Our results indeed illustrate that the yeast homolog of ALG-2, i.e., Pef1p, also binds to Sec31p and modulates COPII coat assembly. Mammalian ALG-2 has also been shown to interact with proline-rich regions of Alix/AIP1 
, annexins VII and XI 
, and tumor susceptibility gene 101 
in a Ca2+
-dependent manner. We demonstrated that Pef1p specifically binds to the proline-rich region of Sec31p (). Thus, the mechanism of the interaction between Pef1p and Sec31p would be similar to that of other PEF proteins. Nevertheless, the Ca2+
dependency for the Sec31p interaction with Pef1p appears to be reversed compared to that previously reported for the ALG-2/Sec31A interaction. The results of our pull-down experiments clearly demonstrated the preferential affinity of Sec31p for Pef1p in the absence of Ca2+
(), while ALG-2 has been shown to bind Sec31A in a calcium-dependent manner. However, the effect on COPII also seems to be reversed: ALG-2 binding stabilizes the membrane association of Sec31A, whereas Pef1p binding prevents the membrane recruitment of Sec13/31p. Although sequence alignment of the C-terminal calcium binding domains of Pef1p shows a high conservation with that of ALG-2, Pef1p has a relatively long N-terminal extension 
, which might potentially contribute to the difference in their behavior. From these data, we postulated that the differing Ca2+
dependency could be explained by a different ER-to-Golgi trafficking route between yeasts and mammals.
In mammalian cells, COPII vesicles that bud from the ER do not directly fuse to the Golgi apparatus; instead, they are thought to tether to each other to form pre-Golgi intermediates, also termed vesicular tubular clusters (VTCs) or ER-Golgi intermediate compartment (ERGIC). The lumen of the ER and Golgi contain a high concentration of free Ca2+
due to Ca2+
-ATPase pump activity 
, and the cytoplasmic Ca2+
concentration immediately adjacent to the membranes of these organelles is reported to be relatively high, possibly because of nonspecific basal leakage of luminal Ca2+
. However, since Ca2+
-ATPases are not incorporated into COPII vesicles, pre-Golgi intermediates exhibit lower Ca2+
. With these features, ALG-2 may stabilize the Sec13/31p coat at ER exit sites via a Ca2+
-dependent interaction to promote COPII coat assembly, while simultaneously encouraging coat disassembly before vesicle fusion with the Ca2+
-poor pre-Golgi intermediates.
In contrast, COPII vesicles in yeasts are thought to tether and directly fuse with the cis
-Golgi. From this picture, if Pef1p in S. cerevisiae
had the same Ca2+
dependence as ALG-2 and behaved as it does in mammalian cells, then Pef1p would inhibit the dissociation of the outer Sec13/31p coat from COPII vesicles at the surface of cis
-Golgi membranes. However, our in vitro
binding experiments showed that Pef1p requires low calcium levels for Sec31p binding, and thus, Pef1p in yeast may not act at the membrane surface of these Ca2+
-rich organelles. Instead, the low concentration of calcium in the cytosol area could promote Pef1p binding to Sec13/31p. Consistent with this hypothesis, we found that GFP-Pef1p exclusively localizes to the cytosol (). Recent work suggests that the inner COPII coat, i.e., the Sec23/24p complex, directly participates in tethering COPII vesicles to the Golgi membranes by binding directly to a component of the tethering complex TRAPP, and therefore, it would be released only after delivery of vesicles to the Golgi compartment 
. In contrast to the inner layer coat, disassembly of the outer COPII coat Sec13/31p does not necessarily occur after vesicle tethering. Therefore, it is possible that Pef1p binding to Sec13/31p in the cytosol would tend to discourage back-recruitment of Sec13/31p that had been released from the COPII vesicles, although the timing of Sec13/31p disassembly and in vivo
triggers for uncoating still remain unknown. Given these assumptions, perhaps the failure to exhibit severe ER-to-Golgi transport defects in the presence of excess Pef1p is not surprising (see below).
We observed that overexpression of either wild-type Pef1p or the Pef1p-E180/181/248A mutant from the GAL1
promoter resulted in growth inhibition, but had no effect on ER-to-Golgi transport (). These findings suggest that excessive levels of Pef1p may influence steps other than ER-to-Golgi transport. Penta-EF-hand proteins have been shown to participate in a variety of calcium-dependent processes in vertebrates 
. Pef1p is the only PEF family protein found in the budding yeast S. cerevisiae
, and is likely to play multiple roles in the control of cellular physiology, such as polar bud growth and cell wall abscission 
. Therefore, we reasoned that the mechanism of the overexpression phenotype of Pef1p might be that the proper balance of components required for these cellular processes was changed, which limits our ability to further examine the contribution of Pef1p to intracellular transport in vivo
. Although we have shown here that the Ca2+
-free Pef1p-E180/181/248A mutant constitutively binds to the Sec13/31p complex, overexpression of this mutant did not prevent in vivo
ER-to-Golgi transport. This result is quite plausible because our light-scattering assays demonstrated that the Pef1p-E180/181/248A mutant delays, rather than blocks, the recruitment of Sec13/31p (), and this mutant is likely to act as a kinetic inhibitor in this reaction. These data are consistent with our results that, unlike ALG-2, GFP-Pef1p-E180/181/248A did not colocalize with Sec13p-mCherry at the ERES ().
Definition of the precise function of Pef1p in yeast cells will clearly require further experimentation, but Pef1p is the first coat-binding protein that negatively regulates coat assembly. The kinetic inhibition of Sec13/31p recruitment may represent a novel regulatory mechanism controlling the rate of ER-to-Golgi transport. Comprehending these refinements and additional levels of control is crucial to our understanding of the molecular basis of ER export in the context of the entire secretory pathway.