ARTICLE pubs.acs.org/jpr
Differential Glycomics of Epithelial Membrane Glycoproteins from Urinary Exovesicles Reveals Shifts toward Complex-Type N-Glycosylation in Classical Galactosemia Simon Staubach,† Peter Schadewaldt,‡ Udo Wendel,§ Klaus Nohroudi,^ and Franz-Georg Hanisch*,†,z †
Institute of Biochemistry II, Medical Faculty, University of Cologne, K€oln, Germany German Diabetic Centre, Department of Clinical Biochemistry and Pathobiochemistry,UKD, University of D€usseldorf, D€usseldorf, Germany § Department of General Pediatrics, UKD, University of D€usseldorf, D€usseldorf, Germany ^ Department of Anatomy I, Medical Faculty, University of Cologne, K€oln, Germany z Center for Molecular Medicine Cologne, University of Cologne, K€oln, Germany ‡
bS Supporting Information ABSTRACT: A variety of genetic variations in the galactose-1phosphate uridyltransferase (GALT) gene cause profound activity loss of the enzyme and acute toxic effects mediated by accumulating metabolic intermediates of galactose in newborns induced by dietary galactose. However, even on a severely galactoserestricted diet, patients develop serious long-term complications of the CNS and ovaries, which may result from damaging perturbations in cell biology caused by endogenously synthezised galactose. Under galactose stress, the cosubstrate of GALT, galactose-1-phosphate, accumulates and disturbs catabolic and anabolic pathways of the carbohydrate metabolism with potential effects on protein glycosylation and membrane localization of glycoprotein receptors, like the epidermal growth factor receptor. To address this issue in view of a cellular pathomechanism, we performed a differential semiquantitative N-glycomics study of membrane proteins. A suitable noninvasive cellular material derived from epithelial plasma membranes was found in urinary exovesicles and in the shed Tamm Horsfall protein. By applying matrixassisted laser ionization mass spectrometry on permethylated, PNGaseF released N-glycans, we demonstrate that GALT deficiency is associated with dramatic shifts from prevalent high-mannose-type glycans found in healthy subjects toward complex-type N-linked glycosylation in patients. These N-glycosylation shifts were observed on exosomal N-glycoproteins but not on the Tamm Horsfall glycoprotein, which showed predominant high-mannose-type glycosylation with M6. KEYWORDS: classical galactosemia, differential glycomics, urinary exosomes, membrane glycoprotein, Tamm Horsfall glycoprotein, N-glycosylation, MALDI mass spectrometry
’ INTRODUCTION Classical galactosemia is caused by a profound deficiency of galactose-1-phosphate uridyltransferase (GALT; EC 2.7.712), the second enzyme in the Leloir pathway of galactose metabolism, resulting in a severely impaired galactose metabolism. Normally, galactose is converted to glucose-1-phosphate and metabolized in the glycolytic pathway, or alternatively, galactose may be activated to UDP-galactose (pyrophosphorylase pathway), the cosubstrate involved in enzymatic galactosylation of glycoproteins and glycolipids. Newborn infants affected with severe GALT deficiency develop a potentially lethal hepatotoxic syndrome induced by dietary (exogenous) galactose, which usually resolves rapidly after the institution of a galactose-restricted diet. In contrast to the rapid recovery of liver disease, the long-term outcome in classical r 2011 American Chemical Society
galactosemia is disappointing. Despite strict adherence to the galactose-restricted diet, a great percentage of patients develop long-term complications, which may result from injurious perturbations in cell biology caused by endogenously synthesized galactose. The long-term complications, comprising cognitive impairment, speech defects, motor function disturbances, and in most female patients, hypergonadotropic hypogonadism, are related to the CNS and ovaries as target tissues. They have their origin in prenatal life and worsen, if at all, only slightly with time.1,2 The possible mechanisms underlying long-term complications comprise the individual or combined effects of excess galactose-1-phosphate, galactitol or other galactose metabolites, Received: July 28, 2011 Published: November 17, 2011 906
dx.doi.org/10.1021/pr200711w | J. Proteome Res. 2012, 11, 906–916
