Novel ubiquitin dependent ER quality control of transmembrane proteins

Our studies on the palmitoylation of ATRs led us to the hypothesis that these receptors might interact with a palmitoylated partner protein (Abrami et al., 2006). We therefore investigated whether LRP6 underwent this lipid modification. We found that LRP6 is palmitoylated on juxtamembranous cysteines, soon after synthesis in the endoplasmic reticulum and that this modification is actually required for its exit from the ER (Abrami et al., 2008). Interestingly palmitoylation deficient LRP6 was monoubiquitinated and this modification of cytoplasmic cysteines was responsible for its retention in the ER, indicating that a ubiquitin dependent retention mechanism operates in the ER (Abrami et al., 2008). The failure in ER exit and the subsequent ubiquitin dependent retention is likely to be a consequence of the detection of palmitoylation deficient LRP6 by an ER quality control mechanism. Our work supports the following model: LRP6 has a very long transmembrane domain (23 residues vs. 21 in the vast majority of type I and II membrane proteins), on the other hand the ER is known to have a membrane that is thinner than that of other cellular membranes in particular the plasma membrane. Thus a hydrophobic mismatch would occur between the LRP6 transmembrane domain and the ER membrane, unless the transmembrane helix of LRP6 was tilted, which would decrease its effective length. Since LRP6 is palmitoylated on cysteines that are very close to the transmembrane domain, we proposed that palmitoylation serves to tilt the transmembrane helix. In agreement with this model we found that shortening of the transmembrane domain of palmitoylation deficient LRP6 releases the mutant LRP6 from the ER, and conversely, shortening the transmembrane domain of WT LRP6 leads to its ER retention (Abrami et al., 2008). This works thus shows the existence of a novel mechanism for regulating the configuration of single transmembrane helices

On the basis of these findings we wish to propose a novel mechanism of ER quality control, that would be somewhat the cytosolic counter part of the calnexin/calreticulin cycle (Figure 1): upon co-translationnal insertion of a protein into the ER membrane, an ER ubiquitin ligase would ubiquitinate cytoplasmic domains on juxtamembranous lysines, this would allow the protein to bind to an ER ubiquitin binding protein, and retention would give the protein time to fold in a manner similar to the calnexin bound monoglucosylated protein that undergoes folding in the lumen of the ER. This interaction would be released by deubiqunating enzymes (DUB). DUBs would also ensure that the protein does not get poly-ubiquinated and targeted for retro-translocation and targeting to the proteasome. If after removal of the ubiquitin, the protein was properly folded, it would distribute to exit sites and leave the ER. If the protein was not yet properly folded, a folding sensor in association with an E3 ligase activity would re-ubiquitinated the protein and send it through the cycle once more. The folding sensor does not necessarily have to be a protein but could involve physico-chemical properties that would lead to differential partitioning into ER membrane domains. Finally after a number of cycles of ubiquitination-deubiquitination, a timer mechanism would allow poly-ubiquitination to occur, possibly by a segregation from the DUBs, and thereby target the terminally misfolded protein to be targeted to the proteasome. Future studies are planned to test this model and identify its components