Phase variation controls expression of Salmonella lipopolysaccharide modification genes by a DNA methylation-dependent mechanism

Abstract

The O-antigen of Salmonella lipopolysaccharide is a major antigenic determinant and its chemical composition forms the basis for Salmonella serotyping. Modifications of the O-antigen that can affect the serotype include those carried out by the products of glycosyltransferase operons (gtr), which are present on specific Salmonella and phage genomes. Here we show that expression of the gtr genes encoded by phage P22 that confers the O1 serotype is under the control of phase variation. This phase variation occurs by a novel epigenetic mechanism requiring OxyR in conjunction with the DNA methyltransferase Dam. OxyR is an activator or a repressor of the system depending on which of its two binding sites in the gtr regulatory region is occupied. Binding is decreased by methylation at Dam target sequences in either site, and this confers heritability of the expression state to the system. Most Salmonella gtr operons share the key regulatory elements that are identified here as essential for this epigenetic phase variation.

Fig. 1

Expression from the gtr^P22 promoter is under control of phase variation in a Dam- and OxyR-dependent manner. Shown are strains with a lacZ transcriptional fusion of gtr^P22 integrated in single copy in the S. Typhimurium LT2 genome. The strain numbers and relevant genotype are shown with the images; mutations in the genome are indicated in sMV139 and sMV136, and the oxyR mutation was complemented with plasmid-encoded oxyR^C199S in sMV154. Dam was over expressed in sMV104 from pTP166 (Marinus et al., 1984). Mutations indicated for sMV147 and sMV171 are in the gtr'-lacZ regulatory region. Blue (Lac+) colonies represent the ON phase and white (Lac−) colonies the OFF phase. A mixture of blue and white colonies indicates phase variation.

Fig. 2

Deletion analyses defines the minimal phase varying regulatory region. A schematic of the gtr region present in each of the strains listed used to define the minimal phase varying region is shown. The base pairs define the length of the region upstream of the +1 transcription start site that is indicated with a line arrow. The Lac phenotype of colonies of the strain is given (Lac+, ON; Lac−, OFF and Lac+/Lac−, phase varying), as well as the level of gtr'-lacZ expression. For phase varying constructs, Miller units are shown as calculated per 100% ON, which is indicated with an asterisk. The positions of the GATC sequences, OxyR binding sites (A, B and C) and RNA polymerase binding site (−10 and −35) are shown on the cartoon below. The four GATC sequences are indicated with a vertical line and the partial gtrA coding sequence with a grey box.

Fig. 7

Regulatory region alignment of predicted phase varying gtr genes. The regulatory regions of gtr^P22, gtr^LT2_I and gtr^PT4_II were aligned. The putative OxyR binding sites are shown along with the consensus OxyR half site binding sequences (Storz et al., 1990; Toledano et al., 1994). GATC sequences are bold and underlined; GATC¹ starts at nt −110; GATC² at −97. GATC³ at −46 and GATC⁴ at −33. The +1 transcription initiation site and −10 and −35 promoter regions are also shown.

Fig. 3

The GATC pairs in the gtr^P22 regulatory region are differentially methylated in phase ON and OFF cells. Southern blot of chromosomal DNA probed with a gtr^P22 regulatory region probe. DNA was digested with MslI and with MboI, DpnI or Sau3AI, as indicated. Control indicates DNA was digested only with MslI. A. Genomic DNA was analysed from cultures with predominantly either cells in the Lac+ (lanes 1–4) or Lac− phase of sMV83 (lanes 5–8). B. DNA analysed from sMV136 (oxyR^-), sMV175 (−10 and −35 only), sMV174 [OxyR(BC) site only] and sMV244 [oxyR^-, OxyR(BC) site only]. C. and D. show the expected band sizes resulting from different digestions of the full length and shorter promoter constructs respectively.

Fig. 4

OxyR^C199S binds to the gtr^P22 promoter region and binding affinity is decreased by GATC methylation. EMSA analysis was performed with increasing amounts of purified OxyR^C199S to unmethylated (‘unmeth’, A–C) or methylated (‘meth’, D–F) gtr^P22 probes. The probes contained the putative OxyR(ABC) binding sites (A and D), only the OxyR(AB) (B and E) or only the OxyR(BC) binding sites (C and F). The concentration of OxyR is indicated. The second band of free DNA represent secondary structure variants of the same DNA sequence.

Fig. 5

OxyR binding confers methylation protection of the GATC pairs in an in vitro methylation protection assay. A. Schematic showing the probe and the possible MboI digest products, labelled as A–C. The top line represents the probe with identification of the 5′ FAM label and position of the GATC sequences. FAM-labelled fragments that are detectable are shown in black; large fragments that are not detectable in this assay are shown in grey. Fragment C can arise by either digestion pattern as indicated: its presence thus does not provide information on the methylation state of GATC³/GATC⁴. B. Correlation between the percentage of DNA bound and the methylation protection of the GATC pairs (see text). The latter is inferred from the quantity of the digestion products A-C as illustrated in (A). Data points are from analysis of gtr^P22 probe and OxyR^C199S at 0, 21, 30, 42 and 105 nM. The percentage of DNA in complex with OxyR^C199S is derived from quantified data from the EMSA. The percentage of each of the three bands indicated in (A) is derived from MboI digest and indicates methylation protection. A linear trendline is shown for each that shows increasing methylation protection with increased amount of DNA-OxyR^C199S complex.

Fig. 6

Quantitative analysis of the expression of gtr^P22 promoter. Expression of various single copy gtr^P22-lacZ reporter constructs in different genetic backgrounds are shown. (A) Expression in a wild-type background, a dam mutant and dam over expression strain (pTP166) as indicated and (B) in an oxyR mutant background. The gtr^P22 sequences are defined as, OxyR(ABC) (nt −278 to +34); −10/−35 (nt −45 to +34); CATC¹/GATG² as GATC mutants (in context of nt −278 to +34); OxyR(BC) (nt −95 to +34) (also see Fig. 2). For phase varying constructs the % ON cells was determined and results are presented as calculated Miller units per 100% ON culture.

Fig. 8

Model for gtr phase variation. A. Cartoon illustrating the protein–DNA interactions at the gtr^P22-like regulatory regions and the methylation state of the GATC pairs for cells in the ON and OFF phase. The promoter is indicated (−10, −35), the GATC pairs (stick-balls; filled for methylated and open for unmethylated state), and the OxyR (light grey ovals) and RNA polymerase (darker shape) interacting with DNA as indicated (also see inset). Line arrow indicates transcription, and open arrow with cross lack of RNA polymerase binding. Dam is not included in the figure, but would not be able to access GATC sites occupied by OxyR. Cartoon not to scale. B. A comparison of the models for pap, agn and gtr Dam-dependent phase variation showing modularity. Shown are GATC sites, transcription start site indicted as above, and the regulatory protein bound (OxyR, light grey ovals; Lrp, dark oval; PapI, triangle) in the ON and OFF phase as indicated (also see inset). Cartoons not to scale.