41588_2024_2000_MOESM1

nature genetics

https://doi.org/10.1038/s41588-024-02000-5

Technica� Report

ChIP-DIP maps binding of hundreds of

proteins to DNA simultaneously and

identifies diverse gene regulatory elements

In the format provided by the

authors and unedited

Supp�ementary information

SUPPLEMENTA

FIGURES

Supplementa

Figure

: ChIP

DIP allows for antibody screening. (A)

Comparison of signal

enrichment for two distinct antibodies targeting SETD2 and two distinct antibodies targeting TBP

mapped within the same pool (52 K562

Antibody

pool) across a genomic region (hg 38;

chr12:6,400,000

7,020,000)

(B)

Metaplot of signal distribution at promoters for two antibodies

targeting TBP shown

(A).

(C)

Metaplot of signal distribution at promoters for the two

antibodies targeting SETD2 shown

(A).

Supplementa

Figure

: ChIP

DIP is robust using various amounts of antibody. (A)

Histogram of

earson correlation coefficients for linear regression of antibody amount vs.

chromatin yield for each individual antibody (n=88).

(B)

Histogram of slope for linear regression

of antibody amount vs. chromatin yield for each individual antibody (n=88).

(C)

Lineplot of

antibody amount (x

axis) versus relative chromatin yield (y

axis). To allow for comparison of

antibodies with various chromatin yields, chromatin yield was normalized to the num

ber of

sequenced reads at the maximum amount of antibody (1.5ug) for each antibody. Only antibodies

with

earson

R>0.95 (n=77) are plotted. Individual data points are shown as x’s. The median for

all antibodies is shown as a solid black line. The

mean +/

is shown as a shaded grey region.

(D)

Lineplot of antibody amount (x

axis) versus read enrichment at known peak sites (y

axis)

(n=35, see

Supplementary

Methods

). To allow for simultaneous plotting of all antibodies, read

enrichment was normalized to the mean read enrichment at per antibody.

The

mean +/

is shown

as a shaded grey region.

Supplementa

Figure

: Normalization of chromatin yield by titrated bead pooling. (A)

Barplot for average chromatin reads per bead associated with each antibody in the mESC 67

Antibody Pool experiment.

(B)

Barplot

for total chromatin yield per antibody in the mESC 67

Antibody Pool experiment, which was performed using titrated bead pooling.

Supplementary Figure 4:

Comparison of protein localization across ChIP

DIP pools of

various protein number and composition

after downsampling to equalize read numbers

)

Correlation heatmap as shown in Figure 2B after downsampling each target to the same number

of reads for each sample

(see

Methods

and

Supplementary

Methods

)

(

)

Heatmap of subtracted

difference between the correlation heatmap shown in (

) and the heatmap shown in

Figure 2B

(

)

Comparison of the relative number of peaks called (y

axis) for each target (x

axis) after read

number equalization. Reference peaks were defined as the number of peaks called in the smallest

pool, the K562 10 A

ntibody

ool

experiment

Supplementa

Figure

: Comparison of protein localization across different amounts of cell

lysate. (A)

Comparison of RNAP II NTD localization across a snRNA gene cluster (hg38,

chr17:58,620,000

58,689,000) generated using various amounts of input K562 cell lysate.

(B)

Comparison of H3K27me3 localization across a genomic region (hg38, chr1:23,850,000

25,850,000) generated using various amounts of input K562 cell lysate.

(C)

Comparison of

localization of

various isoforms of RNAP II at the

EGR1

locus (hg38, chr5:138,455,000

138,480,000) generated using various amounts of input K562 cell lysate.

Supplementary

Figure

Comparison of peak overlap across different amounts of cell lysate.

(A)

Lineplot of percentage peak overlap between

peaks called in ChIP

DIP experiment (35

Antibody Pool) using

lysate from

45 million cell

equivalents

various lower amounts of cell

lysate

(left)

Lineplot of percentage peak overlap between

peaks called in ChIP

DIP experiment

(35 Antibody Pool) using

lysate from

5 million cell

equivalents

various lower amounts of cell

lysate

(right)

(

)

Barplot of maximum percentage peak overlap between

peaks called in

experiment using

lysate from

50,000 cell

equivalents

either 45 or 5 million cell

equivalents

Supplementary

Figure

. Comparison of library complexity for ChIP

DIP at various levels

of cellular input versus CUT&TAG. (A

)

Complexity curves for various targets. For ChIP

DIP, complexity curves are only shown for a given level of input if the sequencing depth was

sufficient to estimate the curve (see

Supplementary

Methods

Supplementa

Figure

. Comparison of FRIP scores at various levels of cellular input

versus CUT&TAG.

