A Proteomic Analysis of Lysosomal Integral Membrane Proteins Reveals the Diverse Composition of the Organelle

MATERIALS AND METHODS

Lysosome (Tritosome) Isolation—

The isolation of Triton WR1339-filled lysosomes has been described previously (6, 12, 13). Briefly rats (200–300-g male Sprague-Dawley from Charles River) were given an intraperitoneal injection of the non-lytic detergent Tyloxapol (85 mg/100 g of animal weight) 5 days prior to sacrifice in a CO₂ chamber. Livers were removed, homogenized, and centrifuged at 1000 × g for 10 min to produce a postnuclear supernatant. A crude organellar pellet was generated by centrifuging the postnuclear supernatant at 34,000 × g for 10 min. The pellet was resuspended in 45% sucrose, layered underneath a discontinuous gradient of 14.3% sucrose and 34.5% sucrose, and centrifuged at 77,000 × g for 2 h. Triton-filled lysosomes (tritosomes) were removed from the 14.3-34.5% interface, diluted with 0.25 m sucrose, and pelleted at 28,000 × g for 30 min. Phenylmethylsulfonyl fluoride (0.1 mm) was present in all solutions.

Electron Microscopy—

Sections of fixed tritosome pellets were processed for electron microscopy according to Leighton et al. (12). Acid phosphatase was stained using the Gomori lead phosphate technique (16).

Lysosomal Subfractionation—

The isolated tritosomes were subfractionated based on solubility as described previously (6). Briefly washed tritosomes were resuspended in 50 mm Tris-HCl, pH 8.0, 0.2 m NaCl and subjected to five freeze/thaw cycles (frozen with dry ice/ethanol and thawed at 37 °C). A membrane fraction was collected by centrifugation for 15 min at 355,000 × g at 4 °C, and the supernatant was removed (soluble/luminal fraction). The crude membrane pellet was treated with 0.1 m Na₂CO₃, pH 11.0 (30 min on ice) and centrifuged as above to give a soluble membrane-associated fraction. The remaining membrane pellet was designated the integral membrane fraction.

Fractionation of Lysosomal Integral Membrane Proteins on CM-Sepharose and Identification by Mass Spectrometry—

The integral membrane fraction was dissolved in 6 m guanidine HCl and 40 mm dithiothreitol. The solution was incubated at 45 °C for 30 min to ensure complete reduction of disulfide bonds. Iodoacetamide was added to give a final concentration of 150 mm and was incubated for 1 h in the dark at room temperature. The protein solution was then twice dialyzed overnight against column buffer (50 mm formic acid, pH 3.6, 8 m urea, 1% Elugent, Calbiochem). Conductivity measurements of the dialysis sac contents and the dialysate were made to ensure complete removal of guanidine. The protein solution was centrifuged at 10,000 × g for 15 min at 18 °C prior to being loaded on a CM-Sepharose cation exchange column pre-equilibrated in column buffer. Bound fractions were collected with a stepwise gradient of 20, 40, 80, and 250 mm NaCl dissolved in column buffer. The protein fractions were acetone-precipitated, and the pellets were dissolved in Laemmli buffer (17). The proteins were separated by standard one-dimensional SDS-polyacrylamide electrophoresis in 12% tube gels (5 mm). The gels (∼12 cm) were recovered; fixed for 2 h in 50% methanol, 10% acetic acid; placed in a gel slicer; and cut into 2–4-mm slices. Slices were then processed for in-gel digestion with trypsin as described previously (18). Recovered peptides were subjected to C₁₈ nano-LC-MS/MS-based separation using a Waters CapLC with a PicoFrit C₁₈ column (New Objective) in line with a Q-TOF mass spectrometer (Micromass).

Protein Identification—

Raw MS/MS data were processed into peak lists (.pkl) using MassLynx Version 3.5 (Micromass). Criteria for this processing were a peptide filter threshold of 2 and a minimum peak width at half-height of 2. The peak lists were analyzed using MS/MS ion search (Mascot, www.matrixscience.com) using the National Center for Biotechnology non-redundant protein (NCBI nr) data base. Some spectra were additionally interpreted by de novo sequencing (Pepseq, Micromass) and sequence tags (PeptideSearch, European Molecular Biology Laboratory Bioanalytical Research Group). All proteins that scored as significant data base “hits” by Mascot were verified by manual inspection of MS/MS data. Proteins that were identified with one unique peptide have MS/MS spectra that contain an interpretable y-ion series suitable for a peptide sequence tag search and have “identity” Mascot peptide scores (Supplemental Table 1). Data base search criteria included mass accuracy of 0.3 Da and one possible missed trypsin cleavage. In some instances, methionine oxidation and guanidinylation peptide modifications were found as well as N-acetylation of the N-terminal peptide from proteins. A non-redundant list of identified proteins was produced by removing duplications through using BLASTP to search each protein versus a data base of all the identified proteins.