Journal of Proteome Research or defective glycosylation of proteins and lipids leading to direct and/or indirect cell or tissue dysfunction.3 A number of reports have demonstrated that GALT deficiency leads to aberrant protein glycosylation especially prior to dietary galactose restriction, but also in a diminished form under a galactose-restricted diet.4 6 Reports on glycosylation abnormalities in galactosemic patients refer so far exclusively to secreted proteins, in particular to plasma glycoproteins such as transferrin and glycopeptide hormones such as follicle-stimulating hormone (FSH).3,6 Transferrin of galactosemic patients prior to dietary galactose restriction was shown to have large proportions of truncated glycans lacking terminal sialic acid and galactose residues in N-glycan antennae.4 At least the distinct glycosylation abnormalities of transferrin prior to dietary galactose restriction is the consequence of the acute hepatotoxic status caused by exogenous galactose. These abnormalities should be different from aberrant glycosylation, which might be related to cell or tissue dysfunction responsible for the development of long-term complications. Rather, for such a harmful action, abnormally glycosylated cellular glycoproteins come into consideration because proper N- and O-glycosylation of glycoproteins is a precondition for proper sorting and trafficking of membrane glycoproteins in order to reach their definite destination and function.7 Likeweise, a defective synthesis of glycoproteins and galactolipids can have a strong impact on normal myelin formation.8 Thus, in the search for the role that aberrant glycosylation may play for the development of long-term complications in galactosemia, the glycosylation of cellular glycoproteins should be studied. Research using patient-derived cellular material is largely confined to blood lymphoblasts or dermal fibroblasts. In the course of preparative work, we searched for an easily acccessible cellular material that could be regarded as of epithelial origin. Human urine has been reported to contain exosomes,9 30 100 nm vesicles originating as internal vesicles within endosomal multivesicular bodies and released by a variety of cells, like dendritic cells and epithelial cells, in the functional context of cell communication or protein export. The membrane of these nanovesicles exhibits a close relationship to the protein composition of the respective plasma membranes, and hence these exovesicles can reveal insight into the proteome and glycome of the exosome forming cell.10,11 In case of urinary exosomes, every renal epithelial cell facing the urinary tract contributes to the pool of exosomes. Urothelial cells are also the origin of the most abundant glycoprotein in human urine, the Tamm Horsfall glycoprotein (THP, also called uromodulin), which is exclusively N-glycosylated. THP is integrated into the plasma membrane via a GPI-anchor and hence represents an integral membrane protein.12 However, it is shed in large quantities into urine, where its concentration can reach 0.3 mg/mL. Both urinary constituents, exosomal membranes and THP, may generally serve as a suitable noninvasive starting material for bio(glyco)marker discovery. This report refers to the first comprehensive study of a cellular N-glycoprotein dysglycosylation associated with classical galactosemia. The results demonstrate that findings from analyses of secretory glycoproteins cannot be regarded as representative for changes in the glycosylation of cellular membrane-bound proteins. Strikingly, the GALT-deficiency does not result in an under-galactosylation of complex-type N-linked chains but instead in a dramatic shift from preponderant high-mannose-type N-glycosylation in control samples to (galactosylated) complextype chains in GALT-deficient patients.
ARTICLE
’ EXPERIMENTAL SECTION Materials
Urine samples (approximately 500 mL, collected in the morning) were obtained from healthy volunteers (5 adult individuals) and from patients with classical galactosemia (12 adult individuals; 6 female, 6 male; approved by the Ethical Commision of the University Clinic of Cologne on November 23, 2010) and kept frozen until use. The galactosemic patients carried the c.A1466G (p.Q188R) mutation in a homozygous fashion and exhibited a very low residual GALT activity characteristic of classical galactosemia. All galactosemic patients were on a severely galactose-restricted diet. Preparation of Urinary Exovesicles
Exosomes (and a minor proportion of other plasma membrane-derived vesicular material) were separated from 200 mL of urine by differential centrifugation at 4 C. The urine samples were initially filtered through Whatmann filters and then centrifuged at 10000g for 30 min. This step was repeated to remove pelletted material, which was accidentally transferred during removal of the supernatant. In the next step, the supernatant was ultracentrifuged at 114000g for 90 min. The pelletted exovesicles were suspended in 200 μL of PBS and transferred into 1.5 mL tubes for a final ultracentrifugation at 135000g for 60 min. The exovesicles obtained by this centrifugation step were stored frozen at 20 C. Electron Microscopy of Exovesicular Preparations
