Satisfacción Programas de Movilidad

Sorting mediated by tetraspanins and TEM are a telling example of the intrinsic relationship between protein sequence, lipid modifications and membrane organization. In the case of CD63, its four transmembrane domains direct insertion into membranes, its YXXΦ motif guides intracellular trafficking, and palmitoylation at cysteine residues determines attachment to membranes of specific characteristics in a cholesterol- dependent manner (Charrin et al., 2003; Israels and McMillan-Ward, 2010; Pols and Klumperman, 2009; Yang et al., 2002). These features give rise to CD63 sorting to specific membrane microdomains, which are also the destination for RhoB chimeras, i.e. ILV of MVB and the plasma membrane, to a varying extent. In fact, co-localization studies in resting cells show almost total overlap between wild-type CD63 constructs and CINCCKVL chimeras, and to a lesser degree with full-length RhoB constructs. As depicted in Figure 51, it could be hypothesized that GFP-8 is intimately connected to TEM through its palmitate residues by a similar cholesterol-dependent mechanism that keeps CD proteins together (Charrin et al., 2003). The lack of further sorting determinants in GFP-8 would render the protein quite dependent on TEM dynamics. Therefore, disruption of CD63 intracellular trafficking such as that which could take place upon overexpression of CD63 constructs, particularly the Y235A mutant, would hinder binding to AP-3 complexes, which could alter CD63 sorting to TEM at MVB and therefore result in a concomitant misplacement of GFP-8. AP-3 has been shown to contribute to the regulated secretory pathway, as AP-3 depletion reduces regulated secretion of secretogranin II, though it elicits an increase in its constitutive secretion as a compensatory mechanism (Asensio et al., 2010). Therefore, a putative indirect association of GFP-8 to AP-3 via CD63 could be abolished in a scenario of altered CD63 sorting such as cells overexpressing CD63 constructs, sending CINCCKVL proteins

towards constitutive exocytosis. In contrast, additional sorting motifs, interactions with effectors, or an active role in endosomal dynamics could be responsible for full-length RhoB localization at endolysosomes without partaking in secretion mechanisms, regardless of changes in CD63 sorting. In any case, these results set forth the possibility that CINCCKVL chimeras could also appear in the extracellular space, which is congruent with their localization at ILV of MVB containing CD63 and possibly destined for exocytosis.

Biogenesis of ILV destined towards exocytosis, which contain CD63, can occur independently of the ESCRT machinery; instead, the curvature-inducing lipid, ceramide, mediates inward budding from the MVB limiting membrane (Trajkovic et al., 2008). Follow-up studies to determine whether CINCCKVL proteins can be delivered to the extracellular space were performed by exogenous treatment with C6 ceramide. These studies revealed that this lipid elicits an “emptying out” of cells co-transfected with GFP-8 and CD63, with a concomitant appearance of fluorescent extracellular material and reduction of GFP-8/mCherry-CD63 co-localization. On the other hand, GFP-RhoB was retained at several subcortical sites below the plasma membrane and did not appear in secreted membranous structures. Full-length RhoB could therefore contain a motif(s) upstream from the CINCCKVL sequence acting as a retention signal to bar its exit from the cell. Indeed, very few instances of human extracellular vesicle- associated RhoB have been described, including one report in exosomes of colorectal cancer cells, another report describing RhoB in exosomes from urine, and three studies in which microvesicles were found to contain RhoB (www.microvesicles.org). Furthermore, under the conditions employed in preliminary assays for isolating an extracellular vesicle fraction by sequential ultracentrifugation, we were unable to detect RhoB (unpublished observations).

