We have shown that Bdellovibrio express two related PBP4-like proteins Bd3459 and Bd0816, evolutionarily adapted for predation, which act on the prey peptidoglycan of diverse bacterial prey at the invasion stage of predation speeding the process and reducing self-competitive, double prey-entry events. Transcription of both genes () is short-lived peaking at 15 minutes into the predatory life cycle. As predatory growth proceeds by a series of prey-hydrolytic and predator-developmental processes, it is likely that the gene products are regulated post-transcriptionally so that their lytic activity is appropriately controlled. This may be by targeted PBP4 protein degradation or by specific inhibitors but determining this is beyond the scope of our study.
The action of the predatory PBP4s is seen as the rounding of the prey cell prior to Bdellovibrio
entry () making an osmotically stable bdelloplast structure but with more deformable peptidoglycan than the original prey cell. Previous studies 
had considered that the increased intracellular space that prey rounding provides allows Bdellovibrio
to overcome space constraints and maximise the yield of progeny Bdellovibrio
per prey invasion. For E. coli
prey we did not find this to be true, (although it may be more important in smaller sized prey) and this infers that prey rounding may be an unintentional by-product of the action of these enzymes which confer other predatory fitness advantages. Indeed, we have shown that the action of the Bd3459 and Bd0816 enzymes enable significantly faster invasions by WT Bdellovibrio
than a double deletion mutant, particularly in prey with a higher proportion of peptidoglycan cross-links (e.g. Acinetobacter
). These enzymes, each and together also inhibit self-competitive “tailgating” prey invasions, further increasing the Bdellovibrio
predatory population fitness.
We consider that the peptidoglycan-de-cross-linking, endopeptidase action of the Bd3459 and Bd0816 enzymes may be important for one or both of the following reasons:
- Reduction of the torsional stress of the prey cell wall pushing out on the Bdellovibrio as it tries to enter the periplasm- making it quicker to enter prey, and reach the inner membrane of the prey and kill it; thus the least time possible is spent by the Bdellovibrio as a “sitting duck” attached to a live prey bacterium which could swim away with it to a non-favourable environment. Because of the de-cross-linking activity of Bd3459 and Bd0816 the Bdellovibrio rapidly enters the prey cell and kills it and then has environmental protection from being inside it. The differing proportion of peptidoglycan cross-linking that the E. coli (33%) and Acinetobacter (61%) possess  may be a contributing reason why there are differing invasion times () between the two strains. Although other factors such as the presence of a capsule or other peptidoglycan differences may be responsible for this, one hypothesis is that in Acinetobacter, more de-cross-linking and therefore time would have to occur to achieve a minimum threshold through which the invading Bdellovibrio can fit. When the Bdellovibrio are attached to the outside of prey (by the Bdellovibrio non-flagellar pole) they are secreting the Bd3459 and Bd0816 enzymes to assist with prey cell wall modification for entry.
- If the first Bdellovibrio invading a prey cell makes its peptidoglycan more “flaccid” and deformable (Video S5), by peptidoglycan de-cross-linking due to due to predatory PBP4-like protein action, as it enters it; then subsequently arriving Bdellovibrio find the prey wall softened. Atomic force microscopy studies on bdelloplasts versus prey cells  support this idea, showing that the spring constant of uninvaded E. coli prey cells was 3 times stiffer than that of bdelloplasts. As leverage inward by retraction of type IV pili bound to prey peptidoglycan is suggested as a predatory entry mechanism , , pili retracting against a less rigid, de-cross-linked prey cell wall (after action of the Bd3459 and Bd0816 enzymes) may be less successful and therefore the second Bdellovibrio will fail to enter. This prevents two Bdellovibrio competing for the same finite nutritional resource in a prey single cell and increases the fitness of the population.
The results from our experiments show that neither of the Bd0816 or Bd3459 proteins are required for prey entry itself. Previous studies 
predicted that endopeptidase activity would not be required to make the pore in the peptidoglycan through which the Bdellovibrio
invades. Those experiments predicted that this role would be carried out by a carefully regulated lytic transglycosylase. Transcriptional studies from our lab 
showed that bd3575
, encoding a lytic transglycosylase, is highly upregulated upon Bdellovibrio
invasion. Deletion of Bd3575 still allowed invasion to occur, so it is possible that there is either another lytic transglycosylase to be found, another mechanism completely is responsible, or it is a combination of Bd3575, Bd0816 and Bd3459 that enable invasions; in this case a triple mutant may be required to abolish invasion.
