Supporting Information Cho et al. 10.1073/pnas.0807227105 SI Materials and Methods automatically by the iCycler iQ optical system software (Bio- Chromatin Immunoprecipitation Coupled with Microarrays (ChIP- Rad). Normalized Ct (⌬Ct) values for each sample were then chip). To identify Lrp- and RNAP-binding regions in vivo, we calculated by subtracting the Ct value obtained for the mock-IP ⌬ ϭ isolated the DNA bound to Lrp protein and RNAP from DNA from the Ct value for the IP-DNA ( Ct CtIP–Ctmock). formaldehyde cross-linked E. coli cells (BOP508 and MG1655, Primer sequences used in this study are available on request. respectively) by chromatin immunoprecipitation with the spe- cific antibodies that specifically recognizes myc tag (9E10; Santa Transcriptional Analysis. Samples for RNA transcript analyses Cruz Biotechnology) and RpoB subunit of RNAP (NT63; were taken from exponentially growing cells. Total RNA and Neoclone), respectively (1). Cells were grown in 2 g/L glucose cDNA preparation was performed as described previously (3). M9 minimal media supplemented with or without 10 mM leucine Each qPCR contained 0.5 ␮M of each forward and reverse at 37°C (2, 3). When the cultures reached an absorbance of Ϸ0.6, primer (qPCR primer set available on request), 150 ng cDNA, the cells were cross-linked by 1% formaldehyde at room tem- and 25 ␮L2ϫ SYBR Master Mix (Qiagen). All qPCR were done perature for 30 min. Following quenching the unused formal- in triplicate on a Bio-Rad Lightcycler with the following profile: dehyde with 125 mM glycine at room temperature for 5 min, the 1 cycle at 95°C for 15 min, 45 cycles at 94°C for 20 sec, 54°C for cross-linked cells were harvested by centrifugation and washed 20 sec, and 72°C for 30 sec. A standard curve was calculated for 3 times with 50 mL of ice-cold TBS. The washed cells were each primer and a PCR efficiency obtained from it. Using the resuspended in 0.5 ml of lysis buffer composed of 50 mM standard curve, the relative cDNA quantity was obtained for Tris-HCl (pH 7.5), 100 mM NaCl, 1 mM EDTA, protease each gene by normalizing it to the quantity of acpP cDNA in the inhibitor mixture (Sigma) and 1 kU Ready-Lyse lysozyme (Epi- same sample. Affymetrix GeneChip E. coli Genome 2.0 arrays centre). The cells were incubated at room temperature for 30 were used for genome-scale transcriptional analyses. cDNA min and then treated with 0.5 ml of 2ϫIP buffer composed of 100 synthesis, fragmentation, end-terminus biotin labeling, and array mM Tris-HCl (pH 7.5), 200 mM NaCl, 1 mM EDTA, 2% hybridization were performed as recommended by Affymetrix (vol/vol) Triton X-100, and protease inhibitor mixture (Sigma). standard protocol (www.affymetrix.com). The lysate was then sonicated 4 times for 20 sec each in an ice bath to fragment the chromatin complexes. Cell debris was Data Analysis. To identify enriched regions in the ChIP-chip data, removed by centrifugation at 13,000 rpm at 4°C for 10 min, and we used the previously developed peak finding algorithm, mpeak the resulting supernatant was used as cell extract for the immu- (4). The log2 ratios of each spot in the microarray were calculated noprecipitation. To immunoprecipitate the protein–DNA com- from the raw signals obtained from both Cy5 and Cy3 channels, plexes, the antibody was then added into the cell extract. For the and then the values were scaled by Tukey bi-weight mean (5). control (mock-IP), 2 ␮g of normal mouse IgG (Upstate) was ChIP-chip analysis is a 2-color array with the reference on the added into the cell extract. They were then incubated overnight array, the data are therefore thought of as a stand-alone at 4°C, and 50 ␮l of the Dynabeads Pan mouse IgG (for c-myc) experiment. The Tukey bi-weight function is used to account for or proteinA (for RNAP ␤ subunit) magnetic beads (Invitrogen) differences in the dyes on the array, whereas RMA is used to was added into the mixture. After 5-h incubation at 4°C, the account for differences between arrays so that the chips can be beads were washed twice with the IP buffer [50 mM Tris-HCl compared. The log2 ratio of Cy5 (IP DNA) to Cy3 (mock-IP (pH 7.5), 140 mM NaCl, 1 mM EDTA, 1% (vol/vol) Triton DNA) for each point was calculated from the scanned signals. X-100, and protease inhibitor mixture], once with the wash Then, the bi-weight mean of this log2 ratio was subtracted from buffer I [50 mM Tris-HCl (pH 7.5), 500 mM NaCl, 1% (vol/vol) each point. Each log2 ratio from triplicate samples was then Triton X-100, and 1 mM EDTA], once with wash buffer II [10 averaged and the Lrp-binding sites were