Figure 1.
Reverse-transcriptase PCR showing relative transcriptional expression of the three B. bacteriovorus HD100 dacB homologues.
RNA was isolated at the indicated time points during one round of synchronous Bdellovibrio infection of E. coli host cells, and primers specific to the genes bd3244 (dacB), bd0816 and bd3459 (dacB-like) were used to amplify an approximately 100 bp internal fragment from each transcript if present. bd3244 expression shows constitutive expression across the Bdellovibrio lifecycle, (including 2–3 hour septation period) as would be expected for a “peptidoglycan-housekeeping” gene. bd0816 and bd3459 show a strong peak of expression at 15 minutes post-infection, when Bdellovibrio has attached to prey cells and is beginning to invade. The expression diminishes rapidly at the 45 minute to 1 hour time point once invasion is complete and the prey bdelloplast is formed, and before septation. This is indicative of predation-specific expression. AP = Attack phase (free-swimming Bdellovibrio); 15, 30, 45 = minutes since start of infection (attachment to and rounding of E. coli host cell into a bdelloplast structure); 1 h, 2 h, 3 h, 4 h = hours since start of infection (host cell resources being degraded and used for Bdellovibrio growth into a filament, followed by septation into multiple progeny and eventually lysis from host cell); Ec = E. coli S17-1 RNA (no Bdellovibrio control); gen = B. bacteriovorus HD100 genomic DNA (positive control); NT = No template control; L = NEB 100 bp DNA ladder, arrowed is 100 bp.
Figure 2.
Electron micrographs and roundness analysis of wild type and dacB knockout-mutant infected E. coli prey cells.
Images were taken 90-minutes post-invasion and the roundness of infected prey cells were analysed using ImageJ software. Invasion by A: Wild-type HD100 (100% of bdelloplasts were rounded, N = 25); B: HD100 Δbd0816 (100% of bdelloplasts were rounded, N = 21); C: HD100 Δbd3459 (75% of bdelloplasts were rounded, N = 24); D: HD100 Δbd0816 Δbd3459 (DKO) double mutant (<10% bdelloplasts were rounded, N = 53). E: Bar graph showing the average roundness coefficient of the bdelloplasts from EM images with error bars showing 95% confidence intervals and statistical analysis of the means compared to WT (ns = not significant; ** = p<0.005; *** = p<0.001). Data is taken from 4 independent experiments. All displayed images were taken from the same experimental repeat at 80 kV (20,000-fold magnification) and stained with 1% PTA for 1 min. Scale bar is 500 µm.
Figure 3.
Histograms and images of B. bacteriovorus HD100 and Δbd0816 Δbd3459 strains infecting E. coli K-12 MG1655 and A. baumannii.
Histograms of mean times for attachment (A) and invasion (B) by B. bacteriovorus HD100 wild type (WT) and Δbd0816 Dbd3459 (DKO) strains infecting E. coli K-12 MG1655 (black fill) and A. baumannii (grey fill); (Mean attachment time - as measured from initial Bdellovibrio contact with outside of prey cell to the start of traversal through the prey cell wall. Mean invasion time - as measured from the start of traversal through the prey cell wall to not being visible outside the prey cell, i.e. being completely within the prey cell) (N = 50 for all experiments). At least two independent experiments were carried out with error bars showing 95% confidence intervals and statistical analyses shown (ns = not significant; *** = p<0.001). Time-lapse images showing prey rounding relative to invasion for C) E. coli and D) A. baumannii by wild type HD100 Bdellovibrio. Arrowed box shows relative time at which rounding first becomes obvious. E. coli begins to round on average 2.3 minutes before invasion (N = 50), whereas A. baumannii begins to round on average 10.7 minutes before invasion (N = 50). E. coli finishes prey rounding concurrently with the completion of invasion, whereas A. baumannii tends to have its rounding completed earlier. Refer to results section for more information. Scale bar is 2 µm.
Figure 4.
Frequency of double invasions of E. coli prey during attack by B. bacteriovorus HD100 wild type, HD100 Δbd0816, HD100 Δbd3459 or B. bacteriovorus HD100 Δbd0816 Δbd3459 (DKO) strains.
Time-lapse microscopy was used to observe invasions when two Bdellovibrio cells were seen to attach to a single prey cell for longer than 10 minutes (longer than the usual ‘recognition’ period) and was taken to be an ‘attempted invasion’ by both cells. The outcomes of invasions (N = 42 for HD100 WT, N = 40 for both HD100 Δbd0816 and HD100 Δbd3459, N = 44 for HD100 Δbd0816 Δbd3459) were tallied into one of three categories; i) ‘single infection’ (black bars; just one Bdellovibrio successfully invades); ii) ‘synchronous infection’ (grey bars; both invade successfully at the same time); or iii) ‘tailgating infection’ (white bars; one invades followed by the other). The double mutant Bdellovibrio strain showed a much higher proportion of double invasions and therefore decreased predatory population fitness due to self competition in single prey. The absence of Bd3459 has less of an effect on double invasions than the absence of Bd0816 although both single mutant datasets show an intermediate effect between WT and the double mutant. Data were taken from at least two experimental repeats. See Materials and Methods for more information.
