Lysosomal Targeting Enhancement by Conjugation of Glycopeptides Containing Mannose-6-phosphate Glycans Derived from Glyco-engineered Yeast

Preparation of M6PgPs from cell wall mannoproteins of glyco-engineered S. cerevisiae

In a previous study, our group constructed a glyco-engineered S. cerevisiae strain overexpressing the YlMPO1 gene (och1Δmnn1Δ/YlMPO1); the OCH1 and MNN1 genes were disrupted to block yeast-specific hyper-mannosylation and immunogenic α(1,3)-mannosylation while the YlMPO1 gene was introduced to increase mannosylphosphorylation¹⁸. Mannosylphosphorylated mannose residues (mannose-1-phosphate-6-O-mannose) can be converted to M6P structures (phosphate-6-O-mannose) by uncapping of the outer mannoses using mild acid hydrolysis (MAH) or uncapping enzyme^7,19. We thought that cell wall mannoproteins, a major component of yeast cell wall, would be a good source for M6P glycans because they can be easily obtained from the harvested cells^18,20. The M6PgPs were prepared from cell wall mannoproteins through pronase digestion and subsequent uncapping, trimming, and purification processes.

The M6PgP preparation procedure is summarized in Fig. 1A. After culturing S. cerevisiae och1Δmnn1Δ/YlMPO1 strain, cell wall mannoproteins were isolated from the harvested cells by hot citrate buffer extraction and subsequent ethanol precipitation. They were digested with pronase, which generated very short glycopeptides together with free amino acids²¹. The glycopeptides were efficiently purified on a graphitized carbon column due to the glycan moiety being much larger than the peptide part. Their glycans were released by PNGase F treatment and analyzed by matrix-assisted laser desorption ionization time-of-flight (MALDI-TOF) mass spectrometry (Fig. 1B-a), which showed four types of glycans: neutral (Man₈GlcNAc₂), mannosylphosphorylated [Man-P-Man₈GlcNAc₂ and (Man-P)₂-Man₈GlcNAc₂], phosphorylated (P-Man₈GlcNAc₂ and P₂-Man₈GlcNAc₂), and mannosylphosphorylated and phosphorylated (Man-P-/P-Man₈GlcNAc₂) glycans. Here, the phosphorylated glycan structures were generated from the unexpected uncapping reaction induced by the acidic matrix preparation step for MALDI-TOF mass spectrometry analysis, which was previously reported²². After MAH for uncapping, all mannosylphosphorylated mannose moieties were converted to M6P structures, which resulted in a simple glycan profile with neutral (Man₈GlcNAc₂), mono-phosphorylated (P-Man₈GlcNAc₂), and bi-phosphorylated (P₂-Man₈GlcNAc₂) glycans (Fig. 1B-b). Through α(1,2)-mannosidase treatment, neutral and bi-phosphorylated glycans were trimmed to Man₅GlcNAc₂ and P₂-Man₆GlcNAc₂ glycans. On the other hand, the mono-phosphorylated glycan was converted to P-Man₅GlcNAc₂ and P-Man₆GlcNAc₂ glycans depending on the phosphorylation site (Fig. 1B-c). Here, P-Man₆GlcNAc₂ glycan can be created only if the phosphorylation site is located on the penultimate α(1,2)-mannose residue of α(1,3)-arm because α(1,2)-mannosidase does not remove the phosphorylated mannose (Supplementary Fig. S1A). In contrast, all α(1,2)-mannoses can be digested to generate P-Man₅GlcNAc₂ glycan when the penultimate α(1,6)-mannose residue of α(1,6)-arm is phosphorylated (Supplementary Fig. S1B). The M6PgPs were purified from the neutral glycopeptides by using titanium dioxide (TiO₂) chromatography (Fig. 1A-d), which was reported to purify negatively charged peptides such as phosphorylated peptides²³ and sialylated glycopeptides²⁴. The M6PgPs were selectively purified because the negative charges of phosphates (Fig. 1B-d) and existence of phosphate groups in the purified M6PgPs was confirmed by alkaline phosphatase treatment (Fig. 1B-e).

The glycopeptides purified by TiO₂ chromatography were also directly analyzed to know the lengths of peptides and amino acid sequences by using the liquid chromatography tandem-mass (LC-MS/MS) spectrometry (Supplementary Figs S2–4). Because pronase digestion generated very short peptides without specific cleavage sites, we could not obtain the amino acid sequence information. Instead, the molecular weights of peptide parts could be determined from the Y0 ion (peptide + H⁺) (Table 1). The Y0 ion masses of 23 identified glycopeptides ranged from 280.06 to 938.51. The average molecular weight of peptide parts was calculated to be 564.95 Da (Supplementary Fig. S3), suggesting that the average peptide length would be ~5.1 amino acids upon considering the average molecular weight of amino acids (110 Da). The 14 glycopeptides were found to contain P-Man₅GlcNAc₂ glycan, which is the strongest peak in the glycan profile analysis by using MALDI-TOF mass spectrometry (Fig. 1B-d). Two glycopeptides carried P-Man₆GlcNAc₂ glycan while the glycopeptide containing P₂-Man₆GlcNAc₂ glycan was not found in proteomics analysis (Table 1) although the P₂-Man₆GlcNAc₂ glycan peak is stronger than the P-Man₆GlcNAc₂ peak in the glycan profile analysis (Fig. 1B-d). Notably, five glycopeptides containing non-phosphorylated high-mannose type glycans were identified (Table 1). Minor amounts of six glycopeptides containing O-linked mannoses or phosphomannoses were also found in the LC-MS/MS analysis (Supplementary Table S1). Here, only two Y0 ion masses (903.56 and 793.52) existed in the six glycopeptides, suggesting that two peptides were attached with the different numbers of mannoses (Man_5–7). Notably, these chain lengths appeared to be considerably long, considering that S. cerevisiae usually forms O-linked Man_1–6 in length²⁵. It was suspected that the only glycopeptides containing long chains of O-linked mannoses were retrieved because the ones harboring short chains were lost during the graphitized carbon column purification step.

