DG-based prediction and experimental confirmation of SYCRP1-binding sites on the Synechocystis genome Katsumi Omagari 1 , Hidehisa Yoshimura 2 , Takayuki Suzuki 2 , Mitunori Takano 3 , Masayuki Ohmori 2,4 and Akinori Sarai 5 1 Department of Virology, Medical School, Nagoya City University, Japan 2 Department of Life Sciences, Graduate School of Arts and Sciences, The University of Tokyo, Japan 3 Department of Physics, School of Science and Engineering, Waseda University, Tokyo, Japan 4 Department of Biological Sciences, Faculty of Science and Engineering, Chuo University, Tokyo, Japan 5 Department of Biochemical Engineering and Science, Kyushu Institute of Technology (KIT), Fukuoka, Japan The cAMP receptor protein (CRP) that was first iden- tified in Esherichia coli exists in many other organisms. SYCRP1 is a cAMP receptor protein found in the cya- nobacterium Synechocystis sp. PCC 6803 [1]. Although E. coli CRP is a global transcription factor controlling 20–100 genes, SYCRP1 has been reported to control only the slr1667–slr1668 operon [2,3]. However, many other genes are expected to be regulated by SYCRP1 because the concentration of cAMP in Synechocystis cell changes under blue-light irradiation [4,5]. A number of methods for predicting binding sites of transcription factors in the genome have been developed over the last three decades. The methods can be classified into three groups according to the type of information used in the prediction [6]: (a) the sequence-based method, (b) the structure-based method, and (c) the DG-based method. The sequence-based method uses the alignment of known binding sequences for screening the database for potential target binding sites [6,7], and relies on sequence information obtained from known binding sites of transcription factors [8]. The structure-based method aligns different DNA sequences on the protein–DNA framework and quantitatively estimates the fitness of the complex structures with those sequences [9]. The DG-based method utilizes the change in the binding free energy, DDG, which is defined as the difference between the binding free energy of a protein to a mutant DNA sequence and that to the consensus DNA sequence, to predict potential target binding sites of a transcription factor [6,10]. The set of DDG values is determined by Keywords additivity; binding free energy change; DNA- binding sites; prediction; regulatory protein Correspondence K. Omagari, Department of Virology, Medical School, Nagoya City University, 1 Kawasumi, Mizuho, Nagoya, 467-8601, Japan Tel ⁄ Fax: +81 52 853 8191 ⁄ 3638 E-mail: usagi525@med.nagoya-cu.ac.jp (Received 13 April 2008, revised 21 June 2008, accepted 30 July 2008) doi:10.1111/j.1742-4658.2008.06618.x DNA-binding sites for SYCRP1, which is a regulatory protein of the cyanobacterium Synechocystis sp. PCC6803, were predicted for the whole genome sequence by estimating changes in the binding free energy (DDG A total ) for SYCRP1 for those sites. The DDG A total values were calculated by summing DDG values derived from systematic single base-pair substitu- tion experiments (symmetrical and cooperative binding model). Of the cal- culated binding sites, 23 sites with a DDG A total value < 3.9 kcalÆmol )1 located upstream or between the ORFs were selected as putative binding sites for SYCRP1. In order to confirm whether SYCRP1 actually binds to these binding sites or not, 11 sites with the lowest DDG A total values were tested experimentally, and we confirmed that SYCRP1 binds to ten of the 11 sites with a DDG total value < 3.9 kcalÆmol )1 . The best correlation coefficient between DDG A total and the observed DDG total for binding of SYCRP1 to those sites was 0.78. These results suggest that the DDG values derived from systematic single base-pair experiments may be used to screen for potential binding sites of a regulatory protein in the genome sequence. Abbreviations CRP, cAMP receptor protein; EMSA, electrophoresis mobility shift assay; ICAP, the consensus DNA sequence for E. coli CRP. Positions within the DNA site are the same as the numbering in [15]. 