Schematic model of the roles of Ub and the Gga proteins in intracellular trafficking of Gap1 and CPS. Upon addition of a favored nitrogen source or an excess of amino acids, the Gap1 permease present at the plasma membrane is K63 ubiquitylated by the Rsp5/Npi1 ligase, but monoubiquitylation of the permease on at least one lysine residue already constitutes a perfectly efficient internalization signal. K63 ubiquitylation of Gap1 is required only for further sorting into the MVB pathway. The short Ub chains are possibly recognized first by the Gga proteins (via the GAT domain) and then by Vps27 (via its Ub-interacting motif domain). If only monoubiquitylated, internalized Gap1 is partially mis-sorted to the vacuolar membrane, and the rest is recycled to the plasma membrane. Good nitrogen sources also promote K63 ubiquitylation of neosynthesized Gap1 present in the Golgi by Rsp5. Under these conditions, Gap1 exits this compartment and is sorted to the late endosome. However, this sorting step does not depend on Gap1 ubiquitylation. It requires the Gga proteins (but not their GAT domain), which likely recognize cis elements exposed by Gap1. Golgi to endosome transport of CPS is also independent of its ubiquitylation. At the late endosome level, K63-Ub chain formation is required for entry into the MVB pathway of both CPS and neosynthesized Gap1, as it is for endocytosed Gap1. This model might also be valid for many other yeast cargoes, including the Fur4, Arn1, and Sit1 permeases. All of these proteins indeed undergo Ub-independent, Gga-dependent sorting from the Golgi to the late endosome, whereas their ubiquitylation is only important for subsequent entry into the MVB pathway (Bilodeau et al., 2004; Blondel et al., 2004; Kim et al., 2007; Erpapazoglou et al., 2008). In some other cases, e.g., for the Tat2 permease and the Pma1 ATPase, cargo ubiquitylation was reported to be required for exit of the TGN (Beck et al., 1999; Helliwell et al., 2001; Umebayashi and Nakano, 2003; Pizzirusso and Chang, 2004). Yet, in these studies, vacuolar delivery was examined using experimental approaches that do not allow for distinguishing between the trafficking steps from the Golgi to the endosome and then from the endosome to the vacuole. In fact, these data are also consistent with a role of Ub only at the endosomal level, where ubiquitylation is absolutely required for MVB sorting of yeast plasma membrane proteins, whether the cargo reaches the late endosome by endocytosis or directly after exiting the Golgi (Reggiori and Pelham, 2001; Blondel et al., 2004; Stimpson et al., 2006; Kim et al., 2007; Stawiecka-Mirota et al., 2007; Erpapazoglou et al., 2008). If this ubiquitylation is defective, the cargo proteins either accumulate at the vacuolar membrane or are redirected to the plasma membrane. This direct late endosome to cell surface route taken by nonubiquitylable Gap1 variants might be equivalent to the alternative secretory pathway of S. cerevisiae involving a transit step through the late endosome (Gurunathan et al., 2002; Harsay and Schekman, 2002). Moreover, rerouting of the Arn1 and Sit1 transporters from the late endosome to the plasma membrane, without passing by the Golgi, has recently been observed under particular physiological conditions (Kim et al., 2007; Erpapazoglou et al., 2008).