A Formvar-coated copper grid was placed on the exosomecontaining suspension for 30 min to allow adhesion of exosomes. Then, the grid was shifted to a drop of fixative for 15 min, followed by washing for 1 min in 0.1 M PBS for five times (after each washing step, excessive liquid was removed with filter paper). Afterward, the grid was applied to a solution containing 2% uranylacetate for approximately 90 s and washed three times on Aqua bidest for 1 min, and finally, the excessive liquid was removed using filter paper. Thereafter, the grid was dried at room temperature. Isolation of Tamm Horsfall Protein
Exovesicle-depleted and neutralized urine was further processed for the isolation of Tamm Horsfall protein (THP) by filtration over a layer of 20 g of diatomaceous earth, as described by Serafini-Cessi et al.13 Loading excessive protein in SDS gel electrophoresis, the presence of only small amounts of contaminating proteins was revealed, which were however identified as unglycosylated (protein identification by LC MS/MS of tryptic peptides as described in ref 14, refer to the detailed protein reports in the Supporting Information). For analysis of N-linked glycans on THP, the protein was digested in-gel after gel electrophoresis, and the eluted glycopeptides were treated with PNGaseF (see below). In-gel digestion with trypsin was performed under standard conditions as applied in proteomics workflows.14 The (glyco)peptides were eluted into two aliquots of 0.5 mL of acetonitrile in water (50%) with rotation for 2 2 h at room temperature. The eluate was heated at 90 C for 10 min to destroy residual protease activity and dried by vacuum rotation prior to addition of PNGaseF. Release and Derivatization of N-Linked Glycans
N-Linked glycans were released from the protein cores by digestion with PNGaseF. In detail, the samples (exovesicles or THP, each corresponding to 10 100 μg protein, or tryptic 907
dx.doi.org/10.1021/pr200711w |J. Proteome Res. 2012, 11, 906–916
Journal of Proteome Research
ARTICLE
Figure 1. Quality validation of exovesicular preparations by electron microscopy and Western blot analysis of fractions from density gradient centrifugation. (a) Electron micrograph of exovesicular preparations from human urine demonstrating the preponderant presence of exosomes (upper and middle panel) and the binding of exosomes to polymerized THP (lower panel). (b) Western blot of fractions from density gradient centrifugation of urinary exovesicles demonstrating the presence of exosomal markers preferentially in the high-density fraction 5 (MUC1, mucin-1; Alix, apoptosislinked gene-2 interacting protein X; hsp70, heat-shock protein 70; Flot-2, flotillin-2; Giα, G protein subunit α; hsp27, heat-shock protein 27). (c) Representative Coomassie-stained SDS-polyacrylamide gels of exovesicular preparations from a galactosemia patient (lane 1) and two controls, c1 and c2 (lanes 2 and 3), showing the presence of THP as a major protein.
glycopeptides from in-gel digestion) were taken up in 50 μL of 50 mM ammonium bicarbonate buffer, pH 8.5, and briefly sonicated. PNGaseF (BioLabs, 250 U) was added, and the reaction mixture was kept at 37 C for 16 h. After drying by vacuum rotation, the glycans were solubilized in 0.1% aqueous trifluoroacetic acid (100 μL) and separated from residual protein/peptides by passage over BondElutC18 (100 mg, glass column) and washing off with 0.9 mL of water. The methylation procedure started after extensive drying of the sample by vacuum centrifugation, followed by vacuum drying in a desiccator over P2O5/KOH for 1 hr. All handlings were performed under argon atmosphere. To the dry sample, 100 μL of base (2.5%, w/v, finely dispersed NaOH in dry DMSO) was added. The sample was briefly sonicated for 1 2 min and incubated for 30 min at room temperature with occasional shaking. Finally, an aliquot of 50 μL of methyl iodide was pipetted to
the frozen reaction mixture, followed by incubation for a further 30 60 min at room temperature. Prior to the extraction step, the reaction mixture can optionally be neutralized with 40 μL of 1 M acetic acid. Extraction of methylated glycans was performed by stepwise addition of 0.3 mL of chloroform and 0.2 mL of water. After vigorous mixing and phase separation, the water layer was removed and repeatedly replaced by at least two further aliquots of 0.2 mL. The chloroform phase was dried under nitrogen, and the glycans were solubilized in methanol prior to application onto the MALDI target. MALDI-TOF/TOF Mass Spectrometry of Methylated Glycans