4786 FEBS Journal 275 (2008) 4786–4795 ª 2008 The Authors Journal compilation ª 2008 FEBS conducting systematic single base-pair substitution experiments, in which each single base pair of the consensus DNA sequence of binding sites is substituted by all other possible base pairs to determine the respec- tive base-pair contributions to the binding free energy of a transcription factor to DNA. To date, DDG values have been measured for six transcription factors, Cro and the k repressor, c-Myb, the ERF domain, E. coli CRP and SYCRP1 [11–16]. The DDG values for c-Myb, for example, have been successfully used not only to pre- dict binding sites that are in agreement with many puta- tive binding sites but also to locate sequences of several new promoters that could be targets for c-Myb [6,10]. In this study, we searched the whole genome sequence for potential binding sites of SYCRP1 that are upstream of ORFs and tightly bound in vitro, using the DG-based method. The potential binding sites were assumed to bind to SYCRP1 only, although other co-factors related to gene regulation might change the sequence pattern of DNA binding sites [17]. SYCRP1 binds tightly to the consensus palindromic DNA sequence of E. coli CRP, T 4 G 5 T 6 G 7 A 8 T 9 C 10 T 11 |A 12 G 13 A 14 T 15 C 16 A 17 C 18 A 19 . Three amino acids (Arg180, Glu181 and Arg185) in E. coli CRP that interact with GC base pairs at posi- tions 5 and 7 through hydrogen bonding are completely conserved [2]. The DDG values for SYCRP1 for the respective base-pair substitutions at positions 4–8 in the consensus sequence have been derived from systematic single base-pair substitution experiments [16]. To increase the accuracy of the prediction, additional DDG values for positions 9–11 in the consensus sequence were measured using an electrophoresis mobility shift assay (EMSA). The measurement enabled us to identify another important base pair involved in specific binding of SYCRP1 that had little effect on the binding of E. coli CRP. For prediction of binding sites of SYCRP1 in the genome sequence, the total changes in binding free energy (DDG A total ) for every 16 bp DNA segment were calculated by summing DDG values for the respec- tive base pairs within the segment. Binding of SYCRP1 to the sites with the lowest DDG A total values was con- firmed by EMSA. It was found that SYCRP1 binds to hitherto unknown sites, and it is suggested that SYC- RP1 regulates genes downstream of the sites. Results Systematic single base-pair substitution experiments for the spacer region in the consensus sequence In order to include the effects of a spacer region for pre- diction of SYCRP1 binding sites, we measured the DDG values of SYCRP1 at positions 9–11 (Fig. 1A) using systematic single base-pair substitution experiments. The spacer region is a segment of DNA flanked by the positions 4–8 that strongly interact with amino acids of each monomer of a SYCRP1 dimer. Figure 2 shows the DDG values for the respective base-pair substitutions at positions 9–11 (this study) and positions 4–8 [16]. A positive DDG value means that the binding affinity is reduced by the base-pair substitution. Substitution of T by A at position 9 caused a 2.0 kcalÆmol )1 increase in the DDG value, which is the largest among the substitu- tions at positions 9–11. This increase is of the same magnitude as those for substitutions at positions 6 and 8. Substitution of T by G at position 9 also showed a non-negligible change in DDG. Substitution of T by C at position 9 and all substitutions at positions 10 and 11 changed DDG values slightly by < 0.5 kcalÆmol )1 , which is smaller than the changes for substitutions at position 4, at which there is no interaction between the base pair and any amino acids of SYCRP1 [16]. Estimation of DDG A total for the whole genome sequence using DDG values derived from systematic single base-pair substitution experiments Using the DDG values for positions 4–8 obtained previ- ously [16] and those for positions 9–11 obtained in this study, we searched the Synechocystis genome for SYC- RP1 binding sites. Figure 3 shows the procedure for the DDG-based prediction. The binding affinity of SYCRP1 to a fragment of 16 bp is estimated as the sum of the DDG values at each position. The window of 16 bp was moved 1 bp at a time along the genome sequence, and the binding affinity of SYCRP1 to each segment was evaluated in terms of the change in bind- ing free energy (DDG A total ). The calculation was based on the assumption of cooperative binding, whereby a symmetrical dimer of SYCRP1 binds to the two half sites in a twofold-symmetrical manner. Figure 4 shows a typical example of the distribution of DDG A total values around genes regulated by SYCRP1 (slr1667–slr1668 operon). The position with the lowest DDG A total value corresponds to the known binding site for SYCRP1. The histogram of DDG A total values for the whole genome (Fig. 5) shows that the DDG A total values ranged from - 0.02 to 33.8 kcalÆmol )1 . The number of sites with low DDG A total values was very small. Sites with DDG A total < 3.9 kcalÆmol )1 were selected as potential binding sites in this study because those sites could be con- firmed to bind to SYCRP1 experimentally. There were seven sites for which DDG A total was < 1.3 kcalÆmol )1 , 17 for which 1.3 £ DDG A total < 2.6 kcalÆmol )1 , and 114 K. Omagari et al. DG-based prediction of DNA binding sites by SYCRP1 FEBS Journal 275 (2008) 