(A)

Scatterplot of FRIP (fraction reads in peaks) vs peak number for

CUT&TAG (left) and ChIP

DIP (right) at various cell numbers or levels of cellular input,

respectively. For ChIP

DIP, only targets with more than 100 peaks were evaluated (see

Methods

Supplementa

Figure

. Simultaneous mapping of all five chromatin states within a single

experiment. (A

Visualization of histone modifications representing various functional

chromatin states that were all mapped within the same ChIP

DIP pool (K562 35 Antibody Pool)

along a genomic region (A:hg38, chr3:120,500,000

130,500,000, B:hg38, chr4:75,000,000

90,000,

000, C:hg38, chr3:137,500,000

153,500,000, D:hg38, chr5:108,000,000

120,000,000 ).

Supplementary

Figure

: De

novo

identification

of transcription factor motifs

(A)

novo

identification

(left) and

the

known

canonical

(right) motifs for transcription factors mapped in

K562 (top) or mESC (bottom). The

de novo

motifs are highlighted in red within the known

canonical motifs.

Transcription factors mapped in the same experiment are indicated by brackets

on the right.

Supplementary

Figure

: Chromatin states corresponding to distinct RNA polymerases. (A)

Comparison of histone profiles for H3K4me2, H3K9ac and H3K56ac at the promoters of RNAP

I, II and III, similar to

Figure 6A

. (Left) Histone modification over the RNAP I

transcribed rDNA

spacer promoter. (Middle/Right) Metaplot of histone profiles at active (blue) and inactive (gray)

promoters for RNAP II (middle) and RNAP III (right).

Supplementary

Figure

: ChIP

DIP detects canonical bivalent domains in mESC. (A)

Metaplot of H3K27me3 (left column) and H3K4me3 (right column) histone profiles at known

bivalent H3K4me3/H3K27me3, known monovalent and unmodified promoters in mESC for

ChIP

DIP (top row) versus traditional ChIP

seq (bottom row)

(B)

Track visualization of H3K4me3 and

H3K27me3 for ChIP

DIP versus traditional ChIP

seq at the known bivalent

gene cluster

(mm10: chr2:74,555,000

74,855,500 )

(C)

Track visualization of H3K4me3 and H3K27me3 for

ChIP

DIP vers

us traditional ChIP

seq at the known bivalent

Lhx2

gene (mm10: chr2:38,249,500

38,550,000)

Supplementary

Figure

: ChromHMM model using histone acetylation marks

(A)

Histone

acetylation mark emission probability matrix for 19

state ChromHMM model. State annotations

(right) were assigned manually based on genomic position enrichments of states.

(B)

Track

visualization of histone acetylation marks (top) and chromatin state annotations (bottom) at

example promoter region (left) versus example intergenic region (right). Histone acetylation marks

are scaled to the same maximum value at both regions. At each re

gion, the chromatin states that

are present are shown with solid lines and a box indicating the exact position; chromatin states that

are absent are listed next to dotted lines.

(C)

Heatmap of genome annotation enrichment of

chromatin states. Enrichment sc

ores are normalized to the maximum and minimum of each

column.

(D)

Heatmap of genomic position enrichment relative to the TSS of chromatin states.

Enrichment scores are normalized to the maximum and the minimum of the heatmap.

SUPPLEMENTARY

NOTES

Note S1. Minimal inter

bead mixing during ChIP

DIP ensures accurate protein

DNA

assignment.

We considered three potential issues in our ChIP

DIP experimental system that could

lead to misassignment between antibodies and chromatin: (1) antibody

ID oligo movement, (2)

antibody movement, (3) chromatin movement (

Extended

Figure

). To evaluate each of these

possibilities, we designed experiments to estimate their frequency.