Matrix-assisted laser desorption ionization (MALDI) mass spectrometry was performed on a UltrafleXtreme instrument (Bruker Daltonics, Bremen, Germany). The permethylated glycans (approximately 500 ng) contained in 20 μL of methanol 908
dx.doi.org/10.1021/pr200711w |J. Proteome Res. 2012, 11, 906–916
Journal of Proteome Research
ARTICLE
Figure 2. Representative MALDI mass spectra of permethylated N-glycans from urinary exovesicles of GALT-deficient patients and healthy control subjects. Mass signals corresponding to a calculated composition of H3+nN2 (with n = 0 6) represent high-mannose-type glycans M3 M9, mass signals corresponding to a calculated composition of HnNn (with n > 3) represent undergalactosylated complex-type species, and mass signals corresponding to a composition of H3+nN2+m (with n = 1 4, m = 2 4) represent complex-type N-glycans. Major signals correspond to the sodium adducts M+Na that have lost NaOCH3. Relative masses marked by an asterisk refer to PSD fragments of N-linked glycans, like S1H1N2+ (registered at m/z 1070.6). MALDI mass spectrum of permethylated N-glycans from urinary exovesicles of a galactosemic patient (a) and of a healthy control individual (b). STD refers to the internal standard LNFPI. 909
dx.doi.org/10.1021/pr200711w |J. Proteome Res. 2012, 11, 906–916
Journal of Proteome Research
ARTICLE
Table 1. Relative Amounts of High-Mannose- and Complex-Type N-Glycans in Exovesicular Membranes from Galactosemic Patients and Healthy Control Subjectsa galactosemic patient N-linked glycans
M
high mannose type
54 (m/z)
composition
1 (%)
2 (%)
4 (%)
5 (%)
6 (%)
1321.7
H4N2
4.4
2.7
9.5
6.6
7.0
4.8
1525.8
H5N2
16.9
9.8
19.6
12.5
16.4
13.0
1729.8 1934.0
H6N2 H7N2
16.1 10.1
27.7 7.8
20.4 7.1
10.6 4.7
29.2 6.1
11.6 5.5
2138.1
H8N2
∑ complex type
3 (%)
4.0
1.2
3.2
2.3
2.1
4.4
51.6
49.2
59.8
36.7
60.8
39.3
2377.2
S1H5N4
8.1
17.2
12.2
12.5
14.0
14.3
2551.3
S1F1H5N4
4.4
3.5
3.4
3.1
1.8
11.6
2738.4
S2H5N4
29.4
17.6
18.5
39.8
17.9
22.5
2912.5
S2F1H5N4
5.6
3.5
2.7
5.1
1.8
8.5
3187.6 3722.9
S2H6N5 S3F2H6N5
0.4 0.4
6.6 2.3
2.7 0.8
1.6 1.2
3.3 0.3
2.4 1.4
48.3
50.8
40.2
63.3
39.2
60.8
11 (%)
12 (%)
∑
galactosemic patient N-linked glycans high mannose type
M
54 (m/z)
composition
8 (%)
9 (%)
10 (%)
1321.7
H4N2
7.7
10.0
9.0
8.3
10.3
8.1
1525.8
H5N2
14.5
8.8
18.8
12.9
34.4
22.8
1729.8
H6N2
12.1
17.5
9.8
8.8
8.9
18.8
1934.0
H7N2
4.8
7.5
2.5
2.5
2.1
6.7
2138.1
H8N2
3.4 42.5
2.5 46.3
2.7 42.9
2.9 35.4
3.8 59.5
2.7 59.1
2377.2
S1H5N4
13.5
15.0
11.2
13.3
6.9
12.1
2551.3
S1F1H5N4
10.6
10.0
4.0
2.9
2.9
4.0
2738.4
S2H5N4
13.5
16.3
36.9
42.1
25.0
24.8
2912.5
S2F1H5N4
12.6
6.3
3.6
5.0
2.9
0.0
3187.6
S2H6N5
4.4
3.8
0.5
0.4
1.7
0.0
3722.9
S3F2H6N5
2.9
2.5
0.9
0.8
1.3
0.0
57.5
53.8
57.1
64.6
40.8
40.9
∑ complex type
7 (%)
∑
healthy control N-linked glycans high mannose type
M
54 (m/z)
composition
2 (%)
3 (%)
4 (%)
5 (%)
1321.7
H4N2
3.6
6.5
5.0
1.8
4.7
1525.8
H5N2
34.1
30.0
30.6
7.9
23.7
1729.8
H6N2
22.9
23.2
22.9
62.8
23.3
1934.0
H7N2
10.3
11.3
12.5
7.9
14.6
2138.1
H8N2
∑ complex type
1 (%)
4.9
4.8
5.3
0.6
7.3
75.8
75.8
76.3
81.1
73.6
6.3 2.7
4.0 6.5
5.6 4.8
3.1 4.9
7.1 4.3
2377.2 2551.3
S1H5N4 S1F1H5N4
2738.4
S2H5N4
11.2
5.4
6.2
1.8
6.7
2912.5
S2F1H5N4
1.8
6.5
4.8
3.7
5.5
3187.6
S2H6N5
1.8
0.7
1.5
3.1
1.6
3722.9
S3F2H6N5
0.5
1.1
0.9
2.4
1.2
24.2
24.2
23.7
18.9
26.4
∑
a % = ratio of single peak intensities to total listed peak intensities. m/z-values correspond to the molecular ions M + Na ( 54). Only major signals were considered in the quantitative evaluation.
were applied to the stainless steel target by mixing a 0.5 μL aliquot of sample with 1.0 μL of matrix (saturated solution of 2,5-
dihydroxy benzoic acid in ACN/0.1% TFA, 1:2). Analyses were performed by positive ion detection in the reflectron mode. 910
dx.doi.org/10.1021/pr200711w |J. Proteome Res. 2012, 11, 906–916
Journal of Proteome Research
ARTICLE
Figure 3. Column diagram showing relative amounts of high-mannose-type vs. complex-type N-glycans on urinary exovesicular membranes from GALT-deficient patients and healthy controls. Left panel: column diagramm of relative amounts of high-mannose-type and complex-type N-glycans with indicated standard deviations. Right panel: box-and-whisker plot of the ratio of relative amounts of complex-type vs. high-mannose-type N-glycans. The box indicates the lower and upper quartiles and the central lines are the medians. The points at the ends of the whiskers show the 2.5% and 97.5% values (centiles). Healthy controls, n = 5; GALT-deficient patients, n = 12.
