Interactions between Wolbachia wBm proteins and host proteins
UPDATE: New publication by Mark Taylor's group (here) follows up this wBm0152 story and finds that it is not secreted (at least, not found in the cytosolic fraction), but found associated with the microbial cell membrane, as may be expected based on studies from other model systems (such as E. coli).
I was excited to see a new publication come out in PLoS Neglected Tropical Diseases focusing on characterizing Wolbachia outer membrane proteins. The paper is titled: “A Potential Role for the Interaction of Wolbachia Surface Proteins with the Brugia malayi Glycolytic Enzymes and Cytoskeleton in Maintenance of Endosymbiosis. ” Because Wolbachia is an obligate intracellular organism for which there is no genetic system, taking a biochemical approach is a good idea, and can be quite fruitful. The paper focuses on the Wolbachia strain wBm found in nematodes that cause elephantiasis (Brugia). Here's a link to the article:
Here's what the authors do, in a nut-shell:1) clone, express, and epitope-tag-purify wBm outer membrane proteins (there are two, paralagous copies of wsp they work on as well as one unrelated OMP, related to the Peptidoglycan Associated Lipoprotein family). They also do the same for a host protein, aldolase.2) They use Elisa binding assays to attempt to show a specific interaction between these outer membrane proteins and crude host extracts (with the negative control of a non-infected nematode extract)3) They use their purified proteins and antibodies made from these proteins to perform a series of overlays to support the specificity of the interactions.4) They perform immuno-gold labeling experiments to show the localization of their proteins in vivo.
So, at first they show that their proteins specifically bind to host extracts. Here's their initial Elisa data, with the non-infected nematode shown as a dashed line. They test three predicted outer membrane proteins here (wBm0100, wBm0152 and wBm0432)
Although interactions between wsp and host extracts have previously been published [1, 2], the authors go on to try and find out what specific host targets wBm0432 and wBm0152 might be binding (they drop wBm0100 for the rest of the study and keep wBm0152). So the next step involves in vivo metabolic labeling. They describe it here: “To identify the possible B. malayi interacting partners of wBm0152 and wBm0432 in vivo, we adapted a method used routinely for protein-protein interaction studies in mammalian cells–in vitro metabolic labeling with L-Photo-Leucine and L-Photo-Methionine amino acids, followed by photo-activated in vivo cross linking, and immune-purification of protein complexes for analysis” This sounds like a protein cross-linking assay – similar to the chemical cross-linking assays of old. Their interpretations are likely affected by the same caveats as chemical cross-linking: a degree of false positives due to non-specific interactions (although the original paper from the Max-Plank that described the technique claimed a higher sensitivity due to the distance between proteins). They use this technique, combined with MS to identify the host interaction partners. The proteins they found to be interacting with their Wolbachia wsp and PAL are also some also very abundant proteins in a cell: actin (which comprises something like 10% of total protein content ) and glycolytic pathway core enzymes. In order to figure out if their cross-linking assay produced specific interactions, they go on to perform an overlay assay.
The overlay (or far-western) is a really cool assay for detecting direct interaction between proteins . It relies on the ability to denature and re-nature the proteins to the extent that you could actually replicate the conditions for them to interact in vitro. Proteins are effectively run out on a gel, blotted in a western, then incubated with the interaction partner (which is “overlayed” onto the membrane in appropriate buffers). The detergent and salt conditions in the buffers can affect the extent of interaction as can the effective “re-naturing”. Here are the results of their far western:
Interactions with glycolytic enzymes and wBm0432 (a wsp paralog)
Ok - so you might think, this is cool, you get some interaction between wBm0432 and these enzymes. One concern I have is, how do we know that wBm0432 wouldn’t interact with many proteins under these conditions? I’d have liked to have seen a non-interactor (a host protein that wasn’t picked up in their screen) as a negative control. Alternatively, how about using a conventional immuno-precipitation and pulling down the aldolase using wBm0432? Moving on, the authors use an overlay, again, in figure 4A. This looks like a figure put in post-review. Likely, that “damned third reviewer!” suggested that they put in a non-interacting wBm protein as a control.