4786–4795 ª 2008 The Authors Journal compilation ª 2008 FEBS 4787 for which 2.6 £ DDG A total < 3.9 kcalÆmol )1 . Of them, we selected sites with a lowDDG A total value upstream or between ORFs as putative binding sites. Twenty-three putative binding sites were obtained (Table 1). The binding site for the slr1667–slr1668 operon, which is regulated by SYCRP1, is included among these sites. Confirmation of SYCRP1 binding to putative binding sites In order to confirm whether SYCRP1 actually binds to the putative binding sites, we performed an EMSA to measure changes in binding free energy (observed DDG total ) for the 11 binding sites with the lowest DDG A total values of the 23 putative binding sites. There were seven binding sites for which DDG A total < 2.6 kcalÆmol )1 and four for which 2.6 £ DDG A total < 3.9 kcalÆ- mol )1 . Figure 6 shows the result of the EMSA experi- ments. The experiments confirmed that SYCRP1 bound all the putative binding sites with DDG A total < 2.6 kcalÆmol )1 . The intensity of the complex band increased when the concentration of SYCRP1 was increased. The increment varied with the DNA sequence to which the SYCRP1 bound. The intensity of the complex band decreased with the increase in DDG A total value. In Fig. 7, we plotted DDG A total versus the observed DDG total and found a high correlation coefficient (0.78). For putative binding sites with DDG A total < 0.5 kcalÆmol )1 , the DDG A total values agreed well with the observed DDG total values. For those sites with 0.5 £ DDG A total < 2.6 kcalÆmol )1 , DDG A total values were twice as large as the observed DDG total values. Among those with 2.6 £ DDG A total < 3.9 kcalÆmol )1 , the DDG A total values of two putative binding sites, sll1874 A B Fig. 1. (A) Systematic single base-pair sub- stitutions of the DNA sequence. The substi- tuted DNA sequences were used to measure DDG values in binding experi- ments. ICAP represents a reference sequence for DG values in this study. Positions 9–11 in ICAP were subjected to systematic single base-pair substitutions. All possible DNA sequences with single base- pair substitutions are shown. (B) DNA sequences used for binding-confirmation experiments: DNA sequences used to con- firm whether SYCRP1 binds to putative binding sites or not are shown. Eleven puta- tive binding sites selected in ascending order of DDG A total are shown. 5′-TGTGATCT-3′ AC 4 4 3 2 1 ΔΔG (kcal·mol –1 ) 0 567891011 G ACT ACG ACT CGT ACG ACGAGT 3′-ACACTAGA-5′ Fig. 2. DDG values obtained in systematic single base-pair substitu- tion experiments. The changes in binding free energy were deter- mined from dissociation constant (K d ) values measured by using EMSA. The sequence shown at the bottom is that of ICAP. DDG values for positions 4–8 were measured by Omagari et al. [16]. Error bars are the standard errors calculated from three indepen- dent experiments. DG-based prediction of DNA binding sites by SYCRP1 K. Omagari et al. 4788 FEBS Journal 275 (2008) 4786–4795 ª 2008 The Authors Journal compilation ª 2008 FEBS and sll1708, agreed well with the observed DDG total values. However, the DDG A total value of the putative binding site slr1928 was three times larger than the observed DD G total value. For slr0733, the free DNA bands and the complex bands were not separated com- pletely because of the tailing from free DNA bands. One possible reason is that the binding of SYCRP1 to slr0733 was weaker than to sll1874 and sll1708, such that the SYCRP1 and DNA complex dissociated dur- ing electrophoresis. Thus, the observed DDG total value for slr0733 may be larger than the predicted DDG A total for that value. Discussion Systematic single base-pair substitution experiments Interactions of SYCRP1 with base pairs in the spacer region, which connects two half sites containing a consensus DNA sequence, were investigated using sys- tematic single base-pair experiments. Those experi- ments showed that the substitutions of T by A or G at position 9 caused the largest significant changes in DDG value in the spacer region. This spacer region is important for binding of SYCRP1 to DNA and pre- diction of potential binding sites. The predicted DDG A total values and observed DDG values exhibited good correlation (correlation coefficient of 0.78). The goodness of fit varied when the values for positions 4–8 were used in this search. These results showed rather weak correlation (correlation coefficient of 0.28). Inclusion of the DDG values for positions 9–11 enhanced the correlation between the predicted DDG A total values and the observed DDG total values. For E. coli CRP, the spacer region does not significantly affect binding [18]. In the E. coli CRP–DNA complex, there is no direct contact between bases and amino acids at these sites [19], and show interactions between amino acids and phosphates which are important for Fig. 3. Procedure for calculating DDG A total for the Synechocystis genome. The DD G values for each base position with respect to three substituted bases define the mutation matrix, as shown in the table. Sequences of length 16 bp were extracted from the genome, and DDG values corresponding to these base pairs were summed. As an example, the DDG values shown in italic in the mutation matrix are summed, giving a DDG A total value for the sample sequence of 1.64 kcalÆmol )1 . Similar calculations were repeated for the whole genome sequence. K. Omagari et al. DG-based prediction of DNA binding sites by SYCRP1 FEBS Journal 275 (2008) 4786–4795 ª 2008 The Authors Journal compilation ª 2008 FEBS 4789 binding [20]. According to the predicted structure of the SYCRP1–DNA complex [2], the base pairs at posi- tions 4–8 may form interactions with an a helix of SYCRP1 and base pairs at positions 9–11 may show no interaction with amino acids. We cannot determine whether interactions between bases and amino acids or other interactions are responsible for these changes from this study alone. Detailed structural information on both SYCRP1 and the SYCRP1–DNA complex would provide clues to clarify this issue. Examination of additivity Binding sites were predicted based on the assumption of additivity of changes in binding free energy in this study. The predicted DDG A total values and observed DDG values exhibited good correlation (correlation coefficient of 0.78). While the additivity assumption provided a certain degree of goodness-of-fit, the pre- dicted DDG A total values were not completely equal to observed DDG values. The predicted values were larger than observed ones. Although the sequence of the binding site (positions 4–19) upstream of sll1268 (No. 2 in Fig. 7) is identical with the consensus sequence, the observed DDG total value was not zero even considering the error bar. However, the observed DDG total value for slr1351 (number 4 in Fig. 7), whose sequence has only single mutation, was about the same as that of the consensus sequence. This indicates that sites outside the binding site have a non-negligible con- tribution to DDG value. In addition, the additivity model assumes that all base–amino acid interactions contribute independently. This assumption seems to hold well for Cro and the k repressor, which bind to DNA through two helix-turn-helix motifs in a homo- dimer. The predicted changes in binding free energy agree quite well with the observed changes for various multiple mutants and operator sequences [11,12]. In the case of Mnt, which is a member of the ribbon- helix-helix family and binds to DNA as a tetramer, and EGR1, a member of the Cys 2 His 2 zinc-finger fam- ily, this assumption does not seem to hold [21–24]. Some transcription factors form protein–protein con- tacts to stabilize DNA binding. Cooperative interac- tions mediated by these protein–protein contacts are required for high levels of binding affinity and specific- ity for many DNA-binding proteins [25]. For example, although MATa1 and MATa2, homeodomain proteins of Saccharomyces cerevisiae, bind to DNA with mod- est affinity and specificity for DNA, the a1⁄ a2 hetero- dimer binds DNA with higher affinity and specificity [26,27]. Such cooperative binding might explain the difference between the observed and predicted values. In the E. coli CRP–DNA complex structure, the CRP dimer binds to twofold-symmetrical DNA sequences symmetrically [19]. Although little is known about the cooperativity by which the SYCRP1 dimer binds to DNA, two models for DNA binding may be considered for binding of SYCRP1. The simplest model involves symmetrical and cooperative binding of SYCRP1 dimer to DNA. In this case, the total change in binding free energy (DDG A total ) is calculated by adding the change in binding free energy (DDG) for the two half sites. Predicted values are larger than observed ones. Fig. 5. Histogram of the DDG A total values for binding of SYCRP1 to the entire genome of Synechocystis based on the calculation of changes in binding free energy for SYCRP1 for every site in the entire Synechocystis genome. The binding is stronger when DDG A total values are lower. Fig. 4. Example of DDG A total calculation. DDG A total values around the slr1667–slr1668 operon regulated by SYCRP1 are shown. The posi- tions of slr1667 and slr1668 are shown at the top; the arrows rep- resent the actual binding site of the slr1667–slr1668 operon. The binding site upstream of the operon has the lowest DDG A total value of those calculated. DG-based prediction of DNA binding sites by SYCRP1 K. Omagari et al. 4790 FEBS Journal 275 (2008) 4786–4795 ª 2008 The Authors Journal