Ionization of crystallized analytes was induced with a pulsed Smart-beam laser (accumulation of about 5000 shots), and the ions were accelerated in a field of 20 kV and reflected at 23 kV.
insight into the patterns of N-linked chains based on molecular masses and the calculated monosaccharide compositions in terms of Fuc (F), Hex (H), HexNAc (N), or NeuAc (S) and based on fragment ions (sequence information from protonated B-type ions according to the nomenclature of Domon and Costello17). In Figure 2, two representative survey spectra of methylated N-glycans from a galactosemic patient and from a healthy control individual are shown. The patterns of N-linked glycans on urinary exovesicles were characterized by the presence of two classes of chains, the highmannose-type glycans comprising mostly M4 to M9 and the complex-type glycans, which were predominantly of the biantennary subtype A2 without core-fucosylation. Minor subtypes were undergalactosylated complex-type species with GlcNAc terminating antennae and glycans of higher antennarity. Qualitatively, the patterns revealed for GALT-deficient patients and the respective healthy controls overlapped to a large extent. On quantitative evaluation of the data, which were based on the summarized signal intensities for high-mannose-type glycans (H3+nN2, with n = 0 6) and complex-type glycans (H3+nN2+m, with n = 1 4, m = 2 4) exovesicles from healthy controls showed patterns of N-glycans that were dominated by highmannose-type vs. complex-type glycans (ratio about 3:1) (Table 1). On the contrary, GALT-deficient patients expressed dramatically increased proportions of complex-type chains on exovesicular glycoproteins with a ratio of about 1:1 (Figure 3). The shift of relative amounts of complex-type vs. high-mannose-type glycans from 0.32 to 1.08 (factor: 3.38) can be regarded as even underestimated because of the presence of THP in exovesicular fractions, which is shown in this study to be glycosylated preponderantly with high-mannose-type glycans in patients and healthy controls (see below). The presence of THP in exovesicular preparations did not, however, strongly influence the ratios of complex- vs. high-mannose-type glycans in samples with high or low THP content (see Table S1, Supporting Information). The variations of ratios (complex- vs. high-mannose-type glycans) measured for independent urine samples from individual healthy control subjects were only small (Table S2, Supporting Information).
’ RESULTS AND DISCUSSION Quality Control of Exovesicular Preparations and Validation of the Analytical Material
Besides exosomes, the exovesicle fraction of human urine could contain other types of vesicular material derived from epithelial plasma membranes of the urothelium, like shedding microvesicles (SMV) or apoptotic blebs (AB).15 However, electron microscopy of crude exovesicle fractions from human urine showed mainly exosome-like, cup-shaped vesicles in the expected size range (Figure 1a). In previous studies, exosomes were often found to adhere to filamentous structures, which are known to form during polymerization of THP.16 This phenomenon likely explains the dominant presence of THP in exovesicular proteins (see Figure 1a,c and below). Filtered exovesicular fractions (cutoff of 100 μm), which were depleted of SMVs, revealed similarly high THP contaminations and no significant alterations of the vesicular N-glycoprofiles but drastically reduced yields and a loss of structural integrity (data not shown). Western blots of fractions from a density gradient centrifugation demonstrated that the exovesicles from human urine comigrate preferentially with high-density fractions characterizing the buoyant density of exosomes and were positive for a series of exosome characteristic marker proteins (Figure 1b). Irrespective of these findings, we refer in this report throughout to urinary exovesicles (without further specification) as an easily acccessible source for the performance of bioanalytics on patient-derived epithelial membranes. N-Glycosylation on Urinary Exovesicles
Exovesicles prepared from 100 mL of human urine were measured to contain about 200 μg of protein on the average (with individual fluctuations between 60 and 600 μg). Enzymatically liberated N-linked glycans were estimated by reference to an internal standard (lacto-N-fucopentaose I) to make up 0.5 5 μg per 100 μg of protein. The permethylated glycans were analyzed by MALDI-TOF and TOF/TOF mass spectrometry to reveal 911
dx.doi.org/10.1021/pr200711w |J. Proteome Res. 2012, 11, 906–916
Journal of Proteome Research
ARTICLE
Figure 4. MS/MS spectra of permethylated high-mannose and complex-type N-glycans registered by MALDI-TOF/TOF MS. (a) MS2 spectrum of the monosialylated biantennary complex-type N-glycan registered at the precursor ion mass (M + Na 54) at m/z 2377. (b) MS2 spectrum of the highmannose-type N-glycan M6 registered at m/z 1730 (corresponding to m/z 1784 NaOCH3). The annotated fragment ions correspond to protonated B-type ions according to the nomenclature of Domon and Costello.17
The most dominant complex-type glycans from exovesicular membranes of GALT-deficient patients were of biantennary structure and lacked core fucosylation (Figure 2 and Table S3,
Supporting Information). Only trace amounts of structures with higher antennarity were found. There was a considerable fraction of N-glycans with complex-type structure that had a 912
dx.doi.org/10.1021/pr200711w |J. Proteome Res. 2012, 11, 906–916
Journal of Proteome Research
ARTICLE
Figure 5. Representative MALDI mass spectra of permethylated N-glycans from urinary THP of GALT-deficient patients and healthy control subjects. (a) Representative MALDI mass spectrum of permethylated N-glycans liberated from electrophoretically purified THP contaminating the exovesicular preparation of a galactosemic patient. (b) The respective sample from a healthy control subject. STD refers to the internal standard LNFPI.