Figure 4A is essentially identical to the assay in Figure 3 (which is truly a wBm0432 overlay assay, not an Aldolase-overlay), except they include a control here (Figure 4A – lane 4) - a Wolbachia periplasmic protein – part of the ABC-type phosphate transport system. The result is that not every wBm protein interacts with aldolase under these conditions. I'm still left wondering: Why didn’t they include the reciprocal overlay, since they seem to have been successful in purifying large quantities of the protein? I think these results are promising, regardless, and are the best part of this paper.
Interactions with actin and wBm0152 (a PAL-like protein)
The authors at this point go on to test the interactions picked up for the other OMP, wBm0152. This protein, if you remember, interacted with actin and with tubulin in their metabolic cross-linking assay. Key to the interpretation of their results is Figure 4B. They have put the following proteins on the blot: actin (~50kD) – lane 1 and wBm152 – lane 2-4 (17kD). They then incubated with binding buffer (lanes 1,2 and 4) and actin (lane 3). Now, without looking at their results, let’s consider what we should expect to see on the membrane (I include a mock Coomassie stained gel for fun, although this was not in the publication):
Instead, what do we find here. We see a significant band, at the molecular weight expected of actin, in the lane that should only have wBm0152 (lane 3). Although they see a small amount of actin binding to the ~18kD lane, no negative control is included (for example, a vector only control prep from their E. coli)One concern is that from lane 4, we see that their wBm0152 prep is not clean (although, importantly, a regular Coomassie gel is not included, just the western). One hypothesis promoted in the article is that these high molecular weight bands are polymers of the wBmPAL and that the tetramer binds actin the best. Unfortunately, no evidence for this polymerization is provided. What about PALs in general, do we have any evidence that wBm0152, or other PALs might act as a polymer? Not that I could find. These proteins are thought to form a complex between the peptidoglycan and the Tol proteins , but in no study has there been evidence shown for a polymerization, although I'd love for the authors to correct me, if I have missed something here. It would indeed be cool to find Wolbachia proteins that interact with host cytoskeletal elements.
Actin is extremely sticky – it interacts, in these kinds of assays, with a large array of proteins, depending on the conditions (For a reference, we have seen actin bind to GFP in this kind of overlay assay). The best way to show a direct interaction with actin is to perform a sedimentation assay. In that kind of assay, you spin the heck out of your protein alone, actin alone, and both together and simply run a coomassie gel on the resulting fractions. If your protein binds to actin, it will be found in the actin-containing pellet. Interesting also is the fact that actin monomers are used in this assay (as their experimental methods don't include a polymerization step or the correct polymerization buffer for actin). Do we expect interaction with actin monomers to be of import? Perhaps, and this would maybe alter their final model (where they suggest interactions with actin filaments).
P.S. what about the study by Mark Taylor and others on wBmPAL and effects on the immune response ? Although the authors do mention this study in their paper, it's hard to incorporate into their model (at the end of the paper).
An update to this post. Lipid II synthesis is required by Wolbachia and the Wolbachia wPal protein localization is perturbed when synthesis of lipid II is inhibited (by fosfomycin), suggesting two interesting things: 1) although peptidoglycan has yet to be observed in Wolbachia, lipid II is required and 2) wPal is likely involved in Wolbachia cell division (See Vollmer et al)
1. Bennuru, S., et al., Brugia malayi excreted/secreted proteins at the host/parasite interface: stage- and gender-specific proteomic profiling. PLoS Negl Trop Dis, 2009. 3(4): p. e410.2. Melnikow, E., et al., Interaction of a Wolbachia WSP-like protein with a nuclear-encoded protein of Brugia malayi. Int J Parasitol, 2011. 41(10): p. 1053-61.3. Lodish, H., A. Berk, and Z. SL, Molecular Cell Biology, 4th Ed2000, New York, NY: W.H. Freeman.4. Wu, Y., Q. Li, and X.Z. Chen, Detecting protein-protein interactions by Far western blotting. Nat Protoc, 2007. 2(12): p. 3278-84.5. Godlewska, R., et al., Peptidoglycan-associated lipoprotein (Pal) of Gram-negative bacteria: function, structure, role in pathogenesis and potential application in immunoprophylaxis. FEMS Microbiol Lett, 2009. 298(1): p. 1-11.6. Turner, J.D., et al., Wolbachia lipoprotein stimulates innate and adaptive immunity through Toll-like receptors 2 and 6 to induce disease manifestations of filariasis. J Biol Chem, 2009. 284(33): p. 22364-78.