Proks have dynamin like molecules!

When I was a grad student, eukaryotes had all the neatest toys ... actin, microtubules, kinesins, dynein, myosin, dynamin, SNAREs ...

OK that's not totally true - bacteria had their version of tubulin (the constituent of microtubules), and it's called FtsZ. Then others found that bacteria had a version of actin, the most known is called MreB. The latest is that prokaryotes have dynamin. (Click here for previous dynamin entries.)

From the paper:

Given the presence of large GTPases with predicted dynamin-like domain organization in many members of the Eubacteria such as E. coli and Bacillus subtilis, it is likely that bacterial dynamins, or BDLPs as we term this class, are not restricted to cyanobacteria. Such observation, combined with our results, suggests a bacterial ancestry for the dynamin superfamily.

And yes, it forms radial tubes in vitro.

And yes, it also it tubulates membranes in vitro.

And yes, they have a crystal structure.

And yes, it's totally cool.

Harry H. Low and Jan Löwe
A bacterial dynamin-like protein
Nature advance online publication 22 November 2006 doi:10.1038/nature05312

PS: What's next? Will we find that proks have membrane trafficking?


Eating Lipids to Fuse Mitos

A couple of weeks back I wrote about dynamins and mitochondrial fusion. Well the latest piece of the puzzle came in ... I just saw a paper in the latest Nature Cell Biology on this very topic. Apparently a mitochondrial version of phospholipase D (MitoPLD) may act downstream of the dynamin like molecule Mfn1 to promote fusion of the outer mitochondrial membrane.

Now remember mitos have two membranes. If two mitos want to get together they must fuse the outer membranes. This requires a dynamin protein (Mfn1 in mammals). After this fusion the two inner mitochondrial membranes can come together and fuse (this second fusion requires yet another dynamin like molecule). Also remember that some dynamins can form spirals and/or rings in vitro. This is weird as dynamins are thought to form spirals the can pinch and thus break off membranes (just like spiral collar around your neck would decapitate you by squeezing your neck and pushing your head away from your torso).

Back to the paper ...

Most of the assays in the paper are at the cellular level (looking for mito aggregates and/or fragmentation), although the authors do confirm that MitoPLD converts cardiolipin to phophatydic acid (PA). Also it looks like Mfn1 first tethers two mitos together and then mitoPLD performs some key step that activates membrane fusion. Beyond this simple explanation things are still murky.


How can mitoPLD stimulate this activity? From the paper:

The requirement for production of phosphatidic acid is reminiscent of the requirement for classic PLD-dependent generation of phosphatidic acid in SNARE-mediated fusion of secretory vesicles with the plasma membrane during many types of regulated exocytosis3, 4 and during sporulation in yeast11. Taken together, these findings reveal a common requirement for a specific manipulation of the lipid environment despite the lack of conservation of the associated protein machinery and mechanism of action. The role of phosphatidic acid remains undetermined -- it may facilitate fusion by generating negative curvature in the opposing bilayers, by recruiting or activating other key proteins or enzymes, or by being further converted to other fusogenic lipids (such as diacylglycerol). Interestingly, phosphatidic acid has been shown to recruit Spo20, the yeast homologue of the mammalian t-SNARE 25, to sites of vesicle fusion through direct interaction28, and to accelerate the rate of mammalian SNARE-driven vesicle fusion in vitro29.

So PA may just promote fusion and may act in several fusion events in cells. Or here is a crazy idea, maybe mitoPLD dissolves some mito membrane (by digesting cardiolipin) and acts as a hole puncher on the mitochondrial outer membrane. This hole puncher activity might be locally contained within Mfn1 rings (see the pic above and the dynamin entry). After the holes are punched, the Mfn1 rings could serve as connector between the two mitos fixtures thus serving as a temporary channel between mitos. The outer membranes of adjacent mitos could then mix ... and presto fusion is complete!

Now of course there are lots of problems. Mfn1 is not known to form rings. There isn't that much cardiolipin in mito membranes and this would be a very messy way of fusing the outer mito membranes ... in any case I'm sure we're missing many components that will shed new light to this process.