compilation ª 2008 FEBS The other model, in contrast with the above sym- metrical and cooperative binding model, is the inde- pendent binding model, whereby either half site adopts a specific or non-specific binding mode independently while binding to DNA. In the non-specific binding mode, the protein binds to DNA but does not Table 1. Putative binding sites of SYCRP1 and the downstream genes. The standard errors were calculated from standard errors of DDG. No. a Locus b Product b Position c Sequence d DDG A total ±SE e 1 sll1520 DNA repair protein (RecN) )568.5 TGTGATCC|AGATCACA 0.0 ± 0.0 slr0442 Unknown protein )194.5 2 sll1268 Unknown protein )153.5 TGTGATCT|AGATCACA 0.0 ± 0.0 3 sll1543 Unknown protein )268.5 TGTGATCT|GGGTCACA 0.3 ± 0.1 slr1667 Unknown protein )251.5 4 sll1247 Unknown protein )158.5 GGTGATCT|AGATCACA 0.7 ± 0.2 slr1351 UDP-N-acetylmuramoylalanyl- D-glutamyl- 2,6-diamino-pimelate- D-alanyl-D-alanine ligase (murF) )92.5 5 sll1577 Phycocyanin b subunit (cpcB) )709.5 TGTGATCT|AAATCACC 1.1 ± 0.2 ssr2848 Unknown protein )93.5 6 slr0992 Hypothetical protein )75.5 TGTGATCT|CCGTCACC 1.6 ± 0.3 7 slr1732 Unknown protein )323.5 GGTGATTC|TAATCACA 2.0 ± 0.2 8 sll1874 Phytochrome-regulated gene (AT103) )394.5 TGTGATTA|TTCTCACA 2.6 ± 0.1 9 sll0702 Unknown protein )207.5 TGTGATGA|CCGTCATA 2.8 ± 0.4 slr0733 Integrase–recombinase protein (xerC) )548.5 10 slr1928 Unknown protein )47.5 TGTGTCCT|GGGTCACT 3.0 ± 0.3 11 sll1708 NarL subfamily )68.5 GGTGATTA|CTATCACG 3.1 ± 0.4 slr1805 Sensory transduction histidine kinase )100.5 12 ssl3335 Secretory protein (SecE) )282.5 GGTGTTGG|AGATCACA 3.1 ± 0.3 13 sll1608 Unknown protein )179.5 AGTGATGT|TTATCATT 3.1 ± 0.4 slr1717 Hypothetical protein )705.5 14 sll1268 Unknown protein )176.5 GGTGACCC|AGACCACT 3.2 ± 0.3 15 ssr2333 Unknown protein ) 91.5 AGTGATTA|TACTCACA 3.3 ± 0.4 16 slr1908 Hypothetical protein )322.5 AATGCTCC|GGGTCACT 3.6 ± 0.4 17 slr1334 Hypothetical protein )83.5 TGTAATTC|TGAGCACA 3.7 ± 0.4 18 slr0869 Unknown protein )201.5 TGTGACTA|CAACCACA 3.7 ± 0.3 19 sll1564 a -Isopropylmalate synthase (leuA) )215.5 TGTGATTG|AGACCATA 3.7 ± 0.3 ssr2802 ABC transporter )142.5 20 sll0533 Trigger factor (tig) )353.5 AATGCCCT|GCGTCACA 3.8 ± 0.4 slr0549 Aspartate b-semialdehyde dehydrogenese (asd) )312.5 21 slr0964 Hypothetical protein )95.5 AGTGCTCC|GGAACACT 3.8 ± 0.5 22 ssl0438 50S ribosomal protein L12 homologue )40.5 TGTGCTAT|TGCTCACG 3.8 ± 0.3 23 slr0054 Diacylglycerol kinase (dgkA) )51.5 TGTAATCC|AGGTTACA 3.8 ± 0.4 a The numbers of the putative binding sites correspond with the numbers shown in Fig. 7. b The genes downstream of putative binding sites. Protein-coding genes of the Entrenz genome database (ftp://ftp.ncbi.nih.gov/genomes/Bacteria/Synechocystis_PCC6803/ NC_000911.ptt) were used for the search. c Position of the center of the putative binding sites relative to the ORF start position. d Sequences of putative binding sites. e Changes in binding free energy and standard errors (DDG A total Æ SE). Fig. 6. Confirmation of SYCRP1 binding to predicted sites using EMSA. We confirmed whether SYCRP1 can bind to 11 putative binding sites selected from 23 sites in ascending order of DDG A total values. The gel images are typical examples. The DDG A total values for these exam- ples become larger from left to right. For lanes 1–4, the final SYCRP1 concentrations are 1, 10, 100 and 1000 n M, respectively. K. Omagari et al. DG-based prediction of DNA binding sites by SYCRP1 FEBS Journal 275 (2008) 4786–4795 ª 2008 The Authors Journal compilation ª 2008 FEBS 4791 recognize the sequence. In this case, simply adding the DDG values for the two half sites is not appropriate, and the following formula is used: DDG B total ¼ÀkT lnðexpðÀDDG l = kTÞþexpðÀDDG r = kTÞÞ ð1Þ where DDG l is calculated by summing the DDG values from the left half sites and spacer, and DDG r is calcu- lated by summing the DDG values from the right half sites and spacer. If the DDG sum for one site becomes too large, its contribution to DDG B total becomes less important. The correlation coefficient between the cal- culated DDG B total and observed DDG total values is 0.87 (Fig. 8). This value is better than that for the coopera- tive symmetrical binding. However, the predicted val- ues for three sites with high binding free energy did not agree with the observed DDG total values. In actual binding, the situation may be somewhere between these two extreme cases, i.e. the binding between SYCRP1 and DNA may take place with intermediate cooperativity between the monomers. The degree of cooperativity may also depend on the sequence of DNA [28] to which SYCRP1 binds. In addition, the validity of the assumption of additivity in calculating DDG (even in each half site) should also be examined in the case of