Figure 5.
Alignment of Bdellovibrio “housekeeping” Bd3244 and predatory Bd0816/Bd3549 proteins versus housekeeping PBP4 of E. coli.
The structural elements of the E. coli PBP4 and the Bd3459 proteins are displayed respectively below and above the alignment. The central structural differences between Bd3459/Bd0816 PBP4s and E. coli PBP4/Bd3244 (reflected in missing/altered domains in the structure (see Figure 8) can be clearly seen.
Figure 6.
Expression of full-length Bd3459 protein alone, but not a S70A active site variant, causes rounding/lysis of E. coli.
Time-lapse microscopy of E. coli TOP10 cells containing either wild type (WT) or active site variant (S70A) Bd3459 protein being overexpressed from an L-arabinose inducible plasmid by addition of arabinose at 0 hours. Cells were immobilised on a YT-agarose pad (YT) or osmotically stabilised M-medium-agarose pad (M), containing 0.2% (v/v) L-arabinose and development over 6 hours is shown. A = Bd3459 (WT) on YT; B = Bd3459 (S70A) on YT; C = Bd3459 (WT) on M; D = Bd3459 (S70A) on M. Osmotically unsupported E. coli expressing WT Bd3459 lyse and osmotically supported E. coli round up. E. coli expressing the S70A protein variant continues to grow normally. Also see Videos S13, S14, S15, S16.
Figure 7.
Activity of Bd3459 and its inhibition by ampicillin.
(A) Pentapeptide-rich peptidoglcyan was incubated with Bd3459, Bd3459 pre-incubated with ampicillin, Bd3459 (S70A) or no enzyme, followed by digestion with cellosyl, reduction with sodium borohydride and analysis of the resulting muropeptides by HPLC. Muropeptides: 4, Tetra; 5, Penta; 44, TetraTetra; 45, TetraPenta. Peaks originating from ampicillin are indicated by a star. Bd3459, but not Bd3459 (S70A) or ampicillin-blocked Bd3459, digested Penta and TetraPenta to Tetra indicative of a DD-endo/carboxypeptidase activity. (B) Structures of the reduced muropeptides. GlcNAc, N-acetylglucosamine; MurNAc(r), N-acetylmuramitol; L-Ala, L-alanine; D-iGlu, D-isoglutamic acid; m-Dap, meso-diaminopimelic acid; D-Ala, D-alanine. (C) Bd3459 and Bd3459 (S70A) were incubated with or without ampicillin, followed by labelling with Bocillin-FL and SDS-polyacrylamide gel electrophoresis. Bocillin-FL was detected by fluorescence (Bocillin-FL); total protein was visualized by staining with Coomassie Blue. Bocillin-FL-binding of Bd3459 was greatly inhibited by pre-incubation with ampicillin; Bd3459 (S70A) bound substantially less Bocillin-FL than Bd3459. Thus, Bd3459 is a penicillin-binding protein and degrades isolated cell wall material.
Figure 8.
Crystal structure of B. bacteriovorus Bd3459 revealing predatory adaptations differing from the “PG-housekeeping” PBP4 of E. coli.
Bd3459 left hand panel, E. coli PBP4 centre panel, coloured thus: TP domain, white; domain II, yellow; domain III, orange; “insert” hairpin, purple; C-terminal “staple”, red. The active site serine residues of Bd3459 (S70 sulfonyl adduct) and E. coli PBP4 (S62) are shown in stick form. Domain III of E. coli PBP4 acts to functionally constrict the active site (for a more attenuated/regulated self-peptidoglycan metabolism), whereas Bd3459 lacks this and thus may act promiscuously on a range of larger peptidoglycan substrates. This would fit with the predatory role of Bd3459 shown genetically. Right hand panel is composed of Bd3459 (blue, with S70 coloured in red) and E. coli PBP4 (orange) superimposed via their transpeptidase domains, illustrating the relatively more solvent-exposed active site of Bd3459 and respective differences in domains II and III.
Figure 9.
Structural comparison detail of overlapping active site residues (SXXK, SXN, KTG) and extra “hairpin” domain of Bd3459 versus E. coli “housekeeping” PBP4 protein.
Comparison of E. coli PBP4 (coloured yellow, active site serine S62, only selected residues shown) and Bd3459 (coloured white, active site serine S70) active sites. The additional hairpin domain of Bd3459 is shown from N292–N303. Despite the more accessible conformation shown in Figure 8, Bd3459 retains the conserved catalytic apparatus of PBP4 enzymes, in a productive (predicted active) conformation (tallying with the activity assays, and observance of a covalent adduct). Insert (right) shows the refined density for the sulphonyl linkage to S70 (1.2σ, 1.45 Å resolution).
Table 1.
Data collection and refinement statistics.