Antibody

ID oligo movement

: If antibody

ID oligos were to dissociate from their original

bead and bind to other beads during the ChIP

DIP procedure, then multiple antibody

oligo types would share a single split

pool barcode, leading to errors in assigning a

chromatin region to the correct protein. If this were the case, we would expect to observe

clusters that contain multiple distinct antibody

ID oligos. To explore this,

for each split

pool barcode

we calculated the proportion of antibody ID oligo reads corresponding to the

maximumly represented antibody

ID oligo type. In a representative experiment, we

observed that 96% of clusters had only a single type of antibody

ID oligo (100% maximum

representation) and 99.4% of

clusters

had at least 80% maximum representation (

Extended

Figure 1C

). Since most split

pool barcodes have unique representation, this suggests that

antibody

ID oligo movement occurs infrequently and should not significantly impact the

accuracy of antibody

chromatin assignments.

Antibody movement

: If antibodies for protein X dissociated from their bead and

reassociated with a distinct protein G bead containing a label for an antibody recognizing

protein Y, then chromatin captured by that antibody would be improperly assigned to

protein Y. To quant

ify the frequency of such events, we performed a ChIP

DIP experiment

where we added

oligo

labeled protein G beads that were not

coupled

to an antibody or

were

coupled

to an IgG antibody and measured the amount of chromatin that was assigned to

these beads. In all cases, we observed minimal detection of chromatin (<0.5%)

on control

beads

. Specifically, we performed a ChIP

DIP experiment with oligo

labeled beads

containing a CTCF antibody, an isotype control IgG antibody, or no antibody (empty). Only

beads containing an antibody (

i.e.,

CTCF, IgG) were present during the IP stages and empty

beads were added in various post

IP, pre

split pool processing steps to determine the

frequency of mixing at each step. If antibodies moved between beads during post

processing, we would expect to find chromatin associated with empty beads. When we

measured the amount of chromatin assigned to each bead, we observed that beads with the

G control antibody received 0.10% (1/1000

) the amount of chromatin compared to the

CTCF antibody (

Extended

Figure 1E

). Empty beads added during the end repair reaction

and dA

tailing reaction received 0.40% and 0.10% the amount of chromatin of the CTCF

antibody beads, respectively. We note that these estimates are likely to be an overestimate

of the true mixing rate bec

ause this experimental design does not distinguish between

chromatin associating due to antibody movement

and

chromatin that may non

specifically

bind to IgG or empty protein G beads. Nonetheless, these results indicate that the impact

of antibody movement on chromatin assignment is minimal, representing no more than

0.5% of all chromatin captured.

Chromatin movement

: If proteins dissociate from their epitope

specific antibodies post IP,

they may specifically bind to other beads containing the same epitope

specific antibodies.

If this movement occurs during the split

pool process, then these chromatin fragments

would

not be assigned because they would lack a paired antibody

ID oligo with the same

barcode.

If this movement occurs prior to split

pool during DNA processing steps, the

chromatin fragments could still be assigned to an antibody.

We estimated the frequency of

chromatin movement using a human

mouse mixing experiment in which we separately

IP’d a human chromatin sample and a mouse chromatin sample using identical pools of

labeled beads. After the IP, we mixed the samples and performe

d the remainder of the

standard ChIP

DIP protocol (DNA processing and split

pool). If chromatin dissociates and

rebinds prior to split

pool then we would expect that it could rebind to beads from the other

species containing the same antibody type as its o

riginal bead. We observed that only ~5%

of reads were assigned to beads of the other species (4.2% and 5.8%,

Extended

Figure

). This suggests that disassociation and reassociation of chromatin

crosslinked proteins

from their epitope

specific antibodies during ChIP

DIP processing steps is minimal.

Nonetheless, in the cases where this does occur it would impact the assignability o

f these

chromatin fragments (sensitivity of detection) rather than resulting in incorrect assignment

(specificity of detection).

Note S2. Guidelines for ChIP

DIP experimental design

Designing

a ChIP

DIP experiment requires selection of multiple experiment

specific parameters

related to the number and identity of antibodies, the execution of split

and

pool barcoding

and

final library sequencing. Below are

guid

elines

to aid the reader in

successfully

planning and

performing

ChIP

DIP

experiments:

(i)

Antibody selection and screening:

To optimize the quality of antibodies used in a ChIP

DIP,

we recommend an initial low

depth

, high throughput screening experiment of candidate antibodies

before proceeding to a deeply

sequenced ChIP

DIP experiment

(see mESC 228 Antibody Pool in

Supplementary

Methods

)

These initial screening

results can be evaluated for both chromatin

yield and

, when references peak sites are available,

peak enrichment per antibody. Antibodies with

poor chromatin capture or that fail to

enrich known binding sites

should be

discarded from future

experiments. Because ChIP

DIP allows

for

use of

multiple antibodies to the same target

protein

within the same pool, it can be used to identify the best performing antibody from a set of

antibodies

(

Supplementa

Figure

)