SYCRP1, for example by conducting systematic double base-pair mutation analysis, to yield a higher level of prediction accuracy. Further investi- gations are necessary to disclose the mechanism of cooperativity in SYCRP1–DNA binding. Putative binding sites and target genes for SYCRP1 Using the DDG values derived from systematic single base-pair experiments, we predicted binding sites for SYCRP1 in the Synechocystis genome. Of the calcu- lated sites, those with DDG A total < 3.9 kcalÆmol )1 located upstream of ORFs were selected as putative binding sites. We obtained 23 putative binding sites, including the known slr1667–slr1668 operon binding site. We confirmed that SYCRP1 binds to ten of the 11 puta- tive binding sites. The upstream region of slr0442, whose expression level decreases in the sycrp1 disrup- tant [2], was found to have a binding site for SYCRP1. Fig. 7. Correlation between predicted and observed changes in binding free energy. DDG A total values were calculated based on the assumption of additivity and the cooperative binding model, whereby changes in the binding free energy due to single base-pair substitutions are summed assuming that a symmetrical dimer of SYCRP1 binds to two half sites in a twofold-symmetrical manner. The broken line is a 45° straight line. The numbers correspond to the sequences in Table 1. Values for number 9 (slr0733 and sll0702) are not shown because its DDG value was larger than 3.9 kcalÆmol )1 . Error bars are the standard errors calculated from three independent experiments. Fig. 8. Correlation between predicted and observed changes in binding free energy using the independent binding model. DDG B total values were calculated based on the independent binding model, whereby independent binding free energies of monomers of SYC- RP1 to each half site were calculated using Eqn (1). The energy is offset by –kTln2 so that DDG B total is zero when DDG l and DDG r are zero. The broken line is a 45° straight line. The numbers correspond to the sequences in Table 1. Values for number 9 (slr0733 and sll0702) are not shown because its DDG value was larger than 3.9 kcalÆmol )1 . Error bars are the standard errors calculated from three independent experiments. DG-based prediction of DNA binding sites by SYCRP1 K. Omagari et al. 4792 FEBS Journal 275 (2008) 4786–4795 ª 2008 The Authors Journal compilation ª 2008 FEBS Based on the functional annotation, some genes down- stream of the putative binding sites are involved in photoresponsibility (e.g. sll1577 and sll1874). Syne- chocysis responds to blue light and this increases the concentration of cAMP in the cell. Therefore, SYC- RP1 might regulate these genes in response to light. In vivo experiments are necessary in order to confirm whether or not those predicted binding sites are actu- ally control regions for SYCRP1. The present results suggest that DDG values derived from systematic single base-pair experiments can be used to screen potential binding sites and target genes on which regulatory proteins act independently at the genome level. Experimental procedures Preparation of SYCRP1 SYCRP1 used in this study was prepared by the method estab- lished by Yoshimura et al. [1]. The purified SYCRP1 was sus- pended in 50 mm Tris ⁄ HCl (pH 8.0), 200 mm NaCl and 50% glycerol, and stored at –80 °C. The concentration of SYCRP1 was measured using a Protein Assay Kit II (Bio-Rad, Hercu- les, CA, USA), and additional confirmation was obtained using the method described by Gill and von Hippel [29]. Systematic single base-pair substitution experiments and confirmation of binding In order to obtain complete DDG values for positions 4–11 for use in prediction of potential binding sites, we measured the DDG values for positions 9–11 by conducting systematic single base-pair substitution experiments based on EMSA. Ten 40 bp DNA double strands with a single protruding base G at the 5¢ ends were prepared (Fig. 1A). The wild- type sequence used for the reference DDG value was the ICAP sequence that contains the consensus DNA sequence of E. coli CRP (5¢-CAACGCAATAAATGTGA TCTA GATCACATTTTAGGCACCC-3¢). The remaining nine sequences were prepared by systematically substituting the bases that are underlined in the ICAP sequence. All DNA strands were commercially synthesized (Operon, Itabashi, Tokyo, Japan) and purified by HPLC. Binding reactions and electrophoresis were performed according to the method previously reported [16]. Briefly, a DNA double strand labeled with [c- 32 P]ATP (Amersham, Piscataway, NJ, USA) at the 5¢ ends was incubated with a gradient concentration of SYCRP1 in a total volume of 30 lL of binding buffer (50 mm Tris ⁄ HCl pH 7.5, 60 mm NaCl, 1 mm EDTA, 8.3% w ⁄ v glycerol, 0.1 mgÆmL )1 BSA) with a final concentration of 20 lm cAMP