A Novel N-Tetrasaccharide in Patients with Congenital Disorders of ...

Papers in Press. Published October 1, 2015 as doi:10.1373/clinchem.2015.243279 The latest version is at http://hwmaint.clinchem.org/cgi/doi/10.1373/clinchem.2015.243279 Clinical Chemistry 62:1 000 – 000 (2015)

Pediatric Clinical Chemistry

A Novel N-Tetrasaccharide in Patients with Congenital Disorders of Glycosylation Including AsparagineLinked Glycosylation Protein 1, Phosphomannomutase 2, and Phosphomannose Isomerase Deficiencies Wenyue Zhang,1† Philip M. James,2,3† Bobby G. Ng,4† Xueli Li,5† Baoyun Xia,1 Jiang Rong,1 Ghazia Asif,1 Kimiyo Raymond,6 Melanie A. Jones,1 Madhuri Hegde,1 Tongzhong Ju,5 Richard D. Cummings,7 Katie Clarkson,6 Tim Wood,8 Cornelius F. Boerkoel,9 Hudson H. Freeze,4 and Miao He5,10*

BACKGROUND: Primary deficiencies in mannosylation of N-glycans are seen in a majority of patients with congenital disorders of glycosylation (CDG). We report the discovery of a series of novel N-glycans in sera, plasma, and cultured skin fibroblasts from patients with CDG having deficient mannosylation. METHOD:

We used LC-MS/MS and MALDI-TOF-MS analysis to identify and quantify a novel N-linked tetrasccharide linked to the protein core, an N-tetrasaccharide (Neu5Ac␣2,6Gal␤1,4-GlcNAc␤1,4GlcNAc) in plasma, serum glycoproteins, and a fibroblast lysate from patients with CDG caused by ALG1 (ALG1, chitobiosyldiphosphodolichol ␤-mannosyltransferase), PMM2 (phosphomannomutase 2), and MPI (mannose phosphate isomerase). RESULTS:

Glycoproteins in sera, plasma, or cell lysate from ALG1-CDG, PMM2-CDG, and MPI-CDG patients had substantially more N-tetrasaccharide than unaffected controls. We observed a ⬎80% decline in relative concentrations of the N-tetrasaccharide in MPICDG plasma after mannose therapy in 1 patient and in ALG1-CDG fibroblasts in vitro supplemented with mannose.

CONCLUSIONS:

This novel N-tetrasaccharide could serve as a diagnostic marker of ALG1-, PMM2-, or MPI-CDG for screening of these 3 common CDG subtypes that comprise ⬎70% of CDG type I patients. Its quantification by LC-MS/MS may be useful for monitoring ther-

1

Department of Human Genetics and 7 Department of Biochemistry, Emory University, Atlanta, GA; 2 Division of Genetics, Department of Medicine, Children’s Hospital Boston, Boston, MA; 3 Department of Pediatrics, Harvard Medical School, Boston, MA; 4 Human Genetics Program, Sanford Children’s Health Research Center, Sanford-Burnham Medical Research Institute, La Jolla, CA; 5 Palmieri Metabolic Disease Laboratory, Children’s Hospital of Philadelphia, Philadelphia, PA; 6 Mayo Biochemical Genetics Laboratory, Mayo Medical School, Rochester, MN; 8 Greenwood Genetics Center, Greenwood, SC; 9 Department of Medical Genetics, University of British Columbia, Vancouver, BC, Canada; 10 Department of Pathology and Laboratory of Medicine, University of Pennsylvania, Philadelphia, PA.

apeutic efficacy of mannose. The discovery of these small N-glycans also indicates the presence of an alternative pathway in N-glycosylation not recognized previously, but its biological significance remains to be studied. © 2015 American Association for Clinical Chemistry

Congenital disorders of glycosylation (CDGs)11 are a group of diseases with highly variable phenotypes and inconsistent clinical features. Since the first description of a CDG in 1980, approximately 100 disorders have been identified (1 ). Most are defects in protein glycosylation, although an increasing number are defects of glycolipid or proteoglycan biosynthesis. Protein glycosylation is the process of adding and remodeling glycans to proteins on either the amide group of asparagine residue (N-glycan) or the hydroxyl group of a serine or threonine residue (O-glycan). Both the structure and function of N- and O-glycans in humans are highly diverse. Excluding O-GlcNAcylated proteins, ⱖ85% of human proteins with a secretory signal sequence are N- or O-glycosylated (2 ). Transferrin is a serum glycoprotein that has 2 N-glycosylation sites, and each glycan contains ⱕ2 terminal, negatively charged sialic acids. The glycoforms of transferrin are widely used as a biomarker of N-glycosylation defects. Alterations in the number or structure of transferrin glycoforms can be detected by isoelectric focusing, HPLC, capillary electrophoresis, or

* Address correspondence to this author at: 3401 Civic Center Blvd, Philadelphia, PA 19103. Fax 215-590-1998; e-mail [email protected]. † W. Zhang, P.M. James, B.G. Ng, and X. Li contributed equally to this study. Received May 30, 2015; accepted August 25, 2015. Previously published online at DOI: 10.1373/clinchem.2015.243279 © 2015 American Association for Clinical Chemistry 11 Nonstandard abbreviations: CDG, congenital disorder of glycosylation; ESI, electrospray ionization; ALG1, asparagine-linked glycosylation protein 1; CH3I, iodomethane; DHB, 2,5-dihydroxybenzoic acid; MRM, multiple reaction monitoring; 2-AB, 2-aminobenzamide.