(Also there is a discussion as to how this process of mito fusion may be connected to apoptosis [i.e. regulated cell death]. Didn't have time to reread up the background, but I'd thought I'd mention it.)


Seok-Yong Choi, Ping Huang, Gary M. Jenkins, David C. Chan, Juergen Schiller & Michael A. Frohman
A common lipid links Mfn-mediated mitochondrial fusion and SNARE-regulated exocytosis
Nature Cell Biology (06) 8:1255-1262


How Doa10p gets into the nucleus, or another freaky experiment done in yeast

I heard about this paper (Deng and Hochstrasser. Nature (06) 443:827-831) and took a look at it over the weekend. Wow! There are lots of goodies in there. And it showcases how manipulable yeast are. (As you can tell I am really jealous of researchers who use yeast as a model system.)

The premise of the paper is not bad either.

There had been some rumours that proteins could get degraded within the nucleus through the ubiquitin/proteosome pathway. Now to some this idea was heretical but this new paper gives some mechanistic info into how this process occurs.

Doa10p is an E3 ligase, that is it is the enzyme that attaches ubiquitin to the protein destined to be degraded. It's also a membrane bound ubiquitin ligase that is involved in the extraction of endoplasmic reticulum (ER) proteins that have a misfolded cytoplasmic domain (i.e. ERAD-C, see this note on ERAD and this note on the 3 types of ERAD). Previously Doa10p had also been implicated in the degradation of mat-alpha2 transcription factor, a nuclear protein involved in turning on genes in response to mating factor. So what's happening?

Doa10p had been localized to the peripheral ER but it was unclear if it diffused into the inner-nuclear membrane. Remember that the ER extends over the nucleus and is continuous with the outer-nuclear membrane (ONM) and inner-nuclear membrane (INM), however to get to the INM proteins have to pass through the nuclear pore complex (NPC). To degrade nuclear proteins, a part of Doa10p must thus pass through the NPC. To confirm this, Deng looked in cells that over express Nup53 and have an over abundance of INM (in these cells the INM looks wrinkled while the ONM is normal). Indeed Doa10p was enriched in the nucleus of these cells indicating that Doa10 does partition into the INM. How is Doa10p crossing the NPC to get to the INM? Well if you remember I wrote an entry on how the Blobel lab discovered that some INM proteins use a nuclear localization sequence (NLS) to cross the nuclear pore, so is this true for Doa10p? At first approximation it doesn't seem like Doa10p has an NLS. Now this may not be a problem, the nuclear pore complex only filters particles that are larger than ~25kDa, anything smaller can freely diffuse across. The authors first speculate that Doa10's cytoplasmic domain may be small enough to passively diffuse in, but if they make Doa10's cytoplamsic domain larger, it still can cross in. So what does it take to get in? Just like the Blobel paper, they screen cells that have mutant forms of each NPC component (called Nups) to figure out how the pore regulates the import of Doa10. Incredibly they find that Doa10 requires different Nups than what had been shown for other INM proteins! This would indicate that membrane proteins travel across the NPC via different mechanisms.

Then they perform a really crazy experiment. The authors want to prove that mat-alpha2 is indeed degraded by nuclear Doa10p. The problem is forcing Doa10p to remain out of the nucleus. To accomplish this feat the authors fuse Doa10p to coronin, an actin binding protein. The result is that Doa10p is retained in the cortical ER where it binds to actin filaments and thus sequestered away from the nucleus. Not surprisingly they find that these cells don't effectively degrade mat-alpha2. And then if cells are treated with actin depolymerizing drug (latrunculin), Doa10p is released from the peripheral ER, enters the nucleus and degrades Mat-alpha2.

Holly crap! Talk about showcasing the manipulability of yeast ... and also the idea of combining genetic manipulation with pharmacology.

So there you have it - the newest twist on how proteins reach the inner-nuclear membrane.

Deng, M. and Hochstrasser M.
Spatially regulated ubiquitin ligation by an ER/nuclear membrane ligase.
Nature (06)443:827-831