identified using mpeak mM Tris-HCl buffer (pH 8.0), 250 mM LiCl, 1% (vol/vol) Triton program (www.stat.ucla.edu/ϳzmdl/mpeak). All signals can be X-100, and 1 mM EDTA], and once with TE buffer [10 mM downloaded from our website (http://systemsbiology.ucsd.edu/ Tris-HCl (pH 8.0), 1 mM EDTA] in order. After removing the publications). Genome-scale data were visualized using Nimble- TE buffer, the beads were resuspended in 200 ␮l of elution Gen’s SignalMap software, and most other analyses were per- buffer [50 mM Tris-HCl (pH 8.0), 10 mM EDTA and 1% SDS] formed with Matlab version 6.5 and Microsoft Excel software. and incubated overnight at 65°C for reverse cross-linking. After For indentifying transcription units in E. coli, we used EcoCyc reversal of the cross-links, RNAs were removed by incubation database (6). with 200 ␮L of TE buffer with 1 ␮L of RNaseA (Qiagen) for 2 h at 37°C. Proteins in the DNA sample were then removed by Lrp DNA-Binding Site Profile. Before attempting to discover the incubation with 4 ␮L of proteinase K solution (Invitrogen) for DNA-binding site signal, we sought to identify as precisely and 2 h at 55°C. The sample was then purified with a PCR purifi- narrowly as possible the chromosomal regions under each Lrp cation kit (Qiagen). ChIP-chip peak where the actual Lrp binding was occurring. We did this in the following manner. For the gene promoter asso- qPCR. To monitor the enrichment of promoter regions, 1 ␮Lof ciated with each peak, we identified the CDS start position and immunoprecipitated DNA was used to carry out gene-specific the positions of all known transcription start sites (TSSs). We qPCR. The quantitative real-time PCR of each sample was then identified the likely Lrp-binding region to be from 300 bp performed in triplicate using iCycler (Bio-Rad) and SYBR green upstream of the most 5Ј TSS (or CDS start in the case of no TSSs) mix (Qiagen). The real-time qPCR conditions were as follows: 25 to 100 bp downstream of the CDS start position. For intragenic ␮L SYBR mix (Qiagen), 1 ␮L of each primer (10 pM), 1 ␮Lof ChIP-chip peaks and those within convergent regions, we used IP or mock-IP DNA, and 22 ␮LofddH2O. All real-time qPCR the chromosomal regions ϩ/Ϫ 200 bp of the peak maximum were done in triplicates. The samples were cycled to 94°C for 15 s, position. The so-extracted chromosomal regions we term ‘‘Lrp- 52°C for 30 s, and 72°C for 30 s (total 40 cycles) on a LightCycler binding regions’’. The large number of discovered Lrp-binding (Bio-Rad). The threshold cycle (Ct) values were calculated regions allowed us to undertake a computational analysis to Cho et al. www.pnas.org/cgi/content/short/0807227105 1of17 understand the DNA signals that lead to Lrp binding. To do this, moter regions of ilvIH, gcvTHP, gltBDF, and dmsABC operons. we developed an algorithm that iteratively learned both the DNA First, the ilvIH operon encodes an acetohydroxy acid synthase motif and the motif separation distribution together and in a that is involved in the biosynthesis of branched-chain amino acids self-reinforcing manner. The discovered DNA motif, which and known to be directly stimulated by Lrp (11). As expected, bears similarity to previously estimated motifs from wild-type significant level (log2 ratio ϳ 3.2) of Lrp association was and SELEX-generated experiments, possesses nucleotide pref- measured at the promoter region of ilvIH operon, and the level erences deemed important for the positioning, stability and of Lrp association was significantly reduced following the addi- accommodation of nucleosomes. We found that available motif tion of 10 mM exogenous leucine (log2 ratio ϳ 1.7). The change finding algorithms were unable to discover a DNA-binding signal in Lrp association completely illustrates the fact that the addition within the Lrp-binding regions that was consistent with what is of exogenous leucine causes a 5- to 10-fold reduction in ilvIH known about the binding signal for Lrp and that was able to promoter activity (12). On the other hand, the levels of Lrp discriminate Lrp-binding regions from random, non-Lrp-peak association were almost identical in the presence and absence of associated chromosomal regions. The Lrp-binding signal is an exogenous leucine at the promoter regions of gcvTHP and especially difficult case because it has high degeneracy (i.e., low gltBDF operons, which catalyze the oxidative cleavage of glycine information content) and because the Lrp protein itself is into CO2,NH3 and 5,10-methylenetetrahydrofolate and 1 of 2 complicated due to its many oligomerization states.