1

Copyright (C) 2015 by The American Association for Clinical Chemistry

electrospray ionization (ESI)-LC-MS (3 ). Although analysis of carbohydrate-deficient transferrin reliably identifies the majority of the N-linked hypoglycosylation disorders, it does not distinguish CDG subtypes, which requires gene sequencing and functional confirmation. Here we show that application of high-throughput MALDI-TOF-MS analysis to total serum N-glycans identifies small N-glycans devoid of mannose. Measuring these N-glycans in plasma and cultured fibroblast lysates distinguishes among unaffected controls and patients with ALG1-CDG, PMM2-CDG, or MPI-CDG, which are the CDG types caused by ALG1 [ALG1 (asparagine-linked glycosylation protein 1), chitobiosyldiphosphodolichol ␤-mannosyltransferase],12 PMM2 (phosphomannomutase 2), and MPI (mannose phosphate isomerase), respectively. Unlike transferrin, a glycoprotein that is predominantly synthesized in the liver, total N-glycans can be measured from samples other than blood, such as urine and fibroblast lysates. This makes N-glycan profiles ideal for biochemical diagnosis, in vitro studies of CDG pathogenesis, and monitoring effectiveness of treatment. Additionally, such analyses can lead to the identification of novel N-glycans with diagnostic and therapeutic significance. Materials and Methods Iodomethane (CH3I), DMSO, 2,5-dihydroxybenzoic acid (DHB), sodium hydroxide, trifluoroacetic acid, sodium borohydrate, and sodium acetate were obtained from Sigma-Aldrich; PNGase F, including denaturing buffer, digestion buffer, and NP-40, from New England Biolabs; Extra-Clean SPE Carbograph columns from Grace Davison Discovery Sciences; Sep-Pak Vac 3cc C18 cartridges from Waters; p-lacto-N-hexaose from V-Laboratories; and methanol, chloroform, and acetonitrile from Fisher Scientific. PREPARATION AND PERMETHYLATION OF PLASMA OR CELLULAR PROTEIN LYSATES

As previously described (4 ), we prepared N-glycan samples by PNGase F digestion at 37 °C for 16 h after the total plasma or serum proteins were denatured by incubation at 100 °C for 10 min. We used 20 ␮L serum or plasma for N-glycan preparation. For cellular N-linked glycan preparation, 100 ␮g total cellular protein from total cell lysate was denatured and precipitated with propanol (1:2 vol/vol) before PNGaseF digestion. The re-

12

2

Human genes: ALG1, ALG1, chitobiosyldiphosphodolichol ␤-mannosyltransferase; PMM2, phosphomannomutase 2; MPI, mannose phosphate isomerase; ALG2, ALG2, ␣-1,3/1,6-mannosyltransferase; ALG11, ALG11, ␣-1,2-mannosyltransferase.

Clinical Chemistry 62:1 (2015)

leased N-linked glycans were purified through a Sep-Pak C18 column, to which they did not bind, followed by binding and release from a carbograph column with acetonitrile as the eluant. All the purified glycans were lyophilized overnight. For permethylation, we crushed 4 NaOH pellets (approximately 375 mg) in 10 mL anhydrous DMSO and added 0.5 mL of this slurry and 0.2 mL CH3I to the dried glycans. The mixture was shaken vigorously for 1 h followed by 5 sequential chloroform/water (600 ␮L/200 ␮L) extractions. The chloroform fractions were pooled and dried for 30 min under nitrogen. The permethylated N-glycans were resuspended in 50 ␮L of 50% methanol and further purified through a C18 Stage Tip (Thermo Scientific), to which they did bind, and were eluted with acetonitrile as previously described (4 ). ANALYSIS OF N-GLYCAN PROFILES BY MALDI-TOF-MS

Each permethylated N-glycan sample (0.5 ␮L) was spotted onto the MALDI plate, and 0.5 ␮L DHB (11 g/L in 50% methanol with 1 mmol/L sodium acetate) was added as a matrix solution. Each spot was then air-dried for 10 min. The data were generated with a MALDITOF-MS 4800 plus (Applied Biosystems) in positive reflector mode with the laser power set at 4880 and the digitizer set at 0.79. Each sample was run 3 times. We selected 20 specific N-glycans for semiquantitative analysis and normalization of intensities to the sum of all the peaks. The ratios of the most abundant glycan (m/z 2792) and the internal standard (m/z 1357 or 926) were recorded to monitor the yield of glycan. We used 4000 series Explorer software (Applied Biosystems) to collect data and integrate peak intensities. All the glycans were identified as singly charged [M⫹Na]⫹. QUANTIFICATION OF N-LINKED GLYCANS BY LC-MS/MS

