6 Discussion
6.6 Roles of different PrP domains in PrP localization and function
Glycosylation and GPI-anchorage
Determining the physiological significance of PrP requires thorough understanding of how its structural elements contribute to its activity and subcellular distribution. Part of my doctoral work addressed this subject by using specific functional readouts in MCF-7 cells and zebrafish embryos. In several cell types, including human cultured enterocytes, HeLa, and neuroblastoma N2a cells, surface expression of PrP is particularly prominent at contact sites and basolateral membranes (Málaga-Trillo et al, 2009; Morel et al, 2004; Sarnataro et al, 2002). This type of accumulation has been previously observed by our group in N2a cells not only for mouse, but also for zebrafish, frog and chicken PrPs (Málaga-Trillo et al, 2009 and unpublished data). Our current analysis in MCF-7 cells and zebrafish embryos revealed that the most dramatic defects in the localization of PrP are caused by lack of the GPI anchor or mutation of the N-glycosylation sites within the globular domain, since the corresponding mutants altogether fail to reach the plasma membrane. This result agrees with previous studies showing that targeted sorting of PrP is determined by a molecular signal encoded within the GPI anchor, and further modulated by N-linked glycans (Chesebro et al, 2005a; Puig et al, 2011). It also correlates with the strongly reduced ability of these constructs to influence zebrafish gastrulation. Being a secreted molecule, it is evident that GPI-anchorless PrP cannot be stabilized at the plasma membrane or accumulate at cell-cell contacts (Chesebro et al, 2005b). In contrast, the reasons behind the impaired cell-surface localization of PrP glycosylation mutants are more complex and difficult to interpret. For
instance, the fact that our PrP glycosylation mutants exhibit defective localization in MCF-7 and zebrafish embryos is consistent with related studies in other mammalian cells (Cancellotti et al, 2005; Korth et al, 2000; Lehmann & Harris, 1997; Puig et al, 2011; Rogers et al, 1990), altogether leading to the straightforward conclusion that glycosylation is essential for the delivery of PrP to the plasma membrane. However, we found that the same glycosylation mutants are correctly localized at the plasma membrane of Drosophila S2 cells (Solis et al, 2013). This is in line with a previous report showing that WT hamster PrP is correctly targeted to the surface of S2 cells despite being incompletely glycosylated (Raeber et al, 1995). Moreover, treatment of CHO and human neuroblastoma cells with tunicamycin efficiently blocks PrP glycosylation but does not influence its trafficking (Lehmann & Harris, 1997; Petersen et al, 1996). Together with more detailed mutational analyses (Neuendorf et al, 2004), these studies strongly suggest that the specific mutations introduced into the consensus sequence Asn-X-Thr -and not the lack of glycosylation per se- result in deficient PrP trafficking in vertebrate cells. The reason for this is unclear, as is the normal trafficking of these mutants in Drosophila cells. Because N-linked glycans are important to ensure proper folding, stability and quality control of glycoproteins in the ER (Vagin et al, 2009), these observations may be explained by differences in biosynthetic processing and folding of proteins between vertebrate and invertebrate cells.
Hydrophobic domain
Unlike our data on the GPI anchor and the glycosylation sites within the globular domain, our experiments with hydrophobic domain mutants (PrP ΔHD) indicate that this central stretch does not influence PrP’s subcellular localization. Zebrafish and mouse PrPs lacking the hydrophobic domain localized -similar to WT PrPs- at cell contacts in MCF-7 cells, zebrafish embryos (this dissertation) and S2 cells (Solis et al, 2013). The WT-like localization pattern of the ΔHD PrP mutants (Δ112-126 for mouse PrP) is thus comparable to the one observed for the mouse PrP ΔCR mutant (Δ105-125) in zebrafish embryos and other cell types (Christensen & Harris, 2009). Another previous report suggested a role for residues 113-133 in the basolateral sorting of mouse PrP in MDCK cells (Uelhoff et al, 2005), in apparent conflict with our data. However, because the deletion used in that study extends C-terminally beyond ours, it is possible that residues 127-133 are responsible for the effect missing in our constructs. Importantly, the WT-like distribution of the PrP ΔHD mutant is in line with our finding that the hydrophobic domain does not affect key functional properties of PrP such as its ability to influence zebrafish gastrulation, form contacts sites and trigger intracellular signals in S2 cells (Solis et al, 2013).
Repetitive and globular domains
Unlike the hydrophobic region, we showed that the repetitive and globular domains act as
important determinants of PrP localization at cell contacts. Key insights on the role of the repetitive domain were provided by zebrafish PrP-1, which –unlike mouse PrP or zebrafish PrP-2- exhibits a naturally patched/dotted distribution along contact sites of MCF-7 and zebrafish embryonic cells. In both these systems, deletion of the repetitive domain induced the loss of PrP-1’s discontinuous pattern and its homogeneous presence along the entire cell contact. This suggests that this region either promotes the protein’s local accumulation at discrete locations or facilitates its exclusion from other, complementary subregions of the cell contact. The former scenario is in line with earlier studies ascribing self-aggregation properties to the repetitive region (Parham et al, 2001; Tank et al, 2007). Interestingly, although deleting the repetitive domain did not affect the per se sorting/transport of PrP to contact sites in MCF-7 and zebrafish embryonic cells, it did impair its accumulation at S2 cell contacts (Solis et al, 2013). This likely reflects the differential regulation of cell contact formation in these experimental models: while formation of S2 cell contacts requires the establishment of PrP trans-interactions, mediated -at least partly by the repetitive domain- MCF-7 and embryonic cell contacts are independently maintained by E-cadherin homophilic interactions.
