1 Supplementary Figures

CRISPR

A B 3

Target F T A Scramble Jun JunD FosL1 Veh FosB KLA 1h IB: ATF3 BATF3 Jdp2 IB: Actin ATF2 BATF FosL2 IB: Jun JunD Jun ATF3 IB: Actin Fos JunB IB: JunD 0 2 4 6 8 10 12 Gene Expression (Log2 RPKM) IB: Actin

C ) ) ) M 14 Up 2-fold 191 14 Up 2-fold 11 M 14 Up 2-fold 72 M K K

Up 4 fold 943 K Up 4 fold 67 Up 4 fold 354 P P

12 P 12 12 R R R 10 10 10 Log2 8 8 Log2 8 Log2 A A A N N

6 N 6 6 R R R ( ( ( O

4 4 O 4 O K Down 2-fold 621 Down 2-fold 58 K Down 2-fold 297 K 3

2 2 D 2 F Down 4 fold 153 Down 4 fold 10 Down 4 fold 102 T Jun Jun A 0 0 0 0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14 Scamble Control (RNA Log2 RPKM) Scamble Control (RNA Log2 RPKM) Scamble Control (RNA Log2 RPKM) Supplementary Figure 1. A genome-wide map of AP-1 activity in macrophages. A. mRNA expression of all AP-1 monomers in Vehicle and 1-hour KLA treated TGEMs by RNA-seq. B. Western blot analysis of ATF3, Jun and JunD expression in CRISPR mediated knockout of ATF3, Jun and JunD in iBMDM cells. C. Scatterplots showing RNA-seq expression between scramble control and CRISPR mediated knockout of ATF3, Jun and JunD.

1 A B C chr19 46,310,000 2.5 Veh 125 _ Veh 40 Shared Nfkb2 KLA 1h ATF3 Fos KLA 1h 35 Jun 2.0 0 _ 30 JunD -seq (x10E4) P I chr19 88,530,000 1.5 25 40 _ Ifnb1

1.0 FosL2 -seq Peaks (Percent) 20 P I 0 _ 15 0.5 chr17: 5,670,000 10 0.0 50 _ Rela JunB Unique Ch 5 Number of Peaks Ch 3

F 100 80 60 40 20 Jun Fos T 0 _ Percentile Threshold JunB JunD A FosL2 5 kb

D E Jun 262-KRMRNRIAASKCRKRKL-276 92,350,000 24,890,000 JunB 272-KRLRNRLAATKCRKRKL-286 chr19: chr17: JunD 237-KRLRNRIAASKCRKRKL-251 27 _ Cxcl10 50 _ Spsb3 Fos 143-RRERNKMAAAKCRNRRR-159 FosL2 130-RRERNKLAAAKCRNRRR-146 ATF3 100-RRERNKIAAAKCRNKKK-107 0 _ 0 _ DNA Contact * *** ** *** ** 27 _ 50 _ ag Count ag Count 0 _ 0 _ 27 _ 50 _ NormT NormT

0 _ 0 _ 5 kb 5 kb

ATF3-ATF3 DBD ATF3-Jun DBD ATF3-Fosf DBD Supplementary Figure 2. Extended characterization of AP-1 binding cistrome. A. Quantification of the number of binding sites for each monomer that are present in vehicle, 1-hour KLA, or shared in both treatment conditions. B. Representative browser shots of ChIP-seq peaks for KLA specific monomers Fos (top), FosL2 (middle), and JunB (bottom). C. Percent of peaks that are unique to each monomer at different thresholds for the number of reads at each peak. 100 indicates 100% of the peaks and 10 indicates the top 10% of peaks when sorting by the number of reads. D. Protein alignment of AP-1 monomer DNA binding domains by Clustal Omega. Black background denotes identity while grey background denotes similar charge. Amino acids involved in DNA binding are indicated by stars. E. Representative browser shots of ChIP-seq peaks from ATF3 DNA binding domain chimeras at an ATF3 specific site (left) and Jun specific site (right).