We analyzed permethylated glycans on an ABSIEX QTRAP 5500 mass spectrometer equipped with a Shimadzu Prominence UFLC system that consisted of 2 LC-20AD solvent delivery units, a DGU-20A3 degassing unit, a SIL-20AC HT autosampler, a CTO-20A column oven, and a CMB-20A system controller. Permethylated N-glycans were separated on a Thermo Scientific Hypersil™ GOLD HPLC Column, 3-␮m particle size, 30 ⫻ 2.1 mm, at a flow rate of 0.25 mL/min. The column was maintained at 35 °C in a column oven. The gradient was started with 55% mobile phase A (2% acetonitrile and 0.1% formic acid in ultrapure water) and 45% mobile phase B (98% acetonitrile and 0.1% formic acid); mobile phase B was increased to 60% at 5 min and to 100% at 5.5 min and reverted to 45% at 13 min to equilibrate for 3 min. The mass spectrometer was operated in positive ESI mode. Nitrogen from an Infinity 1031 Nitrogen generator (PEAK) was used as the curtain gas (25 ␺), ion

Novel N-Tetrasccharide in Congenital Glycosylation Disorders

source gas 1 (45 ␺) and gas 2 (55 ␺), and collision gas (Medium). The ion spray voltage of the source was 5500, and the temperature was set at 600°C. Multiple reaction monitoring (MRM) for each targeted glycan was optimized with infusion of permethylated N-glycans in 80% acetonitrile containing 0.1% formic acid. The MRM transition used for the sialylated tetrasaccharide was 1102/825, and 1149/851 for the pentasaccharide Man3GlcNAc2. The internal standard used for quantification of N-glycans by LC-MS/MS was raffinose with a MRM transition of 681/463. The quantification of N-tetrasaccharide (see below) and disialo-glycan (VLaboratories) was achieved by LC-MS/MS with external standards. SYNTHESIS, PURIFICATION, AND QUANTIFICATION OF N-TETRASACCHARIDE

Chitobiose, a disaccharide GlcNAc␤1,4GlcNAc (SigmaAldrich), was galactosylated with a ␤1,4-galactosyltransferase kit (Sigma) following the manufacturer’s instructions. Briefly, 379 mg Trizma hydrochloride buffer, pH 7.4, and 198 mg MnCl2 were dissolved in 50 mL MilliQ water. To 1.0 mL of the reaction buffer, we added 10 mmol/L chitobiose, 7.3 mg UDP-galactose, 1 mg BSA, 1 ␮L alkaline phosphatase (10 U/mL), and 50 mU ␤1,4galactosyltransferase. The reaction mixture was incubated at 37 °C overnight, and the product was purified on a Sep-Pak C18 column followed by a carbograph column as described above. The addition of sialic acid to Gal␤1,4GlcNAc␤1,4GlcNAc was achieved with ␣2,3sialyltransferase (Sigma) or ␣2,6-sialyltransferase (Sigma). Enzymatic reactions were performed at 37 °C overnight in Tris-HCl buffer (100 mmol/L, pH 8.0) containing 5 mmol/L CMP-sialic acid (CMP-Neu5Ac), 2 U ␣2,3sialyltransferase, or 5 U ␣2,6-sialyltransferase. The product tetrasaccharides from the separate reactions with sialyltransferases (Neu5Ac␣2,3Gal␤1,4GlcNAc␤1,4GlcNAc and Neu5Ac␣2,6Gal␤1,4GlcNAc␤1,4GlcNAc) were purified on a Sep-Pak C18 column and then on a carbograph column. We determined the concentration of the resulting tetrasaccharide by 2-aminobenzamide (2-AB) labeling of the reducing end with a GlycoProfile™ 2-AB labeling kit (Sigma) or aniline after quantitative measurement by HPLC with a fluorescence detector or LC-MS/MS, respectively, as previously described (5, 6 ). We used MRM transition 937/ 671 for the tetrasaccharide labeled with aniline. We used 2-AB labeling for measuring the concentration of synthesized standard and aniline labeling for optimal separation of 2,3- and 2,6-sialylated tetrasaccharide on LC-MS/MS. Results We identified a set of small N-glycans in the plasma of a patient with pathogenic mutations in ALG1, defined as those released by PNGase F, at m/z 559, 763, and 1124

