Parallel evolution leading to impaired biofilm formation in invasive Salmonella strains

Abstract

Pathogenic Salmonella strains that cause gastroenteritis are able to colonize and replicate within the intestines of multiple host species. In general, these strains have retained an ability to form the rdar morphotype, a resistant biofilm physiology hypothesized to be important for Salmonella transmission. In contrast, Salmonella strains that are host-adapted or even host-restricted like Salmonella enterica serovar Typhi, tend to cause systemic infections and have lost the ability to form the rdar morphotype. Here, we investigated the rdar morphotype and CsgD-regulated biofilm formation in two non-typhoidal Salmonella (NTS) strains that caused invasive disease in Malawian children, S. Typhimurium D23580 and S. Enteritidis D7795, and compared them to a panel of NTS strains associated with gastroenteritis, as well as S. Typhi strains. Sequence comparisons combined with luciferase reporter technology identified key SNPs in the promoter region of csgD that either shut off biofilm formation completely (D7795) or reduced transcription of this key biofilm regulator (D23580). Phylogenetic analysis showed that these SNPs are conserved throughout the African clades of invasive isolates, dating as far back as 80 years ago. S. Typhi isolates were negative for the rdar morphotype due to truncation of eight amino acids from the C-terminus of CsgD. We present new evidence in support of parallel evolution between lineages of nontyphoidal Salmonella associated with invasive disease in Africa and the archetypal host-restricted invasive serovar; S. Typhi. We hypothesize that the African invasive isolates are becoming human-adapted and 'niche specialized' with less reliance on environmental survival, as compared to gastroenteritis-causing isolates.

Fig 1. The biofilm phenotypes of representative nontyphoidal and typhoidal Salmonella strains.

Salmonella strains that are known to cause gastroenteritis or enterocolitis (green), invasive disease (blue) or typhoid fever (red) were screened: for the ability to form the red, dry, and rough (rdar) morphotype, which presents as the formation of concentric rings and a wrinkled appearance on the surface of macrocolonies (panels 1 and 2, top); for the presence of multicellular, biofilm aggregates and planktonic cells in liquid cultures grown under biofilm-inducing conditions (middle panel; conical tubes); and for cellulose production, visualized as the white and fluorescent appearance of macrocolonies in the presence of calcofluor white dye (bottom panel).

Fig 2. Detection of CsgD and CsgA protein synthesis in representative Salmonella strains by Western blot.

Whole-cell lysates were derived from multicellular aggregates [MA] or planktonic cells [PC] harvested from flask cultures of nontyphoidal and typhoidal Salmonella. Lysates used for CsgD detection (top panel) were normalized based on total protein concentration. Purified CsgD-6xHis recombinant protein was used as a technical control for CsgD detection. Pooled control samples were derived from combining lysates obtained from the multicellular aggregates of S. Typhimurium 14028 and SL1344 and S. Enteritidis 4931 and 301 strains. Black arrows indicate the detection of CsgA subunit dimers (D) and monomers (M) (bottom panel). The data shown is representative of two biological replicates.

Fig 3. Evaluation of the csgDEFG-csgBAC intergenic region for changes in promoter sequence and activity.

(A) Diagram representing the divergent csg operons. Transcriptional start sites are indicated as black elbow arrows. The square bracket indicates the region analyzed in (B). (B) Multiple sequence alignment of the intergenic and 5ʹ untranslated regions for the csgDEFG and csgBAC operons from Salmonella strains in this study. A neighbour-joining dendrogram was established based on bootstrapping parameters set to 1,000 replicates and a support threshold of 70%. Transcription factor binding sites that have been experimentally verified in Salmonella are indicated above the consensus sequence. Single nucleotide polymorphisms (SNPs) within each strain’s DNA sequence are indicated as coloured rectangles within grey tracks (red, adenine; blue, cytosine; yellow, guanine; green, thymine). Red stars and yellow vertical blocks highlight SNP positions unique to S. Typhimurium D23580 and S. Enteritidis D7795. (C) Promoter-reporter fusion constructs were generated using csgD and csgB promoter sequences derived from the Salmonella strains. The activity of each construct was evaluated in S. Typhimurium 14028 cells during 48 hours of growth. Graphed values represent the maximum reporter activity recorded in this period and is reported as counts per second (CPS). Statistical significance: *, P < 0.05; **, P < 0.01. (D) Constructs derived from S. Typhimurium 14028 csgD and csgB promoter sequences were introduced into each Salmonella strain. Letters above the bars indicate mean values that were statistically similar to (black font) or different from (red font) other mean values. #, values below the activity threshold as established in [24]. Each bar represents the mean value from three to five biological replicates. Error bars represent standard deviations.

