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Fig. 6.

A B Con IgG Con K27ac 8.0 6h TSA IgG 6h TSA K27ac 24h TSA IgG 24h TSA K27ac

6.0 * * * * * * * * 4.0 * * * *

H3K27ac / H3 / H3K27ac 2.0

H3K27ac H3K4me1 H3 0.0

C D E Con Con Con IgG Con K4me1 6h TSA 6h TSA 6h TSA IgG 6h TSA K4me1 24h TSA 24h TSA 8.0 24h TSA IgG 24h TSA K4me1 0.8 12.0

3 * * * 6.0

10 * *

* 0.6 * 9.0 * * *

mAct 4.0 0.4 6.0 mAct * 2.0 H3K4me1 / H3 / H3K4me1

0.2 / cDNA 3.0

cDNA / cDNA 0.0 0.0 0.0 bioRxiv preprint doi: https://doi.org/10.1101/2021.01.05.425373; this version posted January 5, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Fig. 7.

ΔH3K4me1 ΔH3K27ac

H3K4me1 H3K27ac

Chromatin eRNA mRNA remodeling complex