Immunoblots—

Equal amounts of protein (30 μg) from purified tritosomes and the originating rat liver homogenate were separated by SDS-PAGE and transferred to nitrocellulose. Primary antibodies used were anti-rat LAMP-1 (mouse monoclonal, Calbiochem), anti-sialyltransferase 1 (ST6Gal1, kindly provided by J. Paulson), anti-GPP130 (Covance), and anti-calnexin (H-70) and anti-ribophorin 1 (C-15, both from Santa Cruz Biotechnology). Detection was done using species appropriate horseradish peroxidase-conjugated secondary antibodies with ECL and Hyperfilm (Amersham Biosciences).

RESULTS

Triton WR1339-filled Lysosomes (Tritosomes)—

Tritosomes have been extensively characterized as functional secondary (late stage) lysosomes (12, 14, 15, 19). We have previously demonstrated the enrichment of lysosomal markers (β-hexosaminidase activity, Rab7, and LAMP-1) in our tritosome preparations (70-fold enrichment of β-hexosaminidase) and the absence of markers from mitochondria (citrate synthase activity) and ER (calnexin protein) (Refs. 6 and 13 and see below). Electron microscopy of isolated tritosomes (Fig. 1A) indicates that almost all of the structures are Triton WR1339-filled lysosomes containing electron dense cores and little or no other organelles. The membranes of these lysosomes have intense positive staining for acid phosphatase activity (Fig. 1B) further characterizing them as bona fide lysosomes. Electron microscopy of tritosome membranes obtained by freeze/thaw lysis and centrifugation reveals an absence of the electron dense cores and the presence of amorphous membranes (Fig. 1C) that are acid phosphatase-positive (Fig. 1D). There is little or no membrane material evident that is negative for acid phosphatase activity. The absence of contaminating organelles and the reproducibility and reliability of the tritosome preparation along with retention of functional and physiological characteristics makes Triton WR1339-filled lysosomes an ideal choice as a proteomic model.

Separation and Identification of Lysosomal Integral Membrane Proteins—

Prefractionation of a sample into subsets prior to a proteomic-type analysis generally results in a larger array of proteins being identified. Accordingly, after freeze/thaw lysis of tritosomes to remove the soluble proteins, peripherally bound proteins were removed by treatment with sodium carbonate (20) resulting in an insoluble pellet, i.e. the integral membrane fraction. Because of incomplete dissolution of the material in non-ionic detergents (data not shown), likely due to hydrophobic and detergent-resistant domains in the membrane proteins, e.g. NPC1 (21), the integral membrane fraction was dissolved, reduced, S-alkylated in guanidine HCl, and then dialyzed versus 8 m urea and 1% Elugent at pH 3.6. The integral membrane proteins were initially fractionated by cation ion exchange chromatography on CM-Sepharose using a NaCl step gradient. The goal of the fractionation, confirmed by slab gel SDS-PAGE (Fig. 2), was to enrich low abundance proteins and to achieve their segregation from highly abundant acidic proteins, e.g. the LAMPs (for reviews, see Refs. 22 and 23). Therefore we chose the pH of the system so that these and other proteins with very low pI values did not bind (Fig. 2, see “Unbound”) to the adsorbent (under these conditions about 40% of the total protein did not bind).

The proteins from each column fraction were separated by SDS-PAGE in tube gels that were cut into 2–4-mm slices and processed for in-gel digestion, and the peptides were separated and identified by on-line LC-MS/MS. From the thousands of MS/MS spectra acquired, about 13,300 were selected for data base searching resulting in the identification of 215 proteins (non-redundant by BLASTP) in the lysosomal integral membrane protein fraction (Table I). These were organized into ontologies based on their generally accepted biological functions. Hypothetical proteins predicted in the NCBI nr data base from uncharacterized cDNAs were categorized as unknown/uncharacterized unless a conserved domain or orthologue could be found that provided information regarding the function of the protein. For example, “hypothetical protein 5031407H10” was sorted into the vesicular and protein trafficking category as it contained a SAND family (vacuolar targeting) domain (pfam03164).