(Fig. 1, B and C, and Supplemental Fig. 1B, which accompanies the online version of this article at http:// www.clinchem.org/content/vol62/issue1); these were absent in unaffected control plasma (Fig. 1A, online Supplemental Fig. 1A). From the mass of each of these glycans and the fragmentation pattern of the tetrasaccharide (Fig. 1D), we predicted them to be chitobiose (GlcNAc2) (m/z 559), galactosylated chitobiose (Gal1GlcNAc2) (m/z 763), and a sialylated tetrasaccharide (Sial1Gal1GlcNAc2) (m/z 1124), respectively. Because these glycans were lacking in the purified free glycans (online Supplemental Fig. 1C) or the O-glycans released by ␤-elimination of total glycoproteins (data not shown) from the same plasma sample, we concluded that they were N-glycans specifically released by PNGase F digestion. We designated the tetrasaccharide as the N-tetrasaccharide. We also detected small amounts of “non”glycosylated transferrin monomer with the Ntetrasaccharide at observed mass 76 017 (predicted mass 76 016) and “mono”-glycosylated transferrin with an N-tetrasaccharide at observed mass 78 215 (predicted mass 78 222) (Fig. 1E). These glycoforms of transferrin represented only 2%–5% of the signal of the major transferrin species in the ESI-MS profile of ALG1-CDG transferrin. However, ESI-MS analysis of transferrin was an indirect measurement through convolution of the multiple charge envelopes, and the majority of the convolution software used for ESI-MS was not designed for quantitative interpretation of peaks with low abundance compared to the major peaks. Analysis of PNGase F–released material from purified transferrin of the ALG1-CDG patient sample also identified the N-tetrasaccharide (Fig. 1G). The tetrasaccharide was not observed in unaffected controls (Fig. 1F). We hypothesized that the presence of this small N-tetrasaccharide was due to reduced mannosylation of lipid-linked chitobiose in the endoplasmic reticulum. For example, ALG1-deficient yeast accumulates dolichol-linked chitobiose (GlcNAc␤1,4GlcNAc-P-P-Dol). This intermediate can be transferred to nascent proteins, producing N-linked chitobiose (7 ). Because the fragmentation pattern of the N-tetrasaccharide (Fig. 1D) predicted it to have the composition Sial1Hexose1HexNAc2, we hypothesized that N-linked chitobiose-modified proteins reach the Golgi apparatus and act as acceptors for Golgi ␤1,4galactosyltransferase, which adds galactose, and an ␣2,3and/or ␣2,6-sialyltransferase, which adds a sialic acid. We compared the unknown N-tetrasaccharide to 2 synthetic tetrasaccharide standards, Neu5Ac␣2,3Gal␤1,4GlcNAc␤1, 4GlcNAc and Neu5Ac␣2,6Gal␤1,4GlcNAc␤1,4GlcNAc. The results showed that Neu5Ac␣2,6Gal␤1, 4GlcNAc␤1,4GlcNAc had the same retention time and mass fragmentation pattern as the unknown N-tetrasaccharide (Fig. 2, A and B). Thus, we conClinical Chemistry 62:1 (2015) 3

B

559

1124 2431 2227

3603

4000

1800

Y2

B1 B2

B3

B1

Y3

TOF/TOF Fragment

Y1

847

Intensity (cps)

900

Y3

3603

2966

3000

749

2792

1580

2000

398

0 899

2227

1784 1982

30

1621

1094

926

I.S

D PMM2-CDG

2431

60

1376

1124 1171

120

0 1000

4000

472

3000

ALG1-CDG

1000

I.S 1784 1982

2966

2227

2000

500

30

3603

I.S

30

1124

2431

60

1376 1580

60

90

Intensity (%)

763

90

0 1000

C

LMw

Control

1376 1580 1784 1982

Intensity (%)

90

2792

120

2966

120 2792

A

B3

0

2145

2768

3390

4013

300

600

900

1200

76017

75800

76300

80

Control 1103

ALG1-CDG Transferrin

80 Intensity (%)

F

77351

1089

E

79555

75145 1059

40

40

1330

77648 78215

75440 76017

0

1330

1059

0 1601

2204

1059

40

0 999

80

2807

3410

4013

2807

3410

4013

PMM2-CDG

1293

1293

ALG1-CDG

1089 1124

H

2204

1330

1124

40

1124

Intensity (%)

80

1124

1089

G

1601

1249

999

Mass (Da)

999

1601

2204

2807

3410

4013

Fig. 1. Plasma N-glycan and transferrin N-glycan profiles from ALG1-CDG and PMM2-CDG. (A), Control patient. (B), Patient with ALG1-CDG (inset shows low mass range 500 –1200 Da). (C), Patient with PMM2-CDG (inset shows mass range 1100 –1185). (D), Fragmentation of the N-tetrasaccharide by MALDI-TOF/MS analysis. (E), Transferrin profile of a patient with ALG1-CDG (inset shows mass range 75 800 –76 300 Da) by LC-ESI-MS. (F), N-glycan profile of purified transferrin from plasma of a control patient compared with those of ALG1-CDG (G) and PMM2-CDG (H) patients. Insets of F–H show mass range 1059 –1330 Da. All N-glycans shown are permethylated. Profiles here are representative of 20 controls, 10 patients with ALG1-CDG, and 20 patients with PMM2-CDG.

4



B

A

Intensity

7000

C 25 000

1200

10.48

12 500

600

3500

10.48

8.75

5.82 8.75

10.48 8.75

5.82 0

0 0

5

15

10

0

0

5

Retention time (min)

10

15

0

5


10

15


D Intensity

1.20×10+06

0.80×10+06

0.40×10+06

0.00×10+00 0

5

10

15

20

(µM)

E

F ratio

60 CDG-Ik 45

12

/

(µM)

16

CDG-Ia Control

30

8 15

4

0

0 Control

Ia

15 1.2

Ik

Fig. 2. Quantification of N-tetrasaccharide in PMM2-CDG, Ib, and Ik plasma with synthesized tetrasaccharide standard by LC-MS/MS. Chromatograms of synthesized Neu5Ac␣2,6Gal␤1,4GlcNAc␤1,4GlcNAc (A) and Neu5Ac␣2,3Gal␤1,4GlcNAc␤1,4GlcNAc (B) labeled by aniline via reductive amination and measured by LC-MS/MS (MRM transition 937/671). C) Plasma (20 μL) from a patient with ALG1-CDG treated with PNGase F, and total released N-glycans labeled with aniline and measured by the same LC-MS/MS method. (D), Calibration curve of tetrasaccharide with synthesized standard. (E), Concentrations of N-tetrasaccharide in 20 controls (gray), 20 patients with PMM2-CDG (green), and 10 patients with ALG1-CDG (red). (F), Ratio between N-tetrasaccharide and Man3GlcNAc2 in 20 controls, 20 patients with PMM2-CDG, and 10 patients with putative or proven ALG1-CDG. Upper limit of control and PMM2-CDG (1.2) and lower limit of ALG1-CDG (15) are shown with dotted lines.

