The lysosomal system comprises a specialized network of organelles crucial for the sorting, digestion, recycling and secretion of cellular components. With their content of hydrolytic enzymes, lysosomes regulate the degradation of a multitude of substrates that reach these organelles via the biosynthetic or the endocytic route. Gene defects that affect one or more of these hydrolases lead to LSDs (lysosomal storage diseases). This underscores the apparent lack of redundancy of these enzymes and the importance of the lysosomal system in cell and tissue homoeostasis. Some of the lysosomal enzymes may form multiprotein complexes, which usually work synergistically on substrates and, in this configuration, may respond more efficiently to changes in substrate load and composition. A well-characterized lysosomal multienzyme complex is the one comprising the glycosidases β-gal (β-galactosidase) and NEU1 (neuramidase-1), and of the serine carboxypeptidase PPCA (protective protein/cathepsin A). Three neurodegenerative LSDs are caused by either single or combined deficiency of these lysosomal enzymes. Sialidosis (NEU1 deficiency) and galactosialidosis (combined NEU1 and β-gal deficiency, secondary to a primary defect of PPCA) belong to the glycoprotein storage diseases, whereas GM1-gangliosidosis (β-gal deficiency) is a glycosphingolipid storage disease. Identification of novel molecular pathways that are deregulated because of loss of enzyme activity and/or accumulation of specific metabolites in various cell types has shed light on mechanisms of disease pathogenesis and may pave the way for future development of new therapies for these LSDs.
A multifunctional lysosomal enzyme complex
The lysosomal multienzyme complex specified by PPCA (protective protein/cathepsin A), NEU1 (neuramidase-1) and β-gal (β-galactosidase) is tailored to efficiently control the hydrolysis of complex glycoconjugates, starting at their non-reducing end with the cleavage of sialic acids and β-galactosyl moieties. At least another glycosidase, GALNS (N-acetylgalactosamine-6-sulfate sulfatase), is loosely associated with this complex, which may include additional hydrolases needed for the subsequent removal of sugar nucleotides, albeit this has not been experimentally proven [1–4]. Besides their canonical degradative capacity towards specific classes of glycoconjugates, it is becoming apparent that each of the enzymes may exert additional functions either within or outside the complex.
PPCA and galactosialidosis
Mammalian PPCA is a member of the serine protease family of enzymes and, in this capacity, exerts carboxypeptidase, deamidase and esterase activities on selected bioactive peptides, such as substance P, oxytocin and endothelin 1 [5,6]. The enzyme precursor is a zymogen that is activated in the endosomal/lysosomal compartment by partial proteolysis. The three-dimensional structure of the PPCA zymogen, the first determined for a lysosomal enzyme precursor, unravelled the structural mechanism of its catalytic activation . Distinct from its catalytic activity, PPCA has a protective function towards at least two other lysosomal glycosidases, β-gal and NEU1, with which it is found in complex. Early studies using subcellular fractionation, co-purification and co-immunoprecipitation methods in cultured cells and mammalian tissues have demonstrated that interaction of the PPCA zymogen with β-gal and NEU1 occurs soon after synthesis of the proteins . This early interaction assures the regulated trafficking of the two glycosidases to the lysosome as well as their intralysosomal activation and stability. More recently, the use of analytical ultracentrifugation and surface plasmon resonance techniques has enabled a better assessment of the influence of PPCA–NEU1 assembly on the overall biochemical and structural properties of NEU1 . This study suggests a new and complex mechanism in which PPCA competes with NEU1 for the same binding sites, preventing premature oligomerization and catalytic activation of the glycosidase, but allowing the formation of a heterodimeric complex suited for lysosomal targeting . The surface plasmon resonance data of the two enzymes led to the proposition of a structural model of the PPCA–NEU1 heterodimeric complex (Figure 1). These studies also define PPCA as a molecular chaperone for NEU1's lysosomal localization and activation . The determination of the three-dimensional structure of the two enzymes in complex should ultimately explain their interdependent mode of action on target substrates.