2 Merge Cutoff Pearson Correlation >= 0.9 1 2 3 22 2 23 7 3 G G G .2 .2 .2 I I I L F F E E L Jun Fos L L wist2 S T T Atoh1 H H Nfe2 ::Jun Jdp2 T JunB JunD O A A O O L L O FosL1 Creb5 F Bhlha15 Neurog1 B Neurog2 Neurog2 F H H T Fos::Jun B A B Jun var Jdp2 var B JunD var

Merged Motif Merged Motif

0.40 0.30 0.20 0.10 0.00 Correlation Difference Supplementary Figure 3. Curation of the JASPAR motif library to eliminate highly similar motifs. Hierarchical clustering of all motifs in the JASPAR non-redundant library. Colored clades give highly similar motifs (Pearson Correlation ≥ 0.9, indicated by the dotted line) and blue leaf nodes give motifs that are distinct from all other motifs (Pearson Correlation < 0.9). Representative clades are indicated on the left and right (with the motif logos indicated).

3 A C D 3.0 Six3 H1-hESC 2.5 ATF3 Hic Threshold Jun Lhx3 Rora-var JunD 4462 2.0 Nfil3 HepG2 Znf423 GM12878 1864 Zeb1 1.5 149 749 691 Dux4 77 532 Gfi1b 121 1482 1.0 Zfx 3630 4980 Absolute Likelihood Ratio 24174

of Motifs Passing Pax5 15 3202 1.0 1.5 2.0 2.5 3.0 3.5 3.9 Znf410 181 Threshold (-Log10 p-val) NfkB 701 Nkx2-5var2 365 3629 Znf263 29 163 Znf354c Myc 259 0.2 768 Mtf1 40 369 284 B Tead 650 3 rep1 0.0 Hnf1 5953298 F Fox 14226 28921 T

A -0.2 PCC=0.98 Hnf4g Hnf4a SK-N-SH K562 Cebp

0.0 Nfi 0.25 ATF3 rep1 -0.25 Zic Rel TBA model Weight Zbtb33 Pou 0.2 Dbp-Hlf-Tef Nrf1 Elk-Etv

eight 0.0 AP-1 CRE

JunD rep2 PCC=0.97 AP-1 TRE -0.2 Ctcf Irf3/7/8/9

model W PU.1 0.0 A 0.25

-0.25 Pax2 B Jun rep1

T Duxa

TBA model Weight Maf-Nrl eight >20 (positive) Runx MybL2 0.0 0.2 Gata MybL1 >20 (negative) model W

0.0 Egr A

Sp4 B Jun rep2 -0.2 PCC=0.98 Tcfl5 T H C S S 0.0 0.25 E K562 -0.25 12878 JunD rep1 -N- HepG2 K M S H1-h

TBA model Weight G Supplementary Figure 4. Characterization of TBA on individual replicate experiments and JunD ChIP- data from different cell lines. A. Distribution of the absolute ratio of the likelihood value for each motif that has mean likelihood below the threshold indicated on the horizontal axis, when comparing models trained on individual replicate experiments. Likelihood values for each individual replicate are calculated using the likelihood ratio test, and then averaged across 5 cross validation sets. Error bars indicate the standard deviation. B. Comparison of the weights of significant motifs (p<10e-2.5) from TBA models calculated from individual experiments. The similarity of each pair of models, measured by the Pearson correlation coefficient, is annotated in each panel. C. Significance values for the 20 most significant motifs for TBA models trained for JunD binding in a several of cell lines. Red hues indicate motifs positively correlated with binding and blue hues indicate motifs negatively correlated with binding. D. Overlap of JunD binding sites from various cell lines. Labels indicate the number of binding sites that overlap between a combination of the cell lines.

4 A C >4 1 Cebpα 16 MafB 3.0 ) 14

M 2 Cebpβ 17 MafY K 18 Myc

P 3 Ctcf 12 R 4 E2f8 19 Nfe2L2 1.5 14 5 Egr1 20 NfκB2 Positivee 10 6 Egr2 21 Prdm1 5 11 0.0 12 6 2 7 Egr3 22 Rel 8 9 13 8 Etv3 23 Stat2 26 19 9 Fos 24 Tef 18 6 16 1 10 FosL2 25 Tfec 20 17 21 8 26 TgiF1 22 10 11 ID2 4 25 28 12 Irf1 27 Thap1 4 3 29 13 Jun 28 Usf1 Gene Expression (Log2 15 24 A 2 7 14 JunB 29 Zbtb7A L 23 K 27 15 Klf4 0.0 0

0 2 4 6 8 10 12 14 Motif Significance (-log10 pval) 1.5 Veh Gene Expression (Log2 RPKM)