Proteins were identified in gel slices that roughly corresponded to their calculated molecular weights with some exceptions such as known glycoproteins, e.g. macrosialin (predicted to be M_r ∼35,000 but found with an apparent M_r of ∼80,000) and ubiquitin (∼8-kDa monomer but found throughout the molecular mass range). To ensure that our protein separation and identification processes were representative of the true composition of the membrane, we compared the distribution of theoretical protein molecular masses listed in Table I to those predicted from translation of the mouse Mammalian Gene Collection (MGC) full-length cDNA data base (including 13,007 cDNAs). The histograms shown (Fig. 3) clearly demonstrate that our approach is not biased with respect to protein size. It should be noted that the number of proteins of M_r >200,000 identified in the present study is somewhat larger than predicted from the MGC collection, suggesting that membrane proteins constitute a higher proportion of large proteins.

Proteins of the Lysosomal Integral Membrane—

To determine the most abundant lysosomal membrane proteins in our proteomics survey, we used a pseudoquantitative measure of protein abundance consisting of the number of unique peptides assigned divided by the theoretical molecular weight of the protein. The top 15% of proteins in terms of this abundance measure are indicated in boldface type in Table I. These include LAMP-1, LAMP-2, and LIMPII as expected (for a review, see Ref. 22) and V-ATPase V0 subunit D, Rab7, and lysosomal acid phosphatase. Interestingly the lipid raft constituents flotillin-1, flotillin-2, and stomatin (24, 25) were also found in considerable abundance in the lysosomal membrane (21). Other proteins found in relatively high abundance were nicastrin, ferritin light chain, and apolipoproteins (A-V, E, and B). Although the apoB theoretical gene product is very large (>500 kDa), we have likely found the smaller intestine-specific apoB48 species since the peptides identified for apoB are N-terminal to the putative C-terminal splice site of apoB48 (approximately amino acid 2151 of apoB100) (26).

We identified the membrane-bound form of sialyltransferase 1 (ST6Gal1). It is a type II transmembrane protein originally localized to the Golgi but also shown to reside in several post-Golgi structures including the lysosomal membrane (27). When tritosomes were analyzed by immunoblotting for LAMP-1, sialyltransferase 1, and the Golgi protein GPP130, we found that both LAMP-1 and sialyltransferase 1 were substantially enriched in tritosomes, whereas GPP130 was not (Fig. 4, A and B). Our peptide coverage of sialyltransferase 1 includes the two tryptic peptides (Gly²⁹–Lys⁴⁶) on either side of a β-amyloid protein-converting enzyme 1 (Alzheimer’s disease β-secretase activity) cleavage site. Cleavage at this site, residue Gln³⁸, followed by exoproteolysis generates soluble sialyltransferase 1 with a new N terminus (Glu⁴¹) (28). We also found other glycosyltransferases in the lysosomal membrane with somewhat lower abundances (Table I). The enrichment of sialyltransferase 1 but not GPP130 and the presence of other glycosyltransferases in these membranes suggests that a previously unsuspected biological role for such proteins may exist in the lysosomal membrane.

We have previously demonstrated the presence of nicastrin and presenilin 1 (components of the γ-secretase complex) as well as acid γ-secretase activity in the lysosomal membrane (6). Nicastrin was confirmed as a major lysosomal constituent in this survey with the assignment of 13 unique peptides (Table I and Supplemental Table 1). However, we did not identify peptides from other putative components of this complex, e.g. mAph1 or Pen2. Similarly, nicastrin, but no other components of the γ-secretase complex, was found in a proteomic study of melanosomes (a lysosome-related organelle, Table II) (29).

Seventeen proteins of the lysosomal integral membrane fraction were found to have N-terminal acetylation post-translational modifications (Table I and Supplemental Table 1). This includes several proteins involved in trafficking and three proteins in the “unknown” ontology. N-terminal acetylation is a common post-translational modification of proteins in eukaryotes and is usually indicative of the cytosolic topology of the N terminus of the protein. These findings confirm the N-terminal acetylation of Rab6, originally shown in a proteomic study of Golgi vesicles (30), and we now extend this modification to other Rabs (Rab1A and Rab11B) and SNAREs (syntaxin 7 and VAMP8). Although the functional relevance of N-acetylation of these particular proteins is not known, it has recently been reported that N-acetylation is required for the correct membrane targeting of the Arf protein family member Arl3p (31, 32).

Ubiquitin (8 kDa) was identified in seven gel slices in the molecular mass range of ∼30 to ∼150 kDa and therefore exists in the integral membrane fraction as a conjugate to other proteins. Monoubiquitinylation of proteins occurs by conjugation of the C terminus of ubiquitin to a lysine side chain of the target protein, whereas polyubiquitinylation occurs by the successive conjugation of the ubiquitin molecules to lysine residues on the preceding ubiquitin. Lysine 48 of ubiquitin is the most common site for the addition of ubiquitin peptides during polyubiquitin formation (for a review, see Ref. 33). Polyubiquitinylation is generally viewed as a tag for cytosolic proteosome-based degradation, whereas monoubiquitinylation or multimonoubiquitinylation of membrane proteins at the plasma membrane has been reported to be sufficient for their transport to and degradation in the lysosome (34, 35). However, we did not identify the peptide that contains lysine 48 in the present study.

