This study was launched to clarify the effects of glycosylation and complex formation on the properties of PPT1. Previously, the three N-glycosylation sites of PPT1 have been shown to be utilized in non-neuronal cells and their effect on PPT1 activity has been studied by mutagenizing the glycosylation site asparagines to glutamines . Single N → Q mutations were reported to cause only minor effects on PPT1 activity, and double mutations resulted in a minimum of 20% activity. In this study, we utilized a different strategy for generating the glycosylation site mutations. Rather than changing the specific asparagines to another amino acid, we altered the glycosylation site consensus sequence N-X-S/T by mutating serines to alanines. This was done to spare the polypeptide backbone and the side chains from any major distortions [21, 30]. Interestingly, our results differed significantly from the previous analyses. The activity of all the glycosylation site mutants was greatly reduced, and only the N212 glycosylation site mutant could retain a considerable enzyme activity. Thus, our data implicates that glycosylation is important for PPT1 activity, and the most important site for it is N197, followed by N232 and N212.
It has been shown that the glycosylation of asparagine 232 has a fundamental role in the M6P-receptor binding and thus it was proposed to have an effect on the proper lysosomal transport of PPT1 in non-neuronal cells . Our studies on the intracellular transport of glycosylation site mutants did not confirm these conclusions. We investigated both intracellular and secreted PPT1 by immunoprecipitation and analyzed the localization of the enzyme by immunofluorescence. We showed that the same glycosylation mutations that affected the PPT1 enzyme activity also had the most severe impact on the transport of PPT1. When the N197 glycosylation site was mutated, PPT1 was retained in the ER, although the N232 glycosylation site – mostly involved in the M6P-receptor binding – was unaffected. When the N232 glycosylation site was mutated, PPT1 localized to the ER – but also partially to lysosomes. The same mutant was also secreted into the media. Thus, our data suggest that the role of M6P-receptor-mediated lysosomal trafficking might not be as crucial for PPT1 as it has previously been thought.
The relative inefficiency of gene therapy trials to treat INCL suggests that the trafficking of PPT1 may have novel properties. M6P-receptor-mediated trafficking has offered a substantial advance to treat lysosomal storage disorders since most soluble lysosomal enzymes can diffuse in the brain tissue. For example, in glucoproteinoses, gene therapy and bone marrow transplantation have been advantageous in animal models and even in human studies, especially in cases where the M6PR-mediated trafficking of the lysosomal enzyme has been shown to operate in neurons [23, 31–35]. However, the therapeutic studies in Ppt1-deficient mice involving AAV-mediated PPT1, have ameliorated the CNS pathology only in localized areas, suggesting that the neuronal trafficking of PPT1 has undefined features [36–38]. In this study, we could observe several new properties for neuronal PPT1, including altered modification compared to PPT1 in fibroblasts. Both of the cell types analyzed here were infected with the same recombinant PPT1-adenovirus  including the human cDNA, so it is unlikely that our finding would result from different splice variants. One could postulate that the glycosylation is actually similar in these cells, and that the differences seen in the number of the bands would result from some other modification. Further studies are needed to clarify the precise nature of these modifications, but phosphorylation and lipid modifications could be potential study targets. With the antibody internalization assay, trafficking differences could be seen both in fibroblasts and neurons compared with another lysosomal enzyme AGA, which is known to utilize the M6P-receptor. While the AGA antibody was endocytosed into perinuclear lysosomes, the endocytosed PPT1 antibody only partially localized to lysosomes, mostly retaining in small vesicles distributed throughout the cell. We conclude that an increasing amount of evidence pinpoints the existence of an alternative pathway for PPT1 sorting in addition to the M6P-receptor-mediated route.
Molecular interactions of several NCL proteins have been partially resolved. CLN5 has been shown to interact with both CLN2 and CLN3 . CLN6, the protein behind the variant late infantile NCL, was recently shown to form dimers in cross-linkage experiments . However, so far the molecular interactions of PPT1 have not been analyzed in detail. Our data suggest that PPT1 can homodimerize or oligomerize in vivo. Both GFP-PPT1 and PPT1 proteins used in the co-immunoprecipitation assay are highly glycosylated. Glycosylation usually prevents aggregation and thus the interaction is not likely to be due to a bias via aggregation of overexpressed protein. More importantly, the demonstration of the active enzyme in the large molecular weight fraction of the cell homogenate during the size-exclusion chromatography supports a true interaction between PPT1 molecules, either directly or via other molecules. PPT1 has been crystallized as a monomer from a secreted protein sample . Neither could we detect oligomerization in a secreted, monomeric sample. Thus it is possible that the intracellular oligomerization of PPT1 involves other, yet unresolved molecules. The understanding of the physiological relevance of the oligomerization for the action of PPT1 also awaits further studies. Interestingly however, we could detect that mutant PPT1 molecules show a higher degree of oligomerization. This may represent a means to regulate the activation and/or transport of the enzyme. For example, acyl protein thioesterase 1 (APT1), a functional relative for PPT1, forms a dimer which has to dissociate before the enzyme displays activity . This suggests that the dimer formation and dissociation may occur reversibly also in the case of PPT1.
PPT1 activities of mutant proteins have been carefully analyzed earlier from transfected COS-1 cell lysates, purified protein samples and patient lymphoblasts [16, 42]. While severe INCL-mutations produced inactive enzymes, the mutations causing later onset disease showed low residual activity. However, it has not been possible to draw a straight correlation between the level of residual activity and the age of onset, since the late infantile and juvenile phenotype-causing mutations both possessed activities between 6 and 30% of the wild type enzyme. Also in this study, the overexpressed PPT1 carrying the adult phenotype mutation G108R was shown to have only a low activity of 3.5%. The activities of PPT1G108R have earlier been measured in patient fibroblasts and leukocytes and they were similar to those of patients with the early onset disease . As the residual PPT1 enzyme activity of mutant proteins have not correlated with the disease phenotypes, previous studies have proposed that decreased enzyme stability plays a major role in determining the phenotype . A late onset mutation M1I displayed almost full activity (87.5% of wild type) as analyzed by overexpression in COS-1 cells. It has been proposed that the start codon ATA would be utilized in the patients' tissues, resulting in low but clinically relevant expression of the normal enzyme . Our current observations support this suggestion, since also the intracellular localization of this mutant was uniform with wild type enzyme both in HeLa cells and neurons. It can be concluded that the enzyme activity level per se cannot be used to distinguish later onset diseases from each other.
In the current study we could not define the molecular basis for the adult phenotype caused by the G108R mutation. The G108R polypeptides fully colocalized with the ER marker PDI in HeLa cells being well in line with earlier findings . Neuronal G108R was transported further to the extensions but did not colocalize with the presynaptic markers, the target for the wild type and M1I polypeptides. Evidently PPT1G108R can delay the development of the disease until adulthood without reaching the synaptic vesicles. This may implicate that PPT1 has some functional roles along the secretary pathway. Further research is needed to locate the place(s) of action for neuronal PPT1 and to clarify the molecular networks affected by PPT1 deficiency.