Negativee 3.0 >4

B Higher Positively Ranked Motifs

Data in Table 1 ATF3 Jun Fos KLA FosL2 JunB JunD

Data in Table 2 3 F Jun Fos Higher Negatively Ranked Motifs T JunB JunD A FosL2

KLA Supplementary Figure 5. TBA identifies motifs that coordinate the binding of each AP-1 monomer in KLA-1h treatment. A. mRNA expression of transcripts before and after KLA-1h treatment. Differentially expressed (FDR<0.05) factors with known DNA motifs are highlighted in red (up-regulated) and blue (down-regulated) , labeled and listed to the right. B. DNA motifs rank order based on the significance of the motif according to the likelihood ratio test. C. Heatmap representing the negative log10 p-value of each motif that shows a 100 fold likelihood ratio between two monomers when using the likelihood ratio test.

5 )

A M E K

P Irf3/7/8/9 R

PU.1

>5 Jun Fos Atf3 JunB JunD Gene Expression (Log2 FosL2 Balb/c Veh C57Bl6 Veh Balb/c KLA 1h B C57Bl6 KLA 1h

reatment Train

T Monomer Test 0.94 0.94 0.94 0.94 Positive Motifs (Log10 p-val) 0.95 0.94 0.95 0.94 Strain C57Bl/6 0.92 0.94 0.92 0.94 Balb/c 0.94 0.93 0.93 0.94 Treatment Veh 0.95 0.95 0.95 0.95 KLA-1h Monomer 0.94 0.94 0.94 0.94 ATF3 Jun 0.95 0.95 0.94 0.95 Fos JunB 0.93 0.93 0.92 0.94 JunD

deltaSVM AP-1 TRE TBA-2Strain C AP-1 CRE TBA 0.6 Model Predictions

0.4

0.2 speciific binding

A 0.0 L Correlation to Strain K

-0.2 Individual Motifs 3 F Jun Fos T JunB JunD D A

Positive Significant Motifs Negative Motifs (Log10 p-val) 0.14 ATF3 >5 Jun 0.12 Fos JunB 0.10 JunD

0.08

0.06

peaks with mutation in motif 0.04 AP-1 S Spi1-c S 0.02 Cebp 3 F Jun Fos 0.00 T JunB JunD 0.00 0.04 0.08 0.12 A Fraction SS peaks with mutation in motif 6 Fraction Non- Supplementary Figure 6. Leveraging the effects of genetic variation to validate TBA predictions in activated macrophages. A. mRNA expression of monomers before and after 1-hour KLA treatment in C57Bl/6J and BALB/cJ. B. Model performance when varying the strain of the data used for training and testing the TBA model. TBA was trained on AP-1 monomers in either Veh or KLA and on one strain and tested for predictive ability on either strain. C. Predictive performance of various models for predicting strain specific binding of each monomer as measured by the Pearson correlation of a model’s predictions versus the log2 fold change in binding between the Balb/cJ and C57Bl/6J at all AP-1 binding in activated, one hour KLA treated, TGEMs. The performance of models that integrate multiple motifs – deltaSVM, TBA, TBA-2Strain, are represented as diamonds. The correlation of the change in an individual motif’s score (due to a mutation) to strain specific binding is indicated using round points. D. Frequency of mutations in significant motifs (from TBA model, p<10e-2.5) at strain specific versus non-strain specific peaks for each monomer in KLA treated macrophages. E. Heatmap of significance values for motifs that intersected between the One Strain and Two Strain model.

7 2 Supplementary Tables

Motif Mean p-val ap-1 1.00E-50 cebp 1.00E-50 spi1-c 1.00E-50 rel 1.00E-50 egr 1.15E-50 atf7 batf3 creb5 1.64E-33 pax2 2.76E-28 runx 4.26E-25 elk etv 4.48E-23 maf nrl 1.34E-15 irf1 2.70E-13 mzf1 1.14E-07 tcfl5 1.82E-07 spib 2.32E-07 xbp1 1.39E-06 elf 2.77E-06

Supplementary Table 1. Table of highly significant motifs, positively correlated with binding for all AP-1 monomers in KLA treated TGEMs.

Motif Mean p-val zeb1 2.20E-46 yy1 1.53E-20 figla id4 snai2 tcf3 tcf4 4.78E-11 onecut 1.04E-07 tbp 3.04E-07 msc myf6 tfap4 6.38E-07

Supplementary Table 2. Table of highly significant motifs, negatively correlated with binding for all AP-1 monomers in KLA treated TGEMs.

8