cluded that the N-tetrasaccharide had the sequence Neu5Ac␣2,6Gal␤1,4GlcNAc␤1,4GlcNAc. Because a mannosylation deficiency also occurs in patients with PMM2-CDG or MPI-CDG owing to deficiency in GDP-mannose, we hypothesized that the

novel N-tetrasaccharide also occurs in these 2 subtypes of CDG. To explore this possibility, we analyzed the N-glycan profiles from total plasma glycoproteins of a patient with PMM2-CDG and a patient with MPICDG before mannose treatment. As shown in Fig. 1C Clinical Chemistry 62:1 (2015) 5

A

B 1172 1142

1124

1180

C

2000

3602

2967

2605 2792

2396

2967

3602

0 1000

4000

3000

2000

3243

2192

1988

1345

PMM2-CDG

30 3243

2605

2792

2396

1784

1107 60

30

0 1000

1185

Control

2192

1988

60

1580

1089 1172

1142

1784

1580

1089 1172

90

1100

1345

Intensity (%)

90

120

1172

120

3000

4000

D

3410

4013

1.4E4

2792

2410

1180

3602

3243

2792

2967

2431

1866

1580

1172

1118 1124 2396

2192 1988 2070

1580 1346

1145

MPI-CDG postmannose

60

3390

4013

Mass m/z

2966

3603

2431

2227

1635

1982

926

20

3603

2966

2768

Mannose treated

2807

80

1099 40 I.S.

2431

2227

1982

2145

1124 1172

MPI-CDG

1635

1784

899

100

1110

2205

1376

3602

2792

3243

F

1602

7088

2792

1099

0 999

4000

1580 1824

1376

1124 1172

20

I.S. 926

Intensity (%)

3000

2000

E

2967

2605

0 1000

1417

40

2396

2192

1988

1345 30

1752

60

60

1784

80

1180

ALG1-CDG

Untreated

1784

1784

1100

1124.7 1172

100

1142

1124

1172

1580

1124

90

1172

120

0 899

2145

2768

3390

4013

Mass m/z

Fig. 3. Effect of mannose treatment on ALG1-CDG fibroblast N-glycan and plasma N-glycan profiles from a patient with MPI-CDG. (A), Cultured fibroblast N-glycan profile by MALDI-TOF in control cells compared with PMM2-CDG (B) and ALG1-CDG (C) cells. (D), Effect of mannose treatment of an ALG1-CDG fibroblast. Plasma N-glycan profiles in MPI-CDG before (E) and after (F) oral mannose treatment. All N-glycans shown are permethylated.

(PMM2-CDG, plasma N-glycan), Fig. 1H (PMM2CDG, purified transferrin N-glycan), and Fig. 3E (MPICDG, plasma N-glycan), the N-tetrasaccharide was present in patients with PMM2-CDG and MPI-CDG, although its concentration was lower than in patients 6


with ALG1-CDG. In addition, we found that N-linked Man3GlcNAc2 at m/z 1171 and Man4GlcNAc2 at m/z 1375 were increased in patients with PMM2-CDG and MPI-CDG, but not in patients with ALG1-CDG. The plasma N-glycan profiles for patients with PMM2-CDG


Table 1. Biallelic ALG1 mutations in 10 patients.a Patient

a

Mutation 1

Mutation 2

P1

c.IVS11 + 1G>A

c.1079C>T (p.A360V)

P2

c.841G>T (p.V281F)

c.1101C>G (p.H367Q)

P3

c.866 A>G (p.D289G)

c.866A>G (p.D289G)

P4

c.212C>T (p.S71F)

c.221A>T (p.H74L)

P5

c.773C>T (p.S258L)

c.773C>T (p.S258L)

P6

c.293C>T (p.P98L)

c.773C>T (p.S258L)

P7

c.212C>T (p.71S>F)

c.1079C>T (p.A360V)

P8

c.149A>G (p.Q50R)

c.773C>T (p.S258L)

P9

c.866A>G (p.D289G)

c.1312C >T (p.R438W)

P10

c.773C>T (p.S258L)

c.773C>T (p.S258L)

Bold indicates proven mutations; others are putative.

or MPI-CDG were indistinguishable (Fig. 1C and Fig. 3E). Interestingly, analyzing N-glycan profiles of purified transferrin from a patient with PMM2-CDG detected the N-tetrasaccharide; however, increased Man3GlcNAc2 or Man4GlcNAc2 in PMM2-CDG plasma N-glycans was not detected in its transferrin N-glycan (Fig. 1H), suggesting that Man3GlcNAc2 and Man4GlcNAc2 were more abundant in other plasma glycoproteins such as immunoglobulins. We next developed a quantitative assay for the N-tetrasaccharide and for Man3GlcNAc2 using LCMS/MS (Fig. 2, D–F). Measurement of the plasma N-tetrasaccharide and Man3GlcNAc2 in 10 patients with putative or confirmed ALG1-CDG (mutations and putative mutations from patients with ALG1-CDG are shown in Table 1), 20 patients with PMM2-CDG, 1 patient with MPI-CDG, and 20 unaffected controls showed that the N-tetrasaccharide was undetectable in all the controls and present in all the patients with PMM2CDG or ALG1-CDG (Fig. 2E). The concentration of N-tetrasaccharide was the highest for patients with ALG1-CDG (Fig. 2E). In addition, the relative ratio between the tetrasaccharide and Man3GlcNAc2 was increased in patients with ALG1-CDG and within the reference range or low in patients with PMM2-CDG or