(ii) Antibody amounts:

To investigate how ChIP

DIP performs with varying amounts of

antibody, we ran parallel 87

ntibody pool ChIP

DIP experiments where we titrated the amount

of each antibody used (1.5ng, 0.75ng, or 0.15ng for each antibody, respectively, see

Supplementary

Methods

). Across this 10

fold range of antibody amount, we found that most

antibodies (77/88) had a linear increase in chromatin yield (

Supplementa

Figure

) without

a significant change in read enrichment at known peak sites (

Supplementa

Figure

indicating that enrichments measured by ChIP

DIP are robust to variations in antibody amount. In

summary, ant

ibody amounts primarily impact the amount of chromatin captured (yield) but not

the level of enrichment (specificity).

These results indicate that the enrichment measured is

generally robust to the precise amount of antibody used within this range.

As a result, the amount

of each antibody can be determined according to the titrated bead pooling method for read

coverage equalization (see

Supplementary

Methods

(iii) Antibody pool design:

To profile proteins of various abundances, ChIP

DIP employes a

titrated bead pooling strategy (see

Supplementary

Methods

). This allows for antibodies whose

capture efficiency differs by two orders of magnitude (>200X) to be pooled together while

reducing the discrepancy in the resulting per

target sequencing depth (

Supplementa

Figure

However, for antibodies with capture efficiencies differing by greater than three orders of

magnitude (>1000X), we would recommend performing multiple ChIP

DIP experimen

ts, one for

high capture antibodies and one for low capture antibodies, for optimal results.

(iv) Bead Amount:

The bead scale of the experiment

(e.g., number of beads) is selected to ensure

a minimum yield per target. For any given target, we aim for at least 1.5X the minimum number

of usable fragments required by ENCODE standards if the entire library is sequenced (1.5*20

million = 30million). T

o ensure this, we make a conservative assumption of 2 chromatin

reads/bead, which means ~15 million beads and corresponds to at least 5uL of protein G beads

(2.7 million beads/uL) per target. Using titrated

bead pooling (

see

Supplementary

Methods

antibodies with greater chromatin reads/bead have the corresponding number of beads

appropriately scaled down. To increase the total complexity of the experiment, a greater number

of starting beads per target (e.g., 10uL) has also been successfully used

(v) Lysate Amount:

As described in more detail in the main text, ChIP

DIP has been successfully

performed with the lysate corresponding to as little as 50,000 cells. While we observed

a reduced

correlation

of genome wide ChIP signal

for several proteins when generated from low cell inputs

relative to high cell inputs (

Figure 2F

)

, this appears to be due to

a reduction in the amount of

starting lysate which results in a lower final library complexity per target (

Supplementa

Figure

). This reduced yield can impact sensitivity (e.g., number of peaks detected,

Supplementary

gure

) but appears to have minimal impact on specificity (e.g., true vs false peaks detected,

Supplementary

Figure

Specifically,

here are two

possible reasons

that the correlation would decrease in such an

experiment.

Specificity

: Having less cell input and therefore higher antibody to chromatin ratios might

lead to an increase in non

target binding. This would be observed as an increase in the

number of false

positive peaks detected at lower cell numbers. However, in practice, we

do not see this. While we observe fewer significant peaks, virtually all of the peaks that we

detect at these lower cell numbers are also observed in the higher cell nu

mber. For example,

for most proteins

–

including CTCF, H3K4me3 and H3K27me3

–

we observed that

>90%

of peaks detected at 5 x 10

cells are also detected as peaks at higher cellular inputs

(

Supplementary

Figure

Sensitivity

: Having less cell input might lead to lower yield and detection of on

target

protein capture. This would be observed as a decrease in the amount of chromatin captured

for each given protein and a decrease in the number of peaks called at low cell numbers.

practice, this is what we see. For example, we observed that the amount of chromatin

captured for H3K27me3 and RNA Pol II was ~20 times higher when purifying from 5x10

cells compared to 5 x 10

cells (

Supplementary

Table 1,

Supplementary

gure

(vi)

Split

and

pool barcoding complexity:

ChIP

DIP utilizes split

and

pool barcoding to pair

DNA fragments and the corresponding antibody oligonucleotide on the same bead with a shared

barcode. This strategy relies on the fact that each bead receives a unique barcode during split and

pool barcodin

g. To ensure that the probability of shared barcodes on molecules within the same

cluster is high and the probability of a “collision” (two or more beads receiving the same barcode

by chance) is extremely low, the exper

iment is designed such that the total number of possible

barcodes is in vast excess to the number of beads. For a given number of beads (b), we recommend

selecting the number of split

and

pool tags (t) and split

pool rounds (r) such that t

> 10*b. For

more precise selection of parameters, we provide a calculator to estimate the collision frequency

for a given b, t and r based on sampling from a Poisson distribution

alongside our computation

pipeline at

https://github.com/GuttmanLab/chipdip

pipeline

) Sequencing depth:

We recommend an initial low

depth screening sequence run to determine

the distribution of reads across antibodies, estimate the complexity of the experiment and confirm

quality through ChIP

DIP specific quality checks (e.g., empirical cumulative distributi

on of the

proportion oligo reads in each cluster that correspond to the most frequently represented antibody

ID in that cluster). Because individual antibodies cannot be sequenced independently, as the pool

size increases, the number of total sequencing re

ads needed to characterize each antibody increases

as well. We find that the number of sequencing reads scales approximately linearly with the size

of the pool. To confirm that larger pools do not require greater sequencing depth per target, we

downsampled the experiments shown in

Figure 2B

to the same number of reads per target across

all pool sizes. We found that the downsampled tracks were highly correlated (

Supplementary

Figure

), with Pearson correlation coefficients largely unchanged from the initial analysis

(

Supplementary

Figure

), and that there was no trend between the number of peaks called and

the size of the original pool (

Supplementar

Figure

)

iii

) Sample multiplexing:

Split

and

pool barcoding can be used for multiplexing across

different samples (see K562 10, 52 and 35 Antibody Pool, mESC 228, 67 and 165 Antibody Pool,

and mDC 25 Antibody Pool in

Supplementary

Methods

). This can be used to rapidly optimize

experimental conditions (e.g., IP duration, crosslinking conditions, cell numbers), expand the size

of the antibody pool beyond a single 96

well plate, generate biological replicates or simultaneously

profile differe

nt biological samples (e.g., multiple time points in a time course).

Note S3. ChromHMM Model of histone acetylation states.

To investigate the spatial

relationships between histone acetylation marks, we generated a 19

state genome segmentation

model using ChromHMM and 15 different histone acetylation marks (

Supplementary

Figure

). Based on the transition probabilities, we grouped these 19 states and found two large sets

that differed in their genome positioning: set 1 (States 1

8) was promoter proximal, with the

individual states identifying the relationship to the TSS, w

hile set 2 (States 9

16) was promoter

distal, with the individual states demarcating the acetylation peaks and surroundings between them

(

Supplementary

Figure

). State 17 was also promoter proximal but was grouped separately

because of a unique signature with

H2AZac

(

Supplementary

Figure

). Finally, States 18

and 19 corresponded to silent genic and intergenic regions. Remarkably, this model found that

histone

acetylation marks were sufficient to define multiple functional elements (e.g

promoters,

enhancers, gene bodies, silent, intergenic).

Overall, our state

model found that the acetylation marks cover similar genomic loci; there exist

multiple states that have nearly all the marks (

i,e.,

State 1, 2, 9 and 10) and multiple states that

appear alike in composition but differ in relationship to genomic annotations (

i.e.,

State 1 vs State

9) (

Supplementary

Figure

). Comparing sets 1 and 2 (

i.e.,

promoter proximal vs promoter

distal), we found that all acetylation marks are present in both sets, however, some marks are more

enriched in one set over the other.

For example,

H3K9ac

was strongest in set 1 (the promoter

proximal set), while

H3K18ac, H3K27ac

and

H2BK20ac

were enriched throughout set 2 (the

promoter distal set) (

Supplementary

Figure

). Notably

H3K18ac, H3K27ac

and

H2BK20ac

are all targets of the CBP/p300 acetyltransferase

, which is strongly associated with activity at

enhancers

. In

contrast, the GCN5/PCAF subfamily preferentially acetylates H3K9, H3K14 and

–

all marks we see preferentially in set 1. Comparing all 19 states individually, we found that

a small subset of states had greater selective enrichment for specific histone modifications. For

example, State 5 (promoter up/down stream) and State 7 (gene

body) both had greater selectivity

for

H3K9ac

, State 14 was defined by

H2BK20ac

and State 17 was defined by

H2AZac

. Such states

may indicate subtle positional shifts or locations unique to these marks. Correspondingly, by visual

comparison, we saw that H3K

9Ac appears more enriched downstream of the TSS and into the

gene body than other histone acetylation marks.