monosaccharide composition of HnNn (with n = 4 or 5) and fluctuated between 15% and 30% of that fraction between individual samples. These compositions indicate the presence of under-galactosylated species, which were however found in
both sample series from GALT-deficient patients and healthy controls without obvious quantitative correlation with one of the sample groups. The structures of the major components in the fractions of N-linked glycans were corroborated by 913
dx.doi.org/10.1021/pr200711w |J. Proteome Res. 2012, 11, 906–916
Journal of Proteome Research
ARTICLE
Table 2. Relative Amounts of High-Mannose- and Complex-Type N-Glycans on Urinary THP from Galactosemic Patients and Healthy Control Subjectsa galactosemic patient N-linked glycans high mannose type
M
54 (m/z)
composition
1322
H4N2
1526
H5N2
1730 1934
H6N2 H7N2
57.5 7.4
2138
H8N2
∑ complex type
1 (%)
2 (%)
3 (%)
4 (%)
5 (%)
8.6
6.0
15.9
7.5
7.7
20.3
16.8
32.7
22.6
27.3
68.8 6.4
34.1 12.0
54.9 6.0
56.9 7.7
6.2
0.0
5.3
0.9
0.0
100.0
98.0
100.0
91.9
99.6 0.4
2378
S1H5N4
0.0
2.0
0.0
3.1
2552
S1F1H5N4
0.0
0.0
0.0
5.0
0.0
2739
S2H5N4
0.0
0.0
0.0
0.0
0.0
2913
S2F1H5N4
0.0
0.0
0.0
0.0
0.0
3188 3724
S2H6N5 S3F2H6N5
0.0 0.0
0.0 0.0
0.0 0.0
0.0 0.0
0.0 0.0
0.0
2.0
0.0
8.1
0.4
∑
healthy control N-linked glycans high mannose type
M
54 (m/z)
composition
1322
H4N2
5.78
7.47
1526
H5N2
14.05
16.95
1730
H6N2
60.33
62.64
1934
H7N2
7.44
11.49
2138
H8N2
0.00 87.60
0.86 99.41
∑ complex type
1 (%)
2 (%)
2378
S1H5N4
2.73
0.57
2552
S1F1H5N4
0.00
0
2739
S2H5N4
0.91
0
2913
S2F1H5N4
0.00
0
3188
S2H6N5
0.00
0
3724
S3F2H6N5
0.00
0
3.64
0.57
∑ a
% refers to the ratio of single peak intensities to sum of annotated peak intensities. m/z-values correspond to the molecular ions measured as M-54 (loss of sodium methylate). Only major signals were considered in the quantitative evaluation.
MALDI-TOF/TOF MS2 analysis (representative examples shown in Figure 4). The M + Na ( 54) species (loss of sodium methylate) fragment under laser-induced dissociation conditions similar to the protonated MH species, which yield preferentially B-type proton adduct ions from the nonreducing terminal of the glycans in all MS2 spectra. Fragmentation by postsource decay is preferred at C1 O bonds of HexNAc moieties, explaining the restricted set of B ions. What are the potential functional implications of changes in the N-glycoprofiles of membrane glycoproteins? Evidence is accumulating in recent years that the sorting for targeted trafficking of membrane glycoproteins to the apical plasma membrane is under partial control of N- and/or O-glycosylation.7 It can be postulated accordingly that the expression of glycosylated receptors, like the epithelial growth factor receptor, EGFR, in membrane rafts requires specific glycan signals for (galectinmediated) entrapment in specific rafts. A dysglycosylation should have an impact on these sorting and trafficking events and could result in a mislocalization of prominent membrane receptors with concomitant perturbations in the cellular signaling.18 In a
cellular model of GALT deficiency, it has been shown that EGFR is actually lacking in lipid rafts when the cells are grown under galactose stress.19 Other explanations of the observed phenomenon cannot be ruled out, like a potential contribution of disease-associated kidney insufficiencies that lead to secretion of serum glycoproteins and their partial association with exosomal preparations as shown for fetuin and ceruloplasmin.20,21 N-Glycosylation of Urinary Tamm Horsfall Protein
The analysis of urinary THP, the major N-glycoprotein shed from urothelial membranes, offers the chance to get independent evidence on disease-related alterations in cellular N-glycosylation of proteins. However, there is also a necessity to characterize the N-glycoprofiles of this protein because it generally forms a major protein component of exovesicular preparations from urine because of its adherence to the outer surface of exovesicles as a long filamentous protein polymer (Figure 1c). These THP polymers have been claimed to be removable under strong denaturing conditions,16 which was however not confirmed in our hands. 914
dx.doi.org/10.1021/pr200711w |J. Proteome Res. 2012, 11, 906–916