MPI-CDG (Fig. 2F). These 2 measures successfully differentiated ALG1-CDG from PMM2-CDG, MPICDG, and unaffected controls and also differentiated PMM2-CDG and MPI-CDG from unaffected controls. The N-tetrasaccharide was undetectable in the plasma of patients with ALG12-CDG (CDG-Ig), DPAGT-CDG (CDG-Ij), ALG8-CDG (CDG-Ih), ALG6-CDG (CDG-Ic), Man1B1-CDG and DDOST-CDG (CDGIr), or mixed-type CDGs such as PGM1-CDG. We also measured the relative quantity of the N-tetrasaccharide by LC-MS/MS after it was released from purified transferrin. The molar ratio between tetrasaccharide at m/z 1124 and transferrin monomers was 0.24 in ALG1-CDG and 0.02 in PMM2-CDG, vs a ratio of 2 between disialylated N-glycan at m/z 2792 and transferrin monomers in an unaffected control (Table 2). Quantification by LC-MS/MS showed that in purified ALG1-CDG transferrin, this tetrasaccharide was 13.8% of the major biantennary glycans, in contrast to 2% for PMM2-CDG transferrin. In comparison, the MALDITOF profile of N-glycans released from the same ALG1CDG and PMM2-CDG samples showed that the tetrasaccharide comprised 34% and 5% of the biantennary glycans (Fig. 1, B and C). MALDI-TOF also identified the N-tetrasaccharide m/z 1124 in the N-glycan profiles of lysates from cultured fibroblasts of 5 patients with ALG1-CDG and 5 patients with PMM2-CDG, but not control fibroblasts (Fig. 3, A–C). Because fibroblasts do not produce transferrin, the N-tetrasaccharide detected must be present in other cellular glycoproteins. Identification of the N-tetrasaccharide lacking mannose at m/z 1124 in patients with ALG1-CDG, PMM2-CDG, and MPI-CDG suggested that this small N-glycan reflected a deficiency in mannosylation. To test this, we added 500 ␮mol/L mannose to the culture medium of fibroblasts from a patient with ALG1-CDG and an unaffected control. After 16 h, there was an 86% reduction of N-tetrasaccharide in ALG1-CDG cells, along with increased amounts of high-mannose and sialylated N-glycans (Fig. 3D). Addition of 100 ␮mol/L mannose did not produce detectable changes in the profiles of N-glycans, and addition of 200 ␮mol/L mannose pro-

Table 2. Quantitative measurement of N-tetrasaccharide and its transferrin monomer ratio.a N-tetrasaccharide m/z 1124, pmol

Transferrin, pmol

Tetratransferrin

Disialotransferrin

Reference

0.00

684

336

0.00

2.03

PMM2-CDG

5.96

262

242

0.02

1.08

420

242

0.24

1.73

ALG1-CDG a

Disialo-glycan m/z 2792, pmol

57.1

Values are the mean of 2 measurements. The concentrations of N-tetrasaccharide and disialo-glycan were measured by LC-MS/MS using external standards.

Clinical Chemistry 62:1 (2015) 7

duced minimal normalization (data not shown). Given these results, we measured plasma N-glycan from a patient with MPI-CDG before and after the patient’s diet had been supplemented with mannose for 6 months and observed reduced concentrations of the tetrasaccharide m/z 1124 in plasma (Fig. 3, E and F). The carbohydratedeficient transferrin concentrations in this patient also improved markedly after the mannose therapy (data not shown). Interestingly, the increase of Man3GlcNAc2 and Man4GlcNAc2 did not appear to improve. It is possible that mannose supplementation rescued GDP-mannose deficiency in the liver and normalized the glycosylation of transferrin in the patient, but may have had less effect on the GDP-mannose deficiency in other tissues. Discussion Although the N-tetrasaccharide was a minor glycan species on transferrin of patients with ALG1-CDG and PMM2-CDG (Table 2), on the basis of these preliminary studies, it is a potential diagnostic biochemical marker. However, we caution that the MALDI-TOF signal has a very narrow dynamic range and should not be used for quantitative interpretation. Quantification by LCMS/MS should be considered if this tetrasaccharide is used as a biomarker to monitor treatment or disease progression. Small N-glycans have been recently discovered but not well studied. N-linked chitobiose has been observed in primary cancer cells cultured under hypoglycemic conditions as well as in alg1-deficient yeast (8 ), but to the best of our knowledge, this galactosylated or sialylated form of N-linked chitobiose has not been reported previously. The discovery of these small N-glycans points to a previously unrecognized Golgi-associated secretory pathway for N-linked protein glycosylation (Fig. 4). It is known that terminal sialylation of N-linked glycans is important for the survival of hosts during bacterial infection (8 ). The presence of this alternative pathway could be an adaptive way for cells to maintain terminal sialylation under the circumstance of starvation (8 ). Intriguingly, the pentasaccharide Man3GlcNAc2 was increased in all 20 patients with PMM2-CDG and 1 patient with MPI-CDG. Specifically, our data show that the combination of increased N-linked tetrasaccharide and Man3GlcNAc2 excludes a diagnosis of ALG1-CDG and instead supports a diagnosis of PMM2-CDG or MPI-CDG. This arises because the deficiency of GDPmannose in PMM2-CDG or MPI-CDG potentially affects every step of mannosylation in the endoplasmic reticulum (Fig. 4). Mannosylation reactions preceding Man5GlcNAc2 on the dolichol precursor may be more directly affected, since GDP-mannose serves directly as the substrate for these steps. Thus, intermediate-sized glycans can accumulate in plasma N-glycans if the trun8