Fig 4. Biofilm phenotypes and csgD promoter activities following chromosomal replacement of strain-specific csgD promoter sequences in S. Enteritidis D7795 and S. Typhimurium 14028.

(A) Genome engineering was used to replace part of the native csgD promoter sequence in S. Enteritidis D7795 with sequence from S. Typhimurium 14028. The same process was also used to replace part of the native sequence in S. Typhimurium 14028 cells with sequence from S. Enteritidis D7795. The 780-bp csgD promoter region was PCR amplified from the strains and clones listed, followed by digestion with PsiI, which has a recognition site overlapping the S. Enteritidis D7795 SNP-containing region. (B) Macrocolony phenotypes of S. Enteritidis D7795 and S. Typhimurium 14028 clones that either contain or do not contain the identified ‘T’ promoter SNP at position -47. (C) The csgD promoter from each S. Enteritidis D7795 clone was used to generate promoter luciferase reporters that were transformed into S. Typhimurium 14028 cells. Each line represents one biological replicate culture (n = 22) with measured promoter activity (CPS, counts per second) plotted versus time. (D) Maximum reporter activity recorded from csgD promoter luciferase constructs derived from S. Typhimurium 14028 clones and transformed into wildtype S. Typhimurium 14028 cells (n = 11 per reporter construct). The activity of wildtype (WT) S. Typhimurium 14028 and S. Enteritidis D7795 csgD promoter-reporter constructs were included in the assay (n = 4 per each construct). Violin plots show the frequency distribution of the data, with the dotted line representing the median value. ****, P < 0.0001.

Fig 5. Biofilm phenotypes and csgD promoter activities following chromosomal replacement of strain-specific csgD promoter sequences or bcsG mutations in S. Typhimurium D23580 or S. Typhimurium 14028.

(A) Top panel: Maximum activity (CPS; counts per second) was recorded for csgD promoter luciferase reporters derived from S. Typhimurium D23580 clones generated by genome engineering that either contained or did not contain the ‘A’ SNP at position -80 and ‘A’ SNP at position -189 relative to the csgD transcriptional start site. Reporters were transformed into S. Typhimurium 14028 and activity monitored during 48 hours of growth (n = 28 per reporter construct). Wildtype (WT) reporters from S. Typhimurium D23580 (n = 24) and S. Typhimurium 14028 (n = 12) were included as controls. Violin plots show the frequency distribution of the data, with the dotted line representing the median value; **, P < 0.01; ****, P < 0.0001. Bottom panel: 48 h time-course expression profiles for csgD promoter luciferase reporters analyzed in the top panel, except for S. Typhimurium 14028; each line represents one biological replicate culture. (B) Macrocolony phenotypes of S. Typhimurium D23580 and S. Typhimurium 14028 clones containing native or replacement csgD promoter sequences (left panel) or bcsG alleles (right panel).

Fig 6. Conservation of csgD promoter and bcsG single nucleotide polymorphisms in invasive S. Enteritidis and S. Typhimurium lineages.

Maximum likelihood phylogenic trees were constructed from bacterial genome sequences: (A) S. Enteritidis isolates from Feasey et al. [10], and (B) S. Typhimurium isolates from Okoro et al [48], keeping the same general tree shape for comparison purposes. (A) S. Enteritidis isolates were divided into the Central/East African clade (167 isolates) and global epidemic clade (250 isolates), with the distinct region of isolation shown along with presence or absence of the ‘T’ SNP. (B) S. Typhimurium isolates were divided into gastroenteritis-associated and invasive lineages (I and II), with the presence or absence of csgD promoter and bcsG polymorphisms shown.

Fig 7. Functional analysis of the S. Typhi CT18 csgD allele.

(A) The csgD alleles from S. Typhimurium 14028 and S. Typhi CT18 were cloned into p3xFLAG and transformed into S. Typhimurium 14028 ΔcsgD cells. Colony morphology and flask cultures of uninduced cells were evaluated for biofilm phenotypes. (B) Top panel: Whole cell lysates were generated from cells acquired from flask cultures and probed for synthesis of CsgD via Western blot. Bottom panel: A csgB promoter-reporter construct was used to evaluate CsgD activity in S. Typhimurium 14028 ΔcsgD cells harbouring p3xFLAG-csgD^CT18 with or without IPTG induction. (C) Top panel: Presence or absence of multicellular aggregates in flask cultures of S. Typhimurium 14028 ΔcsgD cells containing p3xFLAG constructs (+/- IPTG) or p3xFLAG alone. Bottom panel: Maximum promoter activity of a csgB promoter-reporter construct measured in microbroth cultures corresponding to the tubes shown in the top panel. (D) Biofilm phenotypes of S. Typhimurium 14028 strains that contained the native csgD^14028 allele or the truncated csgD allele from S. Typhi CT18 following genome engineering.