Structural model of the PPCA–NEU1 heterodimer
In humans, only mutations that interfere with the protective function of PPCA have been identified [9,10]. They result in complete loss of NEU1 activity and reduced β-gal activity, hence providing the genetic evidence of PPCA interaction with the two glycosidases. The combined β-gal/NEU1 deficiency is the biochemical hallmark of the neurodegenerative LSD (lysosomal storage disease) GS (galactosialidosis). It is still unclear, however, whether an isolated loss of cathepsin A activity could have pathological consequences, although some authors have demonstrated that an enzyme hydrolysing the C-terminus of endothelin-1 was deficient in tissues from a GS patient, and implicated PPCA as the major endothelin-1-degrading enzyme present in human tissues . GS patients are clinically heterogeneous and differ widely in severity and age of onset of the symptoms. The severe early-onset forms develop a systemic condition associated with fetal hydrops, skeletal dysplasia, visceromegaly, renal and cardiac failure, variable neurological involvement and early death . They have phenotypic manifestations similar to those found in patients with the severe form of sialidosis, a related disorder of glycoprotein metabolism that is caused by an isolated deficiency of NEU1 (see below). The loss of NEU1 activity in both diseases accounts for the storage of essentially the same type of sialylated oligosaccharides and glycopeptides in patients' tissues and body fluids, which probably contribute to disease pathogenesis. Correlation of genotype with disease severity has emerged from structure–function studies of mutant PPCA variants .
The mouse model of GS was the first generated in my laboratory and was one of the first genetically engineered animal models of an LSD . Ppca−/− mice have pathological manifestations reminiscent of patients with GS; symptoms include progressive nephropathy, hepatosplenomegaly and shortened lifespan (~12 months) . In addition, Ppca−/− mice develop severe and progressive ataxia that is associated with gradual loss of cerebellar Purkinje cells [4,11]. This phenotype could either be due to the combined deficiency of NEU1 and PPCA or the isolated deficiency of cathepsin A, although no specific neuronal targets for this enzyme that would explain the loss of Purkinje cells in the GS model have been identified to date. However, the potential physiological esterase/deamidase activity of PPCA in platelets , endothelial cells , heart  and kidney towards several bioactive peptides that regulate blood circulation and blood pressure could justify the occurrence of cardiomyopathy and arterial hypertension that are often diagnosed in patients with GS . A cardiovascular role of PPCA has been recently suggested by a study in a knockin mouse model of PPCA deficiency that carries an amino acid substitution abrogating the catalytic activity, but not the protective function, of the protein .
NEU1 and sialidosis
NEU1 belongs to the family of sialidases or neuraminidases, which are widely distributed in Nature. These enzymes have similar substrate specificity and catalyse the cleavage of ketosidically linked sialic acid residues on glycoproteins, glycolipids (gangliosides) and oligo- or poly-saccharides  The existence of four mammalian sialidases (NEU1–NEU4) that differ in their subcellular localization, substrate specificity and pH optimum underscores the importance of these enzymes in basic cellular processes, including cell proliferation, differentiation and immunity. These processes are strongly influenced by the extent of sialic acid residues on receptors and/or ligands. NEU1 is the most abundant and widely expressed mammalian sialidase, being present in nearly all vertebrate tissues and cell types. Curiously, NEU1 has a higher sequence identity with several well-characterized bacterial sialidases than with the other three mammalian sialidases, which are more similar to each other than to NEU1 [16,17].
Lysosomal NEU1, the second component of the multi-enzyme complex, is the only mammalian sialidase that is involved in two LSDs: sialidosis, caused by structural mutations at the NEU1 locus, and GS. Sialidosis patients with the early-onset form of the disease have a mucopolysaccharidosis-like phenotype with visceromegaly, dysostosis multiplex and mental retardation . Milder late-onset cases are characterized by ocular cherry-red spot and generalized myoclonus, but some of them may suffer from seizures and ataxia .
