Operator sequences for the regulatory proteins of restriction modification systems

Sir, For some type II restriction modification (RM) systems, it has been shown that transcription of the methylase gene (M ) and the restriction endonuclease gene (R ) is regulated by the control gene (C ) product (Tao et al. 1991, J Bacteriol173: 1367–1375). In these systems, C is located directly upstream of R, and in most systems M is located divergently from CR (Fig. 1). The control element is a short (≈80 amino acids) protein containing a helix–turn–helix DNA-binding motif, distantly related to the well-known phage lambda cI regulator. . Schematic representation of the gene order in RM systems that contain a gene for a control element. Arrows indicate the positions of the restriction endonuclease (marked R ), methylase (marked M ) and control element (marked C ) genes. Black triangles indicate the position and orientation of the here-proposed control element operators. Data were taken from GenBank entries X16458 (SmaI), U49842 (Bgl II), X52681 (PvuII), X00530 (EcoRV), X55286 (BamHI), X76192 (MunI) and from Rimseliene et al. 1995, Gene 157: 217–219 (Eco72I). The sequence of the NmeSI system will be submitted to GenBank. The DNA-binding sites from other proteins that bind with helix–turn–helix motifs (e.g. lambda cI and the LysR family of regulatory proteins) are usually 7–9 bp in size, of which only few bases directly interact with the protein. Most of these proteins form dimers, having a binding site of 14–18 bp with a dyad symmetry for at least the directly interacting bases. Based on sequence comparisons of the regions upstream of different C genes, two groups have proposed ‘C-box’ operator sequences of 12 bp and 18 bp respectively (Rimseliene et al. 1995, Gene157: 217–219; Anton et al. 1997, Gene187: 19–27). The proposed C-box consensus sequences are unnecessarily large for a binding site of a monomeric form of the regulator. These sequences are also unlikely binding sites for a dimeric form of the regulatory protein because they lack dyad symmetry. In addition, control elements differing in the recognition helices of their respective helix–turn–helix motifs are unlikely to share identical operator sequences. Dissimilarity of the respective operator sequences is supported by the limited complementation of a bamHIC mutation by multicopy pvuIC and smaIC. In addition, a pvuIC mutation was better complemented by bamHIC than vice versa (Ives et al. 1995, J Bacteriol177: 6313–6315). The C gene product exerts a negative influence on the transcription of M and a positive effect on R transcription (Rimseliene et al. 1995, Gene157: 217–219). The operators for M and R might overlap for RM systems with divergent gene order. However, in RM systems in which the gene order is not divergent, M and C should have separate and most likely identical, operator sequences (Fig. 1). Therefore, we analysed the sequences of the SmaI and NmeSI (recently characterized in our laboratory) systems and the Bgl II system, in which CR and M are orientated convergently and collinearly respectively. For each of these three RM systems, we found identical sequences in opposite orientation in front of the open reading frames of C and M (Fig. 2). These sequences partially overlap the previously suggested C-boxes upstream of the C genes. Given the aforementioned considerations, these sequences are more likely to contain a 7–9 bp control element binding site within them than the previously proposed C-boxes. Noticeably, the recognition helices of SmaIC and NmeSIC are identical, and their sequences in Fig. 2 share the 7 bp sequence TGCTACT. The putative operators in front of R and M are in opposite orientations, suggesting that the orientation of the operator determines whether the regulator exerts upregulation or downregulation. This implies that one operator would be sufficient when the gene order is divergent (Fig. 1). . Alignment of the upstream regions of C and M of the SmaI, NmeSI and Bgl II RM systems. The upstream regions of the M genes are inverted. Distance to the (proposed) start codons of the respective genes are given, except for nmeSIM, where the putative start codon is given in bold typeface. The identical DNA sequences are shaded. The earlier proposed C-boxes are underlined. In conclusion, based on sequence data of RM systems with non-divergent gene order, we postulated operator sequences that are consistent with the dual regulatory role of control elements in RM systems in general.