Contrary to the repetitive region, deletion of the globular domain in all three PrPs (mouse PrP, zebrafish PrP-1 and -2) did not prevent them from reaching the plasma membrane but produced instead a punctate, discontinuous distribution of the proteins at cell contacts of MCF-7 cells and the embryo. This finding suggests that the globular domain promotes the continuous localization of PrP at contact sites, thereby counteracting the clustering effect of the repetitive region. Conceivably, this may be achieved via the stabilization of PrP homophilic trans-interactions along larger regions of the contact site. The fact that WT PrP-1 normally localizes in punctae/patches would suggest that its repetitive region has a stronger clustering activity than that of mouse PrP or PrP-2. In fact, PrP-1 contains larger and more complex repeats than mouse PrP or PrP-2, owing to multiple expansion cycles of this domain during evolution (Cotto et al, 2005; Rivera-Milla et al, 2006). Thus, its globular domain would not be sufficient to ensure a continuous localization pattern. It remains to be established whether the local clusters of PrP-1 at cell contacts define signaling subregions of the plasma membrane, or whether they result from the specific recruitment of PrP-1 to preformed specialized sites. Interestingly, a similar localization pattern has been described for zebrafish Frizzled 7 (Fz7), and shown to modulate the persistence of cell contacts in the gastrula (Witzel et al, 2006). In that study, non-canonical Wnt11 was found to induce the local accumulation of its receptor Fz7 at ‘‘adhesive subdomains’’ within contact sites, thus modulating the endocytosis of the co-localized, atypical cadherin Flamingo. Similar cell contact subdomains were reported in HeLa cells and Xenopus animal caps, where canonical Wnt induces the local aggregation of LRP6-signalosomes at cell contacts to stabilize
β-catenin (Bilic et al, 2007). Hence, it would be interesting to examine whether PrP-2 and mouse PrP -like PrP-1- reside in or induce the formation of such distinct regions within cell contacts, which we may not have been discernible in this study due to the limitations of current microscopy techniques.
Aside from their clear effects on PrP localization, the repetitive and globular domains contribute significantly to PrP function during gastrulation, since the corresponding mutant constructs display a strongly reduced ability to cause an OE phenotype or rescue PrP-1 morphants. Because the presence of copper-binding histidines is not conserved in the repetitive regions of zebrafish PrPs, it is unlikely that these mediate PrP function by inducing copper-dependent endocytosis (Pauly & Harris, 1998). Instead, our results indicate that the repetitive and globular domains support PrP function either by determining its exact position in subregions of the plasma membrane -via opposing effects on local clustering- or by mediating interactions of PrP with itself or other proteins on the surface of neighboring cells.
The importance of the repetitive domain for PrP homophilic trans interactions is underscored by the lack of contact formation in S2 cells expressing the PrP ΔRep mutants (Solis et al, 2013). Accordingly, both the repetitive and globular regions have been reported to mediate PrP/PrP interactions in a yeast two-hybrid system (Hundt et al, 2003).
Physiological vs. pathological functions of PrP domains
Taken together, the present work identifies the GPI-anchor, N-glycosylation and the repetitive and globular domains as the elements determining PrP localization and function.
The lack of an apparent function of the hydrophobic domain during zebrafish gastrulation may seem contradictory to the important neuroprotective role attributed to it by other studies (Biasini et al, 2013; Li et al, 2007). This discrepancy may reflect variations in the molecular set-up of different experimental models and will be discussed in connection to our analysis of the PrP ΔCR mutant in the following section of the Discussion. On the other hand, the dramatic functional impact of mutations/deletions affecting the GPI-anchor or the N-glycosylation sites corresponds to the important roles of these elements in prion pathogenesis. Concretely, the mutation of N-glycosylation sites was shown to confer PrP with prion-like properties (Lehmann & Harris, 1997), whereas mice expressing GPI-anchorless PrP replicated prions but did not become ill (Chesebro et al, 2005b). Our most intriguing finding is perhaps the crucial role of the globular domain in zebrafish gastrulation/cell adhesion, since no functional properties had been attributed to this region of PrP until now, aside from its pathogenic role as a template for PrPSc replication (Moroncini et al, 2004; Morrissey & Shakhnovich, 1999; Norstrom & Mastrianni, 2006; Solforosi et al, 2007). Interestingly, the repetitive domain also contributed largely to PrP’s ability to control embryonic cell adhesion. Consistently, the important role of this domain is inferred by its
ability to trigger neuronal disease when expressed in expanded versions in PG14 and other PrP mutants (Bizat et al, 2010; Chiesa et al, 1998).