Both uncapping and α(1,2)-mannosidase trimming are required for proper lysosomal targeting

For analysis of lysosomal targeting capability, the glycopeptides from the preparation steps were labeled with biotin and then bound to streptavidin-Alexa Fluor 633 conjugate (streptavidin-Alexa633). The resulting complexes of glycopeptide-biotin and streptavidin-Alexa633 were used to treat human fibroblasts for one hour, followed by a cellular uptake analysis (Fig. 2). Flow cytometry analysis shows that the glycopeptides containing phosphorylated glycans (P-Man₈GlcNAc₂ and P₂-Man₈GlcNAc₂) without α(1,2)-mannosidase trimming as well as mannosylphosphorylated glycans were rarely taken up by human fibroblasts. A significant population (28.6%) of fibroblasts endocytosed the M6PgPs (Fig. 2A-c) only after both the outer mannose uncapping and terminal α(1,2)-mannose trimming steps. Notably, simple addition of a TiO₂ column purification step greatly increased cellular uptake efficiency up to 94.4% (Fig. 2A-d).

Fluorescence microscopic images show that the endocytosed M6PgPs (P-Man_5–6GlcNAc₂ and P₂-Man₆GlcNAc₂) were localized to lysosomes; the red fluorescence of M6PgPs was well merged with the green fluorescence of the lysosome marker (Fig. 2B). As shown in fluorescence microscopy analysis, only M6PgPs elaborated by both of the outer mannose uncapping and the terminal mannose trimming steps were efficiently targeted to lysosomes (Fig. 2B-c). Also, the TiO₂ column purification step greatly increased the lysosome targeting capability (Fig. 2B-d). Taken together, pronase digestion, uncapping, mannose trimming, and TiO₂ column purification efficiently produced M6PgPs from cell wall mannnoproteins of glyco-engineered yeast, and these M6PgPs displayed excellent lysosome targeting capability.

Conjugation of a therapeutic enzyme with M6PgPs by using copper-free click chemistry

To prove the usefulness of the M6PgPs, we tried to attach them to a therapeutic enzyme, rGAA (alglucosidase alfa) used for treatment of Pompe disease. rGAA has a low efficacy due to a low M6P glycan content^8,11. Popular chemical crosslinkers containing amine-reactive N-hydroxysuccinimide (NHS) esters were first employed because both rGAA and M6PgPs have primary amines in Lys residues and/or the N-terminal amino acid. However, the use of homo-bifunctional crosslinkers containing NHS ester groups at both ends was not successful and induced aggregations despite various trials with adjustments of molar ratio (data not shown).

In order to avoid aggregation, we designed two-step reactions using two hetero-bifunctional crosslinkers (NHS-PEG₄-Azide and NHS-PEG₄-DBCO). First, the NHS ester of each crosslinker was reacted with primary amines of rGAA or M6PgPs to form covalent amide bonds. In the second reaction, the azide of one conjugate was linked to DBCO of the other conjugate through a copper-free click reaction, which is bioorthgonal²⁶ and also well suited for protein conjugation without impairing activity and function²⁷. This two-step reaction by using NHS-PEG₄-Azide and NHS-PEG₄-DBCO can be achieved through two different reaction routes (Fig. 3A). In the first route, after reaction of rGAA with NHS-PEG₄-Azide, the resulting azide-functionalized rGAA (rGAA-azide) was ligated to DBCO-functionalized M6PgP (DBCO-M6PgP) prepared by conjugation of M6PgP with NHS-PEG₄-DBCO, which generated rGAA-azDB-M6PgP (Fig. 3A-a). In the other route, rGAA-DBCO made by conjugation of rGAA with NHS-PEG₄-DBCO was linked to azide-M6PgP, which generated rGAA-DBaz-M6PgP (Fig. 3A-b).