In the Δbd0816 Δbd3459 mutant, we hypothesise that the cell wall remains rigid for longer and allows secondary invasions mediated by pilus retraction. In this case of extra bdelloplast wall rigidity, how might there be enough room for the Δbd0816 Δbd3459 mutant Bdellovibrio to grow in the periplasm? We do not rule out that there may be other enzymes acting later in predatory development that allow the bdelloplast to expand to accommodate the progeny Bdellovibrio. It is also possible that solely the physical pressure of the internal Bdellovibrio cells pressing upon the stretchable peptidoglycan allows growth (the outward turgor pressure would be favourable for this, unlike the prey invasion process itself which fights against prey cell turgor pressure).
Predatorily specialised bd3459 dacB
-like genes are retained in addition to the single housekeeping bd3244
-like () gene in the genomes of other Bdellovibrios; the marine Bdellovibrio
-like organism Bacteriovorax marinus
has a single predatory homologue (Bms2102) and housekeeping homologue (Bms2612) (Stine et al.
2010, unpublished data; GenBank ID: FQ312005.1). The retention of these genes emphasises the importance of their encoded products for fitness. The retention of duplicate predation-related genes is seen in other important Bdellovibrio
systems like those encoding flagella 
and their motors 
. We speculate that Bdellovibrio
can therefore repair gene damage by internal recombination from a “spare” gene sequence – clearly Bd3459 is more phenotypically active for prey cell wall rounding than Bd0816, and Bd0816 more active for tailgating prevention than Bd3459 so they do not directly substitute for each other totally. Bms2102 and Bd0816 are likely to share the same novel predatory structural adaptations, as observed when comparing the structure of Bd3459 to “regular” PBP4 proteins.
The presumed housekeeping Bd3244 protein has a more conventional PBP4 sequence and includes a predicted classical domain III (). Notably it does not have a Sec sequence like those seen on the other Bd0816 and Bd3459 proteins. Bdellovibrio
has Gram-negative typical A1γ peptidoglycan composition 
, but it may have modifications that render it immune/protected from attack, such as a particularly low number of cross-linked trimers/tetramers that Bd0816 and Bd3459 may cleave, or an immunity protein that acts to protect against self peptidoglycan-modification. Recently a type VI secretion system has been shown to be used by Pseudomonas aeruginosa
to inject peptidoglycan hydrolases into the periplasm of Gram-negative bacteria 
protects itself from self-intoxification with a periplasmic immunity protein. It remains to be determined how Bd0816/3459 are secreted into the prey periplasm and whether this involves novel transport mechanisms or solely Sec machinery. Unfortunately there is little available microscopic and electron-tomographic data on the nature of the junctions between Bdellovibrio
and prey cells upon invasion, nor whether they involve direct membrane fusion junctions between the two outer membranes for example. So at present it is difficult to predict secretion mechanisms, as we cannot yet be sure if Bd0816 and Bd3459 proteins need to be secreted across the Bdellovibrio
outer membrane, while it is invading the periplasm of prey, or whether the Bdellovibrio
outer membrane is temporarily partially disrupted, giving direct access from one periplasm to the other. Further extensive tomographic studies are required inside bdelloplasts to answer this. We are working to determine whether or not Bdellovibrio
peptidoglycan is susceptible to these Bd0816/3459 enzymes or is protected by an immunity protein(s) or by peptidoglycan modification. We are investigating the reason, but can confirm Thomashow and Rittenberg's hypothesis of 1978 that cell wall peptidases shape the prey cells into rounded bdelloplasts and to identify the structure, expression and function of those enzymes.
This work has characterised one pair of predatorily-adapted, PBP4-like enzymes from the arsenal of bacterially degrading and modifying proteins that Bdellovibrio
encodes from its “predatosome” (genes transcribed in response to prey interaction) 
, and shown how their structure has evolved to rapidly modify the cell wall chemistries of diverse bacteria that Bdellovibrio
prey upon to allow rapid, and preferentially singular access to a novel intracellular niche.