for 30 min at room temperature. The DNA concentration was set at a concentration 10- to 1000-fold lower than the K d value. The concentrations of SYCRP1 ranged from 10-fold lower to 10-fold higher than the K d value. The final concentration of SYCRP1 was less than approximately 1000 nm. Samples were quickly loaded onto 10% polyacrylamide gels (acryl- amide:N,N9-methylenebisacrylamide, 50 : 1). Electrophore- sis was performed at a constant voltage (400 V) for 30–45 min in 0.25 · Tris-borate ⁄ EDTA (TBE) with 20 lm cAMP. After electrophoresis, the gels were dried and auto- radiographed using Fujix BAS2500 (Fuji Film, Minato, Tokyo, Japan). From the intensities of the SYCRP1–DNA complex bands and the free DNA bands, the dissociation constant K d and the DDG value were calculated as described by Omagari et al. [16]. Search for potential binding sites for SYCRP1 To search for the potential binding sites for SYCRP1 in the genome, the total change in binding free energy (DDG A total ) for a given segment of the genome sequence was calculated using a mutation matrix as described previously [10]. Figure 3 shows the procedure for this calculation. First, a 16 bp sequence segment was extracted from the +1 posi- tion in the genome sequence. The sequence was compared with the 16 bp consensus sequence of the binding site, and then the DDG values for base-pair substitutions were deter- mined by referring to the mutation matrix for SYCRP1. The total change in binding free energy (DDG A total ) was cal- culated by summing the DDG values at positions 4–19. As the DDG A total value increases, the binding becomes weaker. Next, the position of the 16 bp segment window was shifted by 1 bp at a time, and the same calculations were repeated for the whole genome sequence to investigate the distribu- tion of potential specific binding sites for SYCRP1. Those sites with DDG A total < 3.9 kcalÆmol )1 were selected as poten- tial binding sites. Those sites with DDG A total > 3.9 kcalÆ mol )1 were considered as non-specific binding sites for SYCRP1, because complex bands could not be obtained clearly. Finally, the potential binding sites upstream of or between ORFs were selected as putative binding sites for SYCRP1 in transcriptional regulation. Confirmation of binding SYCRP1 binding to the putative binding sites was experi- mentally confirmed using EMSA. The confirmation was carried out for the putative binding sites with the 11 lowest DDG A total values (Fig. 1B). Eleven DNA double strands of 40 bp with a single protruding base at the 5¢ end labeled with [c- 32 P]ATP were prepared by annealing DNA single strands that had been commercially synthesized (Operon) and purified by HPLC. The double strands have the selected 16 bp putative binding sites in the center. The dis- sociation constant K d and the total change in binding free energy (DDG A total ) for these double strands were measured as previously described [12,16]. K. Omagari et al. DG-based prediction of DNA binding sites by SYCRP1 FEBS Journal 275 (2008) 4786–4795 ª 2008 The Authors Journal compilation ª 2008 FEBS 4793 Acknowledgements We thank Professor A. Suyama for assistance and dis- cussion. This work was supported in part by a grant-in- aid from the 21st century Center of Excellence program (Research Center for Integrated Science) of the Ministry of Education, Culture, Sports, Science, and Technology, Japan. References 1 Yoshimura H, Hisabori T, Yanagisawa S & Ohmori M (2000) Identification and characterization of a novel cAMP receptor protein in the cyanobacterium Synechocystis sp. PCC 6803. J Biol Chem 275, 6241– 6245. 2 Yoshimura H, Yanagisawa S, Kanehisa M & Ohmori M (2002) Screening for the target gene of cyanobacteri- al cAMP receptor protein SYCRP1. Mol Microbiol 43, 843–853. 3 Yoshimura H, Yanagisawa S, Kanehisa M & Ohmori M (2002) A cAMP receptor protein, SYCRP1, is responsible for the cell motility of Synechocystis sp. PCC 6803. Plant Cell Physiol 43, 460–463. 4 Ohmori M & Okamoto S (2004) Photoresponsive cAMP signal transduction in cyanobacteria. Photochem, Photobiol Sci 3, 503–511. 5 Terauchi K & Ohmori M (2004) Blue light stimulates cyanobacterial motility via a cAMP signal transduction system. Mol Microbiol 52, 303–309. 6 Sarai A & Kono H. (2003) DNA-Protein Interactions: Target predictions. In Handbook of Computational Biology (Crabbe MJC & Konopka AK, eds) pp. 241– 278. Marcel Dekker Inc., New York. 7 Stormo GD & Fields DS (1998) Specificity, free energy and information content in protein–DNA interactions. Trends Biochem Sci 23 , 109–113. 8 Frech K, Quandt K & Werner T (1997) Finding pro- tein-binding sites in DNA sequences: the next genera- tion. Trends Biochem Sci 22, 103–104. 9 Kono H & Sarai A (1999) Structure-based prediction of DNA target sites by regulatory proteins. Proteins 35, 114–131. 