Our 19 state ChromHMM genome segmentation results corresponded well with the findings of

our weighted combinatoric NMF model (

Figure

). In our NMF model, we found that TSS

associated combinations (C1 and C2) are defined by H3K9Ac; similarly, in our ChromHMM

model, we found that

H3K9ac

preferred the promoter

associated set of states (set 1). In our NMF

model, we predicted a unique role for

H2AZac

in defining multiple promoter

associated

combinations (C2 and C3); in our ChromHMM model, we found

H2AZac

was se

lectively enriched

in State 17, a promoter

associated state. In our NMF model, we found that promoter distal

combinations are defined by

H2BK20ac

and

H3K27ac

(C4 and C5); in our ChromHMM model,

we found that these two marks prefer the promoter distal set of states (set 2).

SUPPLEMENTARY

METHODS

Cells

, Cell

Culture and Crosslinking

Cell lines.

We used the following cell lines in this study: (i) Female mouse ES cells (pSM44 mES

cell line) derived from a 129 × castaneous F1 mouse cross and (ii) K562, a female human

lymphoblastic cell line (ATCC, Cat # CCL

243).

Primary cells.

Mouse dendritic cells

(mDC)

were derived from bone marrow harvested from 6

week old female C57BL6 mice as previously described.

Briefly, b

one marrow was

harvested

then dissociated into single cells and filtered through 70um cell

strainer. The cells were then

incubated with red blood

cell lysis

buffer for 5

minutes. To differentiate bone marrow to dendritic

cells, bone marrow cells were plated at 200,000 cells/mL in non

tissue culture treated plates. These

cells were supplemented with media on day 2 and day 7. On day 5

cells were harvested and

resuspended in fresh media. On day 8

all the floating cells were collected as mouse bone marrow

derived dendritic cells.

Cell Culture Conditions.

(i) pSM44 mES cells were grown at 37C under 7% CO

on plates coated with 0.2% gelatin (Sigma,

G1393

100ML) and 1.75 mg/mL laminin (Life Technologies Corporation, #23017015) in serum

free 2i/LIF media composed as follows: 1:1 mix of DMEM/F

12 (GIBCO) and Neurobasal

(GIBCO) supplemented with 1x N2 (GIBCO),

0.5x B

27 (GIBCO 17504

044), 2 mg/mL bovine

insulin (Sigma), 1.37 mg/mL progesterone (Sigma), 5 mg/mL BSA Fraction V (GIBCO), 0.1 mM

mercaptoethanol (Sigma), 5 ng/mL murine LIF (GlobalStem), 0.125 mM PD03

25901

(SelleckChem) and 0.375 mM CHIR99021 (SelleckChem). 2i inhibitors were added fresh with

each medium change. Fresh medium was replaced every 24

48 hours depending on culture density,

and cells were passaged every 72 hours using 0.025% Trypsin (Life Te

chnologies) supplemented

with 1mM EDTA and chicken serum (1/100 diluted; Sigma), rinsing dissociated cells from the

plates with DMEM/F12 containing 0.038% BSA Fraction V.

(ii) K562 cells were purchased from ATCC and cultured in

DMEM (Life Technologies, #

11965118), 10% FBS (VWR, #97068

091), 100U/mL Penicillin/Streptomycin (Life

Technologies, # 15140122

), 1mM Sodium Pyruvate (Thermofisher, #11360070), 2mM L

Glutamine (Life Technologies # 25030081) at 37C and 5% CO in 15cm plates (USA Scientific #

5663

9160Q).

(iii) mDC

were cultured as previously described

ells were maintained at 37° C in 5% CO2

humidified incubators.

The media used for culturing and differentiating contains RPMI (Gibco)

supplemented with 10% heat inactivated FBS (Gibco), β

mercaptoethanol (50uM, Gibco), MEM

non

essential amino acids (1X, Gibco), sodium pyruvate (1mM, Gibco), and GM

CSF (20

ng/ml;

Milte

nyi).