Journal of Proteome Research Coomassie-stained SDS-gels of exovesicular proteins show accordingly THP as a major protein, which could have impact on the measured glycoprofiles. For this reason, we analyzed the N-glycans on THP from exovesicles of patients and controls by in-gel trypsin digestion of the electrophoretically separated protein, elution of the glycopeptides, and PNGaseF-catalyzed liberation of the N-linked glycans. In contrast to the total exovesicular membranes, THP was characterized by the predominant expression of high-mannose-type N-glycans (Figure 5). Most dominant was throughout the M6 isoform, which is in agreement with earlier studies on THP glycosylation.22 The N-glycosylation pattern of THP showed no major individual fluctuations and was unchanged in samples from patient urine (Table 2). These data strikingly support the validity of the abovedescribed N-glycosylation shifts on exovesicular membranes because the contaminating THP can only contribute to the pool of high-mannose-type chains in samples but cannot be responsible for the dramatic increases of complex-type chains (see also Table S1, Supporting Information). Why is THP N-glycosylation not fluctuating with GALT deficiency similar to that of other membrane-integral N-glycoproteins in the urothelial membrane? It is known that THP is produced and shed from cellular membranes at high rates, which could imply that it passes the Golgi apparatus only once. We and others have demonstrated for single-passage glycoproteins, like secreted isoforms of human MUC1, that these proteins carry preferentially high-mannose-type N-glycans compared to their membrane-tethered isoforms expressing more complex-type chains.23 The latter often recycle after re-endocytosis through the Golgi compartments or the trans-Golgi network back to the plasma membrane,24 a process that could be accompanied by further processing of initially formed high-mannose-type to complex-type glycans.
ARTICLE
will hopefully give insight into cellular changes and pathomechanisms induced by endogenous galactose-stress via a protein dysglycosylation.
’ ASSOCIATED CONTENT
bS
Supporting Information Detailed protein reports and Tables S1 S3. This material is available free of charge via the Internet at http://pubs.acs.org.
’ AUTHOR INFORMATION Corresponding Author
*Tel.: +49 221 478 4493. Fax: +49 221 478 7788. E-mail: franz.
[email protected].
’ ACKNOWLEDGMENT This study was supported by the Galactosemia Initiative Germany. We acknowledge the kind support by Christian Hoffmann, who provided high-quality electron microscopic pictures of exovesicle preparations. ’ REFERENCES (1) Bosch, A. Classical galactosaemia revisited. J. Inherited Metab. Dis. 2006, 29 (4), 516–525. (2) Berry, G. T.; Elsas, L. J. Introduction to the Maastricht workshop: Lessons from the past and new directions in galactosemia. J. Inherited Metab. Dis. 2011, 34 (2), 249–255. (3) Fridovich-Keil, J. L.; Gubbels, C. S.; Spencer, J. B.; Sanders, R. D.; Land, J. A.; Rubio-Gozalbo, E. Ovarian function in girls and women with GALT-deficiency galactosemia. J. Inherited Metab. Dis. 2011, 34 (2), 357–366. (4) Charlwood, J.; Clayton, P.; Keir, G.; Mian, N.; Winchester, B. Defective galactosylation of serum transferrin in galactosemia. Glycobiology 1998, 8 (4), 351–357. (5) Sturiale, L.; Barone, R.; Fiumara, A.; et al. Hypoglycosylation with increased fucosylation and branching of serum transferrin N-glycans in untreated galactosemia. Glycobiology 2005, 15 (12), 1268–1276. (6) Gubbels, C. S.; Thomas, C. M.; Wodzig, W. K.; et al. FSH isoform pattern in classic galactosemia. J. Inherited Metab. Dis. 2011, 34 (2), 387–390. (7) Potter, B. A.; Hughey, R. P.; Weisz, O. A. Role of N- and O-glycans in polarized biosynthetic sorting. Am. J. Physiol.: Cell Physiol. 2006, 290 (1), C1–C10. (8) Stoffel, W.; Bosio, A. Myelin glycolipids and their functions. Curr. Opin. Neurobiol. 1997, 7 (5), 654–661. (9) Pisitkun, T.; Shen, R. F.; Knepper, M. A. Identification and proteomic profiling of exosomes in human urine. Proc. Natl. Acad. Sci. U. S. A. 2004, 101 (36), 13368–13373. (10) Moon, P. G.; You, S.; Lee, J. E.; Hwang, D.; Baek, M. C. Urinary exosomes and proteomics. Mass Spectrom. Rev. 2011No. DOI: 10.1002/ mas.20319. (11) Simpson, R. J.; Lim, J. W.; Moritz, R. L.; Mathivanan, S. Exosomes: Proteomic insights and diagnostic potential. Expert Rev. Proteomics 2009, 6 (3), 267–283. (12) Rindler, M. J.; Naik, S. S.; Li, N.; Hoops, T. C.; Peraldi, M. N. Uromodulin (Tamm Horsfall glycoprotein/uromucoid) is a phosphatidylinositol-linked membrane protein. J. Biol. Chem. 1990, 265 (34), 20784– 20789. (13) Serafini-Cessi, F.; Bellabarba, G.; Malagolini, N.; Dall’Olio, F. Rapid isolation of Tamm Horsfall glycoprotein (uromodulin) from human urine. J. Immunol. Methods 1998, 120, 185–189. (14) Albert, T. K.; Laubinger, W.; Muller, S.; et al. Human intestinal TFF3 forms disulfide-linked heteromers with the mucus-associated