cated precursor is available. We did not observe any Man1GlcNAc2 or Man2GlcNAc2, which could indicate that the Km of GDP-mannose for the ␣1,3/1,6mannosyltransferase [encoded by ALG2 (ALG2, ␣-1,3/ 1,6-mannosyltransferase)] is lower than that of either ␤1,4-mannosyltransferase (encoded by ALG1) or ␣1,2 mannosyltransferase [encoded by ALG11 (ALG11, ␣-1,2-mannosyltransferase)]. In support of this, Man4GlcNAc2, which as the dolichol-linked precursor is another substrate of ALG11, is also increased in PMM2CDG and MPI-CDG (Figs. 1C and 3E). The possible alternative pathways in PMM2-CDG and MPI-CDG are described in Fig. 4. Our data show that increased concentrations of the N-tetrasaccharide, together with a concentration of Man3GlcNAc2 that is within the reference range or low, could be potentially diagnostic for ALG1-CDG, because in ALG1-CDG only the ␤1,4-mannosyltransferase is deficient and there is no increase of other downstream intermediates from downstream mannosylation steps in the endoplasmic reticulum. The ability to detect the N-tetrasaccharide and Man3GlcNAc2 by LC-MS/MS or MALDI-TOF-MS provides a reliable biochemical diagnostic screen for ALG1-CDG and PMM2-CDG or MPI-CDG that is important because it facilitates diagnosis of the 3 most common type I CDGs (10 ). N-glycan profiling by MALDI-TOF analysis is available as a routine clinical test and is often used as one of the first-tier screening tests in combination with transferrin analysis. Although whole-exome and whole-genome sequencing greatly facilitate diagnoses of rare genetic disorders, establishing a diagnosis for patients with CDG remains challenging because the majority of patients with CDG carry rare missense mutations of undefined functional significance (10 ). The fact that there are 14 pseudo genes of ALG1 makes molecular analysis of ALG1-CDG more difficult. The discovery of biochemical biomarkers should therefore greatly improve the efficiency of diagnosing these CDGs. Although transferrin glycoforms with this tetrasaccharide could be detected in some patients with ALG1-CDG by routine transferrin analysis with ESIMS, the concentrations detected are very low and are undetectable in PMM2-CDG or MPI-CDG. Other transferrin test methods such as isoelectric focusing or HPLC are less analytically sensitive than ESI-MS and likely will not detect these glycoforms. Therefore, the combination of routine transferrin and N-glycan analysis by mass spectrometric analysis of plasma or serum is a better first-tier screening strategy for N-glycosylation disorders than transferrin analysis alone. So far, carbohydrate-deficient transferrin has been the predominant biomarker for the diagnosis and treatment monitoring of patients with CDG. The majority of cells, however, do not express transferrin, and this has


Fig. 4. Proposed alternative glycosylation pathways in PMM2-CDG or MPI-CDG due to GDP-mannose deficiency. The dolichol-linked precursor for N-glycosylation is synthesized by addition of mannose from GDP-mannose (GDP-Man) synthesized in the cytoplasm. Patients with PMM2 and MPI have deficiencies indicated by the red X in the biosynthesis of GDP-Man. N-glycosylation of a glycoprotein is initiated in the lumen of the rough endoplasmic reticulum by transfer of a glycan precursor from its dolichol-linked precursor to Asn residues (-Asn-X-Ser/Thr) in a nascent polypeptide through ribosomal-directed synthesis. Three possibilities for abnormal glycosylation are shown in the upper panel, involving transfer of 2 N-acetylglucosamine(GlcNAc) residues, transferof Mannose3GlcNAc2 (Man3GlcNAc2), andtransfer of Man4GlcNAc2. After N-glycosylation in the endoplasmic reticulum, glycoproteins move to the Golgi apparatus, where they are subject to further modification by addition of galactose (Gal) and sialic acid (Neu5Ac) by appropriate glycosyltransferases. After completion of glycosylation, the glycoproteins may move to the plasma membrane and be secreted. Orange arrows indicate upregulation in the production of these unusually glycosylated forms in the plasma, serum, and fibroblasts of patients with PMM2-CDG or MPI-CDG.