In accordance with the major role of the lysosome in transporting small molecules out of its degradative milieu, we identified (a) four proteins that contain an endomembrane protein of 70 kDa (EMP70) domain, which may have small molecule membrane transport functions (36), (b) proteins of the solute carrier family 25 (including the adenine nucleotide transporter and phosphate transporter), (c) the putative copper uptake transporter SLC31 member 1 and the histidine transporter SLC15 member 14, and (d) members of the ATP-binding cassette family. As well, the NPC1 cholesterol transporter and components of the vacuolar ATP proton pump were also found. Aside from the V-ATPase, most of the transporters identified were assigned five or fewer peptides suggesting that these are low abundance proteins in the lysosomal membrane.

Membrane Proteins Common to Lysosomes, Phagosomes, and Melanosomes—

We compiled a BLAST data base of the proteins identified from published proteomic studies of the phagosome (37) and melanosome (29) and searched it with each of the 215 proteins from our present survey. Forty proteins in our lysosomal integral membrane fraction were found also either in phagosomes and/or melanosomes (Table II). Proteins such as LAMP-1, subunits B and E of the V-ATPase, voltage-dependent anion channel 1, Rab7, flotillin-1, stomatin, and ubiquitin were common to all these organelles. The predominantly ER proteins GRP78 (BiP) and protein-disulfide isomerase were also found in phagosomes and melanosomes. Three other Rab proteins, Rab2, Rab5C, and Rab11B, were also in common as well as flotillin-2, LIMPII, and NPC1. In our survey, we identified a number of other ER, cytosol, and Golgi proteins (mostly classified as biosynthesis or metabolism in Table I) in the lysosomal membrane protein fraction. Although these proteins had somewhat lower abundances in terms of the number of peptides identified, some of these proteins have been reported to be associated with phagosomes and/or melanosomes (Table II) and are therefore likely not from vesicular contaminants of the preparations but arise in lysosomes and these related organelles through common trafficking pathways. It has been demonstrated previously that phagosomal membrane may arise from ER membrane, and the maturation of the phagosome to phagolysosome coincides with a decrease in the quantity of ER proteins (37, 38). In agreement with its level of maturation, immunoblots of tritosomal proteins show that the ER membrane protein calnexin is undetectable, while ribophorin 1 is present at a low level (Fig. 4C). This suggests that ER membrane and its associated proteins may be involved in early lysosomal structures and are lost at variable rates during maturation of the organelle.

DISCUSSION

The approach of organelle-based proteomics allows the concentration and thus the identification of otherwise rare cellular proteins while at the same time offering clues to their in vivo function based on their localization. This technique is dependent on the quality and yield of the organelle in question. Since lysosomes are heterogeneous in terms of the native densities, they are difficult to purify by conventional methods. Triton-filled lysosomes overcome this problem by decreasing the overall densities of a large population of lysosomes in rat liver, allowing milligram quantities of highly purified organelles to be isolated in a single step. Additionally since tritosomes are secondary lysosomes they represent a population of relatively stable, end stage organelles. Thus their protein composition may be considered constant and constitutive, i.e. they are less a reflection of the transitional phase(s) of the life cycle of lysosomes.

The 215 proteins reported here to our knowledge represent the largest number of proteins identified from a membrane fraction of the lysosome to date. A previous study identified 27 proteins from a soluble lysosomal fraction that bound to a mannose 6-phosphate affinity column (39). However, lysosomal membrane proteins are generally not transported to the organelle via this pathway as reflected by the highly sialylated state of their N-linked oligosaccharides. Our combination of ion exchange chromatography and one-dimensional SDS-PAGE followed by in-gel trypsin digestion and mass spectrometry (LC-MS/MS) provides a high degree of protein separation. It is robust, offers reasonable reproducibility, and can be used for comparing protein profiles from multiple complex protein samples.