10 Deng QL, Ishii S & Sarai A (1996) Binding site analysis of c-Myb: screening of potential binding sites by using the mutation matrix derived from systematic binding affinity measurements. Nucleic Acids Res 24, 766–774. 11 Takeda Y, Sarai A & Rivera VM (1989) Analysis of the sequence-specific interactions between Cro repressor and operator DNA by systematic base substitution experiments. Proc Natl Acad Sci USA 86, 439–443. 12 Sarai A & Takeda Y (1989) Lambda repressor recog- nizes the approximately 2-fold symmetric half-operator sequences asymmetrically. Proc Natl Acad Sci USA 86, 6513–6517. 13 Tanikawa J, Yasukawa T, Enari M, Ogata K, Nishim- ura Y, Ishii S & Sarai A (1993) Recognition of specific DNA sequences by the c-myb protooncogene product: role of three repeat units in the DNA-binding domain. Proc Natl Acad Sci USA 90, 9320–9324. 14 Hao D, Yamasaki K, Sarai A & Ohme-Takagi M (2002) Determinants in the sequence specific binding of two plant transcription factors, CBF1 and NtERF2, to the DRE and GCC motifs. Biochemistry 41, 4202– 4208. 15 Gunasekera A, Ebright YW & Ebright RH (1992) DNA sequence determinants for binding of the Escheri- chia coli catabolite gene activator protein. J Biol Chem 267, 14713–14720. 16 Omagari K, Yoshimura H, Takano M, Hao D, Ohmori M, Sarai A & Suyama A (2004) Systematic single base- pair substitution analysis of DNA binding by the cAMP receptor protein in cyanobacterium Synechocystis sp. PCC 6803. FEBS Lett 563, 55–58. 17 Cameron AD & Redfield RJ (2006) Non-canonical CRP sites control competence regulons in Escherichia coli and many other gamma-proteobacteria. Nucleic Acids Res 34, 6001–6014. 18 Pyles EA, Chin AJ & Lee JC (1998) Escherichia coli cAMP receptor protein–DNA complexes. 1. Energetic contributions of half-sites and flanking sequences in DNA recognition. Biochemistry 37, 5194–5200. 19 Parkinson G, Wilson C, Gunasekera A, Ebright YW, Ebright RE & Berman HM (1996) Structure of the CAP-DNA complex at 2.5 angstroms resolution: a com- plete picture of the protein–DNA interface. J Mol Biol 260, 395–408. 20 Shanblatt SH & Revzin A (1986) The binding of catab- olite activator protein and RNA polymerase to the Esc- herichia coli galactose and lactose promoters probed by alkylation interference studies. J Biol Chem 261, 10885– 10890. 21 Man TK & Stormo GD (2001) Non-independence of Mnt repressor–operator interaction determined by a new quantitative multiple fluorescence relative affinity (QuMFRA) assay. Nucleic Acids Res 29, 2471–2478. 22 Bulyk ML, Johnson PL & Church GM (2002) Nucleo- tides of transcription factor binding sites exert interde- pendent effects on the binding affinities of transcription factors. Nucleic Acids Res 30, 1255–1261. 23 Benos PV, Bulyk ML & Stormo GD (2002) Additivity in protein–DNA interactions: how good an approxima- tion is it? Nucleic Acids Res 30, 4442–4451. 24 Benos PV, Lapedes AS & Stormo GD (2002) Is there a code for protein–DNA recognition? Probab(ilistical)ly Bioessays 24, 466–475. 25 Berggrun A & Sauer RT (2001) Contributions of dis- tinct quaternary contacts to cooperative operator bind- ing by Mnt repressor. Proc Natl Acad Sci USA 98, 2301–2305. DG-based prediction of DNA binding sites by SYCRP1 K. Omagari et al. 4794 FEBS Journal 275 (2008) 4786–4795 ª 2008 The Authors Journal compilation ª 2008 FEBS 26 Jin Y, Mead J, Li T, Wolberger C & Vershon AK (1995) Altered DNA recognition and bending by inser- tions in the alpha 2 tail of the yeast a1 ⁄ alpha 2 home- odomain heterodimer. Science 270, 290–293. 27 Li T, Stark MR, Johnson AD & Wolberger C (1995) Crystal structure of the MATa1 ⁄ MAT alpha 2 homeod- omain heterodimer bound to DNA. Science 270, 262– 269. 28 Gromiha MM, Siebers JG, Selvaraj S, Kono H & Sarai A (2004) Intermolecular and intramolecular readout mechanisms in protein–DNA recognition. J Mol Biol 337, 285–294. 29 Gill SC & von Hippel PH (1989), Calculation of protein extinction coefficients from amino acid sequence data. Anal Biochem 182, 319–326 (erratum appears in Anal Biochem 189, 283). K. Omagari et al. DG-based prediction of DNA binding sites by SYCRP1 FEBS Journal 275 (2008) 4786–4795 ª 2008 The Authors Journal compilation ª 2008 FEBS 4795 . DG-based prediction and experimental confirmation of SYCRP1-binding sites on the Synechocystis genome Katsumi Omagari 1 , Hidehisa Yoshimura 2 ,. sites. Figure 3 shows the procedure for the DDG-based prediction. The binding affinity of SYCRP1 to a fragment of 16 bp is estimated as the sum of the DDG values at each position. The window of. shown. The posi- tions of slr1667 and slr1668 are shown at the top; the arrows rep- resent the actual binding site of the slr1667–slr1668 operon. The binding site upstream of the operon has the