Neu1−/− mice manifest a multisystemic disease similar to severe sialidosis . One of the phenotypes that these mice share with sialidosis patients is a time-dependent enlargement of the spleen, with features consistent with splenic EMH (extramedullary haemopoiesis). While investigating the molecular bases of this phenotype, we discovered a new function of NEU1 as a negative regulator of lysosomal exocytosis . We found that, in haemopoietic cells, NEU1 controls the extent of sialylation and turnover of LAMP (lysosomal-associated membrane protein) 1. Loss of NEU1 function results in hypersialylation of LAMP1 at the lysosomal membrane, which changes the fate of these organelles, making them poised to preferentially dock at the PM (plasma membrane) and engage in lysosomal exocytosis (Figure 2). The downstream effect is excessive exocytosis of lysosomal content, particularly serine proteases, in the bone niche, which culminates in the premature degradation of VCAM-1 (vascular cell adhesion molecule 1), and loss of BM (bone marrow) retention . These findings identify sialidosis as the only known genetic disease in which lysosomal exocytosis is exacerbated . Given the multisystemic nature of the disorder and the fact that lysosomal exocytosis is a basic physiological process in many cell types, we predicted that abnormal exocytosis may underlie other phenotypic abnormalities in sialidosis.
Immunofluoresence detection of LAMP-1 and confocal microscopy of NEU1-deficient macrophages
In fact, enhanced lysosomal exocytosis in Neu1−/− mice could contribute to the profound hearing loss and muscle atrophy characteristic of the disease in sialidosis-affected children [21,22]. Progressive morphological abnormalities and vacuolization of the marginal cells of the stria vascularis in the cochlea of these mice is associated with oversialylation of LAMPs, most importantly LAMP1. The increased expression and apical localization of LAMP1 in these cells suggests exacerbation of lysosomal exocytosis into the endolymph, resulting in reduced endolymphatic potential and failure to transduce sound through sensory hair cells .
In skeletal muscle tissue, loss of NEU1 activity and consequent exacerbation of lysosomal exocytosis of fibroblasts may underlie the observed abnormal remodelling of the ECM (extracellular matrix) and excessive proliferation of connective tissue . Muscle fibres juxtaposed to the expanded perimysium and epimysium become infiltrated by fibroblasts and components of the ECM, leading to their progressive fragmentation and ultimately destruction . Together, these findings may offer the molecular basis for the muscle weakness and atrophy associated with sialidosis in children.
β-Gal and GM1-gangliosidosis
Although both genetic and biochemical studies support the strict requirement of NEU1 for PPCA, it is less clear to what extent β-gal, the third component of the multienzyme complex, relies on PPCA for its function. Although its interaction with PPCA aids the trafficking of the enzyme to the lysosomes and influences its stability, the partial deficiency of β-gal in GS patients and the apparent normal β-gal activity in the GS mouse model suggest a more loose dependence of this enzyme on PPCA. In fact, β-gal maintains catalytic activity outside the complex. The enzyme is largely responsible for the intralysosomal degradation of GM1-ganglioside, a major glycosphingolipid of neuronal membranes, and keratan sulfate [23,24]. Deficiency of β-gal in humans leads to GM1-gangliosidosis, a severe generalized neurodegenerative disorder affecting primarily the CNS (central nervous system), and to Morquio B syndrome, a milder condition characterized by skeletal dysplasia and corneal clouding. The major clinical signs of GM1-gangliosidosis include progressive neurological deterioration, macular cherry-red spot, facial dysmorphism, hepatosplenomegaly and generalized skeletal dysplasia .