It is worth noting that a study using numerous chimeric proteins postulates that cytosolic proteins with a tendency to oligomerize, when attached to a lipid anchor, become secreted into the extracellular medium. Indeed, a GFP-tagged, oligomeric, cytosolic protein with a RhoB-related isoprenylation and mono-palmitoylation sequence attached (CCKVL) is detectable in an exosome/microvesicle fraction, albeit to a much lesser degree than the positive controls containing either an N-myristoylation tag or a PIP-binding domain (Shen et al., 2011). However, throughout this work we

have used different fluorescent proteins besides GFP, some of which are theoretically expressed as monomers and therefore would not necessarily follow this oligomerization trend per se, e.g. mCherry, which stands for monomeric Cherry (Shaner et al., 2004). Interestingly, the Hrs domain 2xFYVE is practically incapable of eliciting a secretory phenotype in the study mentioned above. It could be hypothesized that the potential RhoB-Hrs interaction hinted at by Hrs KD cells guides RhoB constructs into ILV destined towards degradation at the lysosome, hindering its entry into exocytic ILV into which GFP-8 is sorted and retaining GFP-RhoB within the cell, as described above. These distinct phenotypes only appear upon altering cellular homeostasis, such as ESCRT depletion or C6 ceramide treatment, but not under resting conditions. Whether these distinct ILV subpopulations exist within the same MVB will be the subject of further studies, though co-transfection of CINCCKVL and RhoB constructs has proven highly toxic, perhaps due to lipid precursor depletion (results not shown).

Considering that lipidated CINCCKVL constructs are fused directly to GFP and have no other RhoB-derived sequences, the differences encountered between these proteins and full-length RhoB constructs point to other sorting, signaling, or interaction sequences upstream of the lipidation sites. A close look at the RhoB sequence does indeed bring up interesting sequences and residues as hypothetical recognition motifs, as shown in Figure 52.

Figure 52. RhoB sequence and possible sorting motifs upstream of its lipidation region. The full-length human RhoB sequence is shown. The lipidation motif, CINCCKVL is highlighted in blue, while possible AP-interaction motifs are boxed in red or black. The YDYLE sequence fits with a putative motif recognized by AP-4, namely “YX[FYL][FL]E”. The second is a putative YXXΦ motif that is recognized by several AP complexes. The “KRXX” sequence is a hypothetical AP-3 binding motif. Arrows mark a possibly ubiquitinated lysine residue and a serine that becomes phosphorylated. See text for details.

Indeed, it has been reported that serine 185 of RhoB, close to its palmitoylation cysteines (see Figure 52), is phosphorylated by casein kinase-1, which hampers stress fiber formation and EGF receptor stabilization (Tillement et al., 2008). As discussed above, a few amino acids upstream, there is a lysine residue included in the GFP-22 construct that could be amenable to ubiquitination. RhoB also possesses a sequence matching a recently described, non-canonical AP-3 interaction motif of interferon- induced transmembrane (IFITM) protein 1, i.e. the dibasic patch “KRXX”, as shown in Figure 52. Interestingly, two possible YXXΦ motifs appear in regions upstream of the CAAX box that are unique to RhoB. As seen in Figure 52, that which is closest to the C- terminus, “YGSQ”, follows a rather canonical pattern and could hence be recognized by several AP complexes (Bonifacino and Traub, 2003). Furthermore, this sequence includes the phosphorylatable serine and its recognition by AP complexes could therefore depend on its phosphorylation state. The other YXXΦ motif, “YDYLE”, appears further upstream (Figure 52), beginning with Tyr154, and is a striking match to the motif that mediates AP-4 binding for trans-Golgi to endosome sorting of the amyloid precursor protein, namely YX[FYL][FL]E (Burgos et al., 2010; Park and Guo, 2014). RhoB interaction with AP-4 would hypothetically be consistent with exit from the Golgi after palmitoylation, en route towards its endosomal localization.

Preliminary localization studies were carried out with mutants of tyrosines belonging to these hypothetical YXXΦ motifs, namely, GFP-RhoB-Y154A and GFP- RhoB-Y183A. However, there were no apparent changes in subcellular localization in basal conditions compared to wild-type GFP-RhoB by live confocal microscopy observation. Further studies are needed to address the possibility that response to stimuli such as EGF or alteration of endocytic components such as those used throughout this work would yield discrete phenotypes, or to assess the behavior of these mutants at different time points to detect whether their trafficking is delayed or expedited due to hypothetical disruption with AP complexes.

3. Small GTPases in pathological scenarios

3.1 RhoB constructs appear at endolysosomes in cell models with impaired