Cell Harvest

(i) For harvesting pSM44 mESCs, cells were trypsinized by adding 5 mL of TVP (1 mM EDTA,

0.025% Trypsin, 1% Sigma Chicken Serum; pre

warmed at 37C) to each 15 cm plate and rocking

gently for 3

4 min until cells start to detach. 25 mL of wash solution (DMEM

12 supplemented

with 0.03% GIBCO BSA Fraction V, pre

warmed at 37C) was added to each plate to inactivate

the trypsin. Detached cells were transferred into 50 mL conical tubes, pelleted at 330 g for 3 min,

washed in 4 mL of 1X PBS per 10 million cells a

nd then pelleted in 1X PBS in preparation for

crosslinking. (ii) For harvesting K562s, the cell suspension was transferred to 50mL conical tubes,

pelleted at 330 g for 3 min, washed with 4 mL of 1X PBS per 10 million cells and then pelleted in

1X PBS in pr

eparation for crosslinking.

Cell Treatment.

mDCs were stimulated with 100 ng/mL lipopolysaccharide (LPS, rough, ultra

pure E. coli K12

strain, Invitrogen) and collected at 0 hours, 6 hours and 24 hours post

treatment

Cell Crosslinking

Cells were crosslinked in suspension with 1%

formaldehyde (FA)

for 10 min at room temperature.

For both cell lines, during crosslinking steps and subsequent washes, volumes were maintained at

4 mL of buffer or crosslinking solution per 10 million cells. Pelleted cells were resuspended in

1mL

of 1X PBS per 10 million cells and pipetted to disrupt clumps of cells. Next, cells were

crosslinked in suspension in a final volume of 4 mL of 1% formaldehyde (Pierce 28906) diluted

in 1X PBS per 10 million cells and rocked gently for 10 min at room temperature. Formaldehyde

was immediately quenched with addition of 200

of 2.5 M glycine (Sigma G7403

250G) per 1

mL of 1% FA solution and incubated with gentle rocking for 5 min at room temperature. Cells

were then washed three times with 0.5% BSA in 1X PBS that was kept at 4C. Finally, aliquots of

10 million cells were prep

ared in 1.7 mL tubes; these cell aliquots were pelleted, flash frozen in

liquid nitrogen and stored in

80C until

lysis.

Nuclear Isolation and Chromatin Preparation

Nuclear Isolation.

Crosslinked cell pellets (10 million cells) were lysed using the following nuclear isolation

procedure: cells were incubated in 0.7 mL of Nuclear Isolation Buffer 1 (50 mM HEPES pH 7.4,

1 mM EDTA pH 8.0, 1 mM EGTA pH 8.0, 140 mM NaCl, 0.25% Triton

100

, 0.5% NP

40,

10%

glycerol

, 1X

Protease Inhibitor Cocktail (

PIC

)

for 10 min on ice. Cells were pelleted at 850

g for 10 min at 4C. Supernatant was removed, 0.7 mL of Lysis Buffer 2 (50 mM HEPES pH 7.4,

1.5 mM EDTA, 1.5 mM EGTA, 200 mM NaCl, 1X PIC) was adde

d and the sample was incubated

for 10 min on ice. Nuclei were obtained after pelleting and supernatant was removed (as above).

Then, 550 uL of Lysis Buffer 3 (50 mM HEPES pH 7.4, 1.5 mM EDTA, 1.5 mM EGTA, 100 mM

NaCl, 0.1% sodium deoxycholate, 0.5% NLS, 1X

PIC) was added and the sample was incubated

for 10 min on ice prior to sonication.

Chromatin fragmentation and size analysis.

Chromatin was fragmented via sonication of the nuclear pellet using a Branson needle

tip sonicator

(3 mm diameter (1/8’’ Doublestep), Branson Ultrasonics 101

148

063) at 4C for a total of 2.5 min

at 4

5 W (pulses of 0.7 s on, followed by 3.3 s off). To che

ck the resulting DNA size distribution,

a small aliquot of 20uL of sonicated lysate was then added to 80uL of Proteinase K buffer ((20

mM Tris pH 7.5, 100 mM NaCl, 10 mM EDTA, 10 mM EGTA, 0.5% Triton

100

, 0.2% SDS)

and reverse crosslinked at 80C for 30 m

inutes. DNA was isolated using Zymo IC DNA Clean and

Concentrator columns and eluted in water. 10uL of purified DNA was then run for 10 minutes on

a 1%

agarose

gel (Invitrogen

™

Gel

™

EX Agarose Gels, 1%, Cat.No. G402021). Fragments