The mouse model of this disease closely recapitulates the CNS phenotype seen in patients and develop tremor, ataxia, abnormal gait and paralysis of the hind limbs . Studies on this mouse model resulted in the discovery of previously unknown molecular pathways that are induced by GM1 accumulation and result in neurodegeneration and neuronal apoptosis. We found that defective lysosomal degradation of GM1 leads to an increased amount of this ganglioside at the ER (endoplasmic reticulum) membranes, where it induces depletion of ER Ca2+ stores and activation of the unfolded protein response [26,27]. We subsequently demonstrated that GM1 accumulates specifically in GEMs (glycosphingolipid-enriched microdomains) of the MAMs (mitochondria-associated ER membranes), which are the sites of apposition between ER and mitochondria. This finding suggests that accumulation of GM1 favours the formation of contact sites between the ER and the mitochondrial membranes (Figure 3). At these GEMs, GM1 interacts with the IP3R (InsP3 receptor) Ca2+ channel and, in turn, influences Ca2+ flux between these organelles. The ensuing mitochondrial Ca2+ overload ultimately activates the mitochondrial leg of apoptosis. These exciting results enabled us to propose a novel mechanism of Ca2+-mediated apoptotic signalling, where a lipid rather than a protein acts as molecular effector of both ER-stress-induced and mitochondrial-mediated apoptosis . Overall, these studies may help us to uncover new functions of GM1 and β-gal in normal cell physiology, and to design more efficacious therapies for this devastating disease.
GM1 accumulation in β-Gal−/− neurons promotes the formation of MAMs
Therapy for LSDs is based on the mechanism of cross-correction, i.e. on the capacity of soluble lysosomal enzyme precursors to be secreted and recaptured by cells, be routed to the lysosomes and function normally (reviewed in ). On the basis of their molecular and biochemical bases, sialidosis, GS and GM1-gangliosidosis bring in different challenges from the standpoint of therapy. However, the mouse models of these diseases have made possible to test different therapeutic approaches, including transplantation of normal or overexpressing BM, ERT (enzyme-replacement therapy), SRT (substrate-reduction therapy), as well as ex vivo and in vivo gene therapy . The results of these studies have reiterated that the same therapeutic approach might not be successful in all LSDs and that any therapeutic attempt in different LSDs should be accompanied by a thorough understanding of the molecular bases of pathogenesis. For instance, in our model systems, we found that BM transplantation improved the phenotype of Ppca−/− mice, but failed in Neu1−/− mice because of lack of engraftment of the donor BM [4,30]. ERT studies with recombinant enzyme in the GS model resolved lysosomal storage in their systemic organs , whereas the same approach used in the sialidosis model was only partially successful because the recombinant enzyme elicited an immune response . SRT has proven efficacious for ameliorating disease symptoms in the GM1-gangliosidois mouse model .
BM-mediated ex vivo gene therapy was also tested in both Ppca−/− and β-Gal−/− mice with very encouraging results, including partial correction of the CNS phenotype [11,34]. An interesting observation was that, in the β-Gal−/− mice, the microgliosis that ensues at the site of neuronal apoptosis  initiates an inflammatory response accompanied by activation of specific cytokines and chemokines. We found that this pathological feature creates a local microenvironment for the recruitment into the CNS of genetically modified BM monocytes overexpressing β-Gal. The resulting cross-correction of neighbouring neurons led to reduction of GM1 storage, and hence reversal of activation of the unfolded protein response and partial rescue of the CNS phenotype.
Lysosomes in Health and Disease: A Biochemical Society Focused Meeting held at Charles Darwin House, London, U.K., 13–14 May 2010. Organized and Edited by Frances Platt (Oxford, U.K.) and Paul Pryor (York, U.K.).
central nervous system
lysosomal-associated membrane protein
lysosomal storage disease
mitochondria-associated ER membrane
protective protein/cathepsin A
This work was supported by the National Institute of Diabetes and Digestive and Kidney Diseases [grant number DK05025], the National Institute of General Medical Sciences [grant number GM060950], the National Institute of Neurological Disorders and Strokes [grant number NS065951], the National Institute of Arthritis and Musculoskeletal and Skin Diseases [grant number AR049867], the Cancer Center Core Grant [grant number CA021765], the Assisi Foundation of Memphis, the Jeweler's Charity Fund, and the American Lebanese Syrian Associated Charities.