One interesting finding was the relatively high abundance of lipid raft proteins in the lysosomal membrane. Lipid rafts are small membrane domains enriched in cholesterol and sphingolipids that have the ability to sort, concentrate, and organize proteins for various cell signaling events. They have recently been shown to concentrate SNARE proteins involved in vesicle fusion (40) and have been found associated with late endosomal (41), lysosomal (21), and phagosomal membranes (42). Interestingly a coincident increase in the quantity of lipid rafts in phagosomes with their maturation into phagolysosomes was demonstrated (42); this is in agreement with our present finding of an apparent high abundance of lipid rafts on mature lysosomes. Their presence in the lysosomal membrane suggests major signaling and sorting functions for even this late stage organelle. Consistent with this we have identified the rat orthologue of MEK binding partner 1 protein, a scaffold component of the endosomal ERK pathway of the mitogen-activated protein kinase cascade and which is known to bind to p14 (not identified in this study) on the cytosolic face of late endosomes/lysosomes (43), a process required for the endosomal activation of ERK1/2 (44). Similarly, cellular repressor of E1A-stimulated genes (CREG) was identified confirming a previous study of lysosomal mannose 6-phosphate receptor-binding proteins (45). CREG is a secreted glycoprotein that promotes differentiation of pluripotent stem cells (46) and inhibits cell growth in a mannose 6-phosphate/insulin-like growth factor II receptor-dependent manner (47). We also identified LRG-47, a macrophage interferon-γ-inducible GTPase required for pathogen-containing phagosomes to fuse with lysosomes (48). Also identified were syntaxin 7, syntaxin 8, vti1b, and VAMP8. These occur in a complex that forms on late endosomes, and this complex is required for fusion events with other endosomes (49) and lysosomes (50). Collectively these findings suggest a crucial role for the lysosomal membrane in intracellular communication and signaling events.

A well described function of the lysosomal membrane has been the transport of the products of degradation to the cytosol for reuse by the cell. However, few proteins linked to these activities in the lysosome have been identified. A notable group of proteins that we have identified in this report are those predicted to have transport functions, e.g. EMP70 domains, or the those in the solute carrier family. These proteins can now be assessed for molecular transport function with respect to the lysosomal membrane and its luminal contents.

Although de novo biogenesis of lysosomes is not well understood, identification of proteins in common among lysosomes and their related organelles may help elucidate how they are formed. As a starting point, many well known proteins of the lysosomal membrane identified in this study were also identified in other proteomic surveys of the phagosome and melanosome (Table II). Another clue to begin to understand the processes of lysosome biogenesis and maintenance arises from our finding of the so-called “grey-lethal” (Gl) protein. Gl has been implicated in a severe form of human malignant autosomal recessive osteopetrosis (51). In mice, mutation of this gene causes a defect in coat color through “clumping” of pigment granules (melanosomes) in melanocytes (52) and a defect in bone resorption caused by defective osteoclast maturation (53), i.e. this protein affects the function of specialized lysosomes (for reviews, see Refs. 4 and 54). The localization to the lysosomal membrane of Gl protein and a large number of other proteins with no previously ascribed function will help greatly in elucidating their biological roles.

From this proteomic survey, it is clear that the protein composition of the lysosomal membrane is influenced by contributions from Golgi, ER, and plasma membrane components. Highlighting this was the identification in the integral membrane fraction of Golgi transferases, e.g. sialyltransferase 1 (27) and other glycosyltransferases, some of which have been shown to have post-Golgi localizations (for a review, see Ref. 55). As well, some abundant ER proteins were found, e.g. protein-disulfide isomerase, ribophorin 1, and cytochrome P450s, but interestingly no calnexin and plasma membrane receptor proteins, e.g. asialoglycoprotein receptors and Toll-like receptor 3, were found. Although delivery of proteins and cargo to the lysosome from endosomes and the trans-Golgi network is well established, autophagy, a process of sequestering cytoplasm and organelles for degradation, also delivers protein to the lysosome presumably using LAMP-2 as a receptor (56). Many abundant ER and cytosolic proteins such as GRP78, protein-disulfide isomerase, cytochrome P450, and betaine-homocysteine S-methyltransferase have been found to be associated with autophagic vacuoles (57–59).

In summary, the lysosome is traditionally thought to be the end point of the endocytic pathway and to have a chiefly nonspecific degradative capacity for its luminal contents. Through our lysosomal proteomic strategy it seems clear that the membrane is a very diverse structure that receives membrane and protein contributions from a variety of subcellular sources and pathways. Through vesicular trafficking and intracellular membrane dynamics, the high degree of diversity of the proteins of the lysosomal membrane is indicative of a highly complex organelle. The diverse set of proteins in the lysosomal membrane could be organized into distinct domain structures comparable to those currently envisioned for earlier endosomal compartments (60) that are demarcated by at least two different Rab proteins (61). Since we identified several different Rabs in the lysosomal membrane, it remains to be determined whether each of these Rab proteins is involved in the organization of additional protein sets and different domains in the lysosomal membrane.