A Profusion of Sorting Pathways

Retention in the Golgi

Little is known about how proteins are retained in the Golgi complex. For the budding yeast protein Mnn1p, either the transmembrane domain or the lumenal domain is sufficient for retention. On page , Reynolds et al. report that the MAP kinase Hog1p is needed for retention mediated by the lumenal domain. Hog1p is part of a pathway for sensing hyperosmotic stress, and Reynolds et al. find that other components of this pathway are necessary for localization of the Mnn1p lumenal domain.

Without Hog1p, the lumenal domain is secreted, whereas the full-length protein slips from earlier Golgi compartments to the trans-Golgi network. Whether this affects the α-1,3-mannosyltransferase function of Mnn1p is not known. It is certainly possible, since the lumen of later Golgi compartments may differ in acidity, divalent cation concentrations, or the availability of sugar nucleotide donors.

A change in Mnn1p activity could alter the cell wall, which contains a large number of mannoproteins. Reynolds et al. suggest that Hog1p's effects on Mnn1p localization may be one way that Hog1p loosens the cell wall to allow expansion and growth. The Hog1p pathway is known to induce the expression of an exoglucanase, which digests various components of the cell wall, and to repress the protein kinase C pathway, which turns on cell wall biosynthesis in budding yeast. Clearly, Hog1p is not the retention protein that Reynolds et al. were originally seeking, but its discovery in this context raises intriguing questions about how Golgi protein localization might be dynamically regulated.

From Early Endosomes to Golgi

Evidence of a trafficking pathway from early endosomes to the Golgi is presented by Mallard et al. on page 973. This is distinct from the Golgi-bound pathway taken by the cation-independent mannose 6-phosphate receptor (CI-MPR). After delivering lysosome-bound proteins to the late endosome, the CI-MPR returns directly from late endosomes to the Golgi.

Mallard et al. use the movements of Shiga toxin B-subunit to define the pathway. The B-subunit is internalized at 19.5°C. It remains in early endosomes, already partitioning away from the bulk fluid phase in tubular structures that often contain the AP1 coat protein γ-adaptin. After a shift to 37°C, the B-subunit moves directly to the Golgi with a half-time of 19 min. The authors never see any B-subunit in late endosomes or lysosomes.

Candidates for endogenous proteins that may use this pathway include TGN38, a protein whose function is not known, and the protease furin, which operates in several compartments from the Golgi to the plasma membrane.

Two Ways to Make a Synaptic Vesicle

Synaptic vesicles (SVs) must be rapidly recreated after they have released their contents by fusing with the plasma membrane. In vivo evidence suggests that dynamin, clathrin, and the AP2 adaptor complex are involved in forming synaptic vesicles directly from the plasma membrane. But an in vitro model of SV formation identified endosomes as the source of SVs; in this case, SV formation requires ARF1 and the AP3 adaptor but not dynamin or AP2.

Shi et al. solve this contradiction on page 947. Although they set out to observe an earlier step in the in vitro pathway (the generation of endosomes [the SV precursor] from plasma membranes), what they saw was formation of SVs directly from the plasma membrane, as is seen in vivo. Apparently, the method originally used to load membranes with labeled SV components favored the labeling of endosomes. Shi et al. now observe that SVs form from both endosomes and plasma membrane with equal efficiency. The plasma membrane pathway requires dynamin, clathrin, and AP2, but the AP3/endosome pathway does not appear to require clathrin. There have been conflicting reports on whether clathrin is involved in AP3-based trafficking.

Several questions remain. For the plasma membrane pathway to work, there must be a mechanism for sorting proteins into different membrane domains before vesicle formation. And, says senior author Regis Kelly, “what we don't have is what AP3 does in real neurons.” Mice mutant for an AP3 component are viable but have a variety of neurological defects. So the plasma membrane pathway appears to be the major pathway in most neurons, whereas the endosome pathway may operate only in certain types of neurons, or at particular levels of synaptic activity.

Nuclear Import without Importins

On page , Jenkins et al. show that signal-mediated transport of the HIV Vpr protein into the nucleus does not require exogenous energy or any soluble proteins, including the importins. Classic nuclear localization sequences bind to importin α, which binds importin β, which binds the nuclear pore complex (NPC). The GTPase Ran is needed for this transport pathway. Nuclear transport mediated by a family of importin β–related proteins is also thought to rely on Ran function since these proteins contain Ran binding sites. (Transport is not Ran's only role: on page 1041, Nakamura et al. find that overproduction of a Ran-binding protein leads to ectopic microtubule nucleation sites.)

Vpr transport does not involve Ran or the importins— neither a dominant-negative Ran nor the importin β–binding fragment of importin α prevent Vpr transport—but the transport pathways may converge. Excess importin β blocks Vpr transport, and excess Vpr fragments block importin-mediated transport of other proteins. Thus, Vpr and importin α may be binding the same NPC target.

Vpr increases the productivity of an HIV infection, but it is not known which step is accelerated. One possible site of Vpr action is nuclear entry of the HIV preintegration complex. Although this process does not absolutely require Vpr, it may be facilitated by Vpr's direct access to NPC proteins.

Kinesin Inhibition

Kinesin in vitro stays glued to a microtubule and moves processively, in contrast to myosin moving on actin. But like myosin, most of the kinesin in the cell is soluble. Verhey et al. (page ) suggest that the kinesin light chain prevents most of the motile heavy chain from binding microtubules.

With kinesin binding only rarely to microtubules, its apparent microtubule-stimulated ATPase activity is low. Proteolysis of kinesin heavy chain increases the ATPase activity, so David Hackney suggested that the tail domain folds back to inhibit the motor domain. Hackney went on to show that at least some kinesin is found in a folded conformation, although he did not demonstrate a correlation between this conformation and inhibition.

Verhey et al. find evidence to support the folding model: a portion of the heavy chain's tail is needed for self-inhibition. However, they identify light chain as a key to the inhibition in cells. Overexpressed heavy chain decorates microtubule-like structures in cells and can be isolated bound to taxol-stabilized microtubules. Coexpression of light chain eliminates both heavy chain's in vivo localization and in vitro binding to microtubules.

Inhibition can be released in vitro by a pH shift from 7.2 to 6.8, for both overexpressed and endogenous kinesin. The pH shift does not involve a change in the amount of folded or extended forms.

Artificial acidification in vivo also induces movement, in this case of lysosomes and late endosomes to the cell periphery. A global pH change is highly unlikely to occur naturally, but leakage of hydrogen ions from the acidic interior of organelles could activate motors locally.

By William A. Wells, 1095 Market Street #516, San Francisco, CA 94103. E-mail: wells@biotext.com