Clinical Chemistry 62:1 (2015) 9

hindered the use of cellular models to develop treatments for CDG. Although alternative cellular expression systems have been developed (11 ), they have been technically difficult to implement and maintain. Our observation that the N-tetrasaccharide is easily measurable in fibroblast lysates from patients with ALG1-CDG or PMM2-CDG therefore highlights the potential for identifying biomarkers usable in cellular models. Consistent with the diagnostic value of this biomarker, cell lines carrying certain ALG1-CDG mutations demonstrate a response to mannose supplementation. The applicability of mannose treatment for patients with ALG1-CDG remains unknown because not all the ALG1-CDG mutant cell lines respond to mannose supplementation. For instance, cells homozygous for S258L do not respond to mannose supplementation (data not shown), and the concentration of mannose used in vitro may be greater than the concentration that can be achieved in vivo (12 ). Because different glycoproteins carry different types of complex N-glycans, monitoring changes of different N-glycan biomarkers could also help us understand the delivery of mannose therapy to different organs in patients with CDG. Our finding that only N-tetrasaccharide was reduced after mannose therapy without reduction in Man3GlcNAc2 or Man4GlcNAc2 indicates the potential limitation of oral mannose therapy in patients with CDG, and that future studies are warranted to investigate the effects of mannose therapy on glycoproteins in different human tissues. In summary, we have identified a novel sialylated N-tetrasaccharide that completely lacks mannose and have shown that it is a useful biomarker for the detection of deficient mannosylation in N-glycosylation by MALDI-TOF-MS or LC-MS/MS analysis. The discov-

ery of this N-tetrasaccharide also implies an alternative Golgi apparatus–associated secretory pathway of protein N-glycosylation. Because of our limited understanding of genetic defects that lead to GDP-mannose deficiency, our study included only 3 known CDG subtypes, ALG1CDG, PMM2-CDG, and MPI-CDG. As new CDG subtypes are being discovered, a similar mannose deficient N-glycan profile may be seen in other new defects in the same or adjacent pathways.

Author Contributions: All authors confirmed they have contributed to the intellectual content of this paper and have met the following 3 requirements: (a) significant contributions to the conception and design, acquisition of data, or analysis and interpretation of data; (b) drafting or revising the article for intellectual content; and (c) final approval of the published article. Authors’ Disclosures or Potential Conflicts of Interest: Upon manuscript submission, all authors completed the author disclosure form. Disclosures and/or potential conflicts of interest: Employment or Leadership: W. Zhang, Emory University; H.H. Freeze Sanford-Burnham Medical Research Institute. Consultant or Advisory Role: None declared. Stock Ownership: None declared. Honoraria: None declared. Research Funding: H.H. Freeze, the Rocket Fund and R01 DK99551. Expert Testimony: None declared. Patents: None declared. Other Remuneration: T. Wood, BioMarin Pharma. Role of Sponsor: The funding organizations played no role in the design of study, choice of enrolled patients, review and interpretation of data, or preparation or approval of manuscript. Acknowledgments: We gratefully acknowledge the participation of the families in this research.

References 1. Freeze HH, Chong JX, Bamshad MJ, Ng BG. Solving glycosylation disorders: fundamental approaches reveal complicated pathways. Am J Hum Genet 2014;94:161– 75. 2. Steentoft C, Vakhrushev SY, Joshi HJ, Kong Y, VesterChristensen MB, Schjoldager KT, et al. Precision mapping of the human O-GalNAc glycoproteome through SimpleCell technology. EMBO J 2013;32:1478 – 88. 3. Lacey JM, Bergen HR, Magera MJ, Naylor S, O’Brien JF. Rapid determination of transferrin isoforms by immunoaffinity liquid chromatography and electrospray mass spectrometry. Clin Chem 2001;47:513– 8. 4. Xia B, Zhang W, Li X, Jiang R, Harper T, Liu R, Cummings RD, He M. Serum N-glycan and O-glycan analysis by mass spectrometry for diagnosis of congenital disorder of glycosylation. Anal Biochem 2013;442:178 – 85.

10


5. Xia B, Feasley CL, Sachdev GP, Smith DF, Cummings RD. Glycan reductive isotope labeling for quantitative glycomics. Anal Biochem 2009;387:162–70. 6. Xia B, Royall JA, Damera G, Sachdev GP, Cummings RD. Altered O-glycosylation and sulfation of airway mucins associated with cystic fibrosis. Glycobiology 2005;15: 747–75. 7. Isono T. O-GlcNAc-specific antibody CTD110.6 crossreacts with N-GlcNAc2-modified proteins induced under glucose deprivation. PLoS One 2011;6:e18959. 8. Joziasse DH, Lee RT, Lee YC, Biessen EA, Schiphorst WE, Koeleman CA, van den Eijnden DH. alpha3Galactosylated glycoproteins can bind to the hepatic asialoglycoprotein receptor. Eur J Biochem 2000;267: 6501– 8. 9. Timal S, Hoischen A, Lehle L, Adamowicz M, Huijben

K, Sykut-Cegielska J, et al. Gene identification in the congenital disorders of glycosylation type I by wholeexome sequencing. Hum Mol Genet 2012;21:4151– 61. 10. Matthijs G, Schollen E, Heykants L, Grunewald S. Phosphomannomutase deficiency: the molecular basis of the classical Jaeken syndrome (CDGS type Ia). Mol Genet Metab 1999;68:220 – 6. 11. Losfeld ME, Soncin F, Ng BG, Singec I, Freeze HH. A sensitive green fluorescent protein biomarker of N-glycosylation site occupancy. FASEB J 2012;26: 4210 –7. 12. Sone H, Shimano H, Ebinuma H, Takahashi A, Yano Y, Iida KT, et al. Physiological changes in circulating mannose levels in normal, glucose-intolerant, and diabetic subjects. Metabolism 2003;52:1019 –27.