’ CONCLUSION It was demonstrated that a dysglycosylation of cellular membrane-bound proteins occurs in galactosemic subjects adhering to a galactose-restricted diet. Thus, it seems that aberrant glycosylation is a consequence of (the relatively high amounts of) endogenously synthesized galactose. The shift in expression from preferentially high-mannose-type to complex-type glycans reflects an increased processing rate in early Golgi compartments comprising three steps: the trimming by α-mannosidases I and II, the GlcNAc transfer by βGlcNAc-transferases GlcNAc-T I and -T II, and the galactosylation by the β4Gal-T. The complexity of the entire trimming and reglycosylation process makes it unlikely that only the cosubstrate levels of UDP-Gal, which may be altered in galactosemic patients, are the primary and direct cause of these dramatic changes. Other specific or global alterations on the proteomic level will have to be identified by differential proteomics. Severe cellular pathomechanistic effects can be expected to result from the described N-glycoprofile changes. As N-linked glycans form signals in the sorting for targeted trafficking of membranous glycoproteins, profound effects on the (apical) membrane localization and function of glycosylated receptors in lipid rafts can be expected. In accordance with the abovereported findings and their hypothetical implications, we are currently performing a differential proteomic study based on the GALT-deficient fibroblast cell model (grown under galactosefree and under galactose stress conditions) to reveal insight into changes of the lipid raft (glyco)proteomes. These analyses 915
dx.doi.org/10.1021/pr200711w |J. Proteome Res. 2012, 11, 906–916
Journal of Proteome Research
ARTICLE
FCGBP protein and is released by hydrogen sulfide. J. Proteome Res. 2010, 9 (6), 3108–3117. (15) Simpson, R. J.; Jensen, S. S.; Lim, J. W. Proteomic profiling of exosomes: Current perspectives. Proteomics 2008, 8 (19), 4083–4099. (16) Fernandez-Llama, P.; Khositseth, S.; Gonzales, P. A.; Star, R. A.; Pisitkun, T.; Knepper, M. A. Tamm Horsfall protein and urinary exosome isolation. Kidney Int. 2010, 77 (8), 736–742. (17) Domon, B.; Costello, C. E. A systematic nomenclature for carbohydrate fragmentations in FAB-MS/MS spectra of glycoconjugates. Glycoconjugate J. 1988, 5, 397–409. (18) Staubach, S.; Hanisch, F. G. Lipid rafts: Signaling and sorting platforms of cells and their roles in cancer. Expert Rev. Proteomics 2011, 8 (2), 263–277. (19) Slepak, T. I.; Tang, M.; Slepak, V. Z.; Lai, K. Involvement of endoplasmic reticulum stress in a novel Classic Galactosemia model. Mol. Genet. Metab. 2007, 92 (1 2), 78–87. (20) Zhou, H.; Pisitkun, T.; Aponte, A.; et al. Exosomal Fetuin-A identified by proteomics: A novel urinary biomarker for detecting acute kidney injury. Kidney Int. 2006, 70 (10), 1847–1857. (21) Moon, P. G.; Lee, J. E.; You, S.; et al. Proteomic analysis of urinary exosomes from patients of early IgA nephropathy and thin basement membrane nephropathy. Proteomics 2011, 11 (12), 2459– 2475. (22) van Rooijen, J. J.; Voskamp, A. F.; Kamerling, J. P.; Vliegenthart, J. F. Glycosylation sites and site-specific glycosylation in human Tamm Horsfall glycoprotein. Glycobiology 1999, 9 (1), 21–30. (23) Parry, S.; Hanisch, F. G.; Leir, S. H.; et al. N-Glycosylation of the MUC1 mucin in epithelial cells and secretions. Glycobiology 2006, 16 (7), 623–634. (24) Litvinov, S. V.; Hilkens, J. The epithelial sialomucin, episialin, is sialylated during recycling. J. Biol. Chem. 1993, 268 (28), 21364–21371.
916
dx.doi.org/10.1021/pr200711w |J. Proteome Res. 2012, 11, 906–916