The M6PgP-conjugated rGAAs prepared through two different routes (rGAA-azDB-M6PgP or rGAA-DBaz-M6PgP) were analyzed by SDS-PAGE (Fig. 3B). The conjugation reactions of rGAA with crosslinkers were accomplished at three different molar ratios (1:5, 1:10, 1:20) as rGAA has 16 primary amines (15 Lys residues and one N-terminal α-amino group) at the maximum while M6PgP was conjugated with 10-fold molar excess of crosslinkers for completion of reaction. The resulting rGAA-azide or rGAA-DBCO was reacted with four fold molar excess of DBCO-M6PgP or azide-M6PgP, respectively; therefore, the used molar ratios were 1:5:20, 1:10:40, and 1:20:80 (rGAA:NHS-PEG₄-Azide:DBCO-M6PgP or rGAA:NHS-PEG₄-DBCO:azide-M6PgP). In the rGAA-azDB-M6PgP conjugation route, all reaction products showed one band around 100 kDa, which is a similar size as rGAA (Fig. 3B-a). In contrast, two new bands (~250 kDa and ~120 kDa), besides the ~100 kDa band, were found in rGAA-DBCO as well as rGAA-DBaz-M6PgPs (lane 3–6 of Fig. 3B-b), which suggested that the newly formed bands were generated from the first conjugation reaction of rGAA with NHS-PEG₄-DBCO rather than the second reaction.

The M6P glycans of M6PgP-conjugated rGAAs were analyzed by blotting with domain 9 of the cation-independent MPR (Dom9), which was reported to bind M6P glycan with high affinity^17,28. The rGAA displayed ~100 kDa band in Dom9 blots (lane 2 of Fig. 3C) while the rGAA treated with PNGase F displayed no band (lane 1 of Fig. 3C), indicating that Dom9 recognizes N-glycan structure. Conjugation of rGAA-azide with DBCO-M6PgP enhanced the intensities of ~100 kDa band (lane 4–6 of Fig. 3C-a). The band intensities of 1:10:40 and 1:20:80 (rGAA:NHS-PEG₄-Azide:DBCO-M6PgP) reaction products were higher than that of the 1:5:20 reaction product. On the other hand, the rGAA-DBCO conjugation with azide-M6PgP resulted in formation of the aggregated products (lane 3–6 of Fig. 3C-b).

Since SDS-PAGE and Dom9 blotting results suggested the existence of nonspecific aggregated products, especially in rGAA-DBCO conjugates, we investigated whether rGAA conjugates can maintain GAA enzyme activity (Fig. 3D). The rGAA-azide and rGAA-azDB-M6PgPs showed similar activities with non-modified rGAA, whereas rGAA-DBCO and rGAA-DBaz-M6PgPs had significantly decreased activity. This decrease would be caused solely by conjugation of rGAA with NHS-PEG₄-DBCO because GAA-DBCO and GAA-DBaz-M6PgP displayed similar low values. Since conjugation of rGAA with NHS-PEG₄-DBCO induced aggregation problems involving activity decrease, the rGAA-azDB-M6PgP reaction route was selected for efficient M6PgP conjugation.

N-Glycan analysis of M6PgP-conjugated rGAA

N-glycans of rGAA and three rGAA-azDB-M6PgPs produced from the different molar ratio reactions were released by PNGase F treatment and analyzed after 2-aminobenzoic acid (2-AA) labeling by using high performance liquid chromatography (HPLC) (Fig. 4A), which was optimized for M6P glycan analysis in our previous work^19,22. Glycan peaks separated in HPLC were collected and their masses were measured for identification by using MALDI-TOF mass spectrometry (Fig. 4C); the glycan peak identities were further confirmed by α(1,2)-mannosidase and/or alkaline phosphatase digestions (data not shown). Also, mass spectra of whole 2-AA-labeled glycans were analyzed as a crosscheck (Supplementary Fig. S5). As reported^10,12, rGAA has high-mannose type (peak 1 and 3), non-sialylated (peak 2), and sialylated complex type bi-antennary (peak 4, 5 and 6) glycans together with very low content of M6P glycans (peak 7 and 9); relative content of the M6P glycan peaks, calculated from the integrated peak areas, was just 3 ± 0.4% (Fig. 4B). Owing to low intensities in rGAA glycan analysis, the mono-phosphorylated glycan peaks in close location were collected together and assigned as the peak 7 (P-Man₅GlcNAc₂ and P-Man₆GlcNAc₂), while the peak 9 was assigned as bi-phosphorylated glycan (P₂-Man₆GlcNAc₂). The N-glycan profile obtained from M6PgP-conjugated rGAA showed that the intensities of M6P glycan peaks greatly increased. Here, the masses of peak 7′ and 8 correspond to P-Man₅GlcNAc₂ and P-Man₆GlcNAc₂ glycans, respectively. Because peak 8, observed only in M6PgP-conjugated rGAA, was generated by α(1,2)-mannosidase digestion, its phosphorylation site should be located at the terminal mannose of α(1,3)-branch. The highest relative content (49 ± 5.1%) of M6P glycans was observed in M6PgP-conjugated rGAA generated by the 1:10:40 (rGAA:NHS-PEG₄-Azide:DBCO-M6PgP) molar ratio reaction while the 1:5:20 and 1:20:80 reaction products displayed 32 ± 4.1% and 48 ± 4.3% M6P glycan contents, respectively. Therefore, in the following experiments, we used M6PgP-conjugated rGAA generated by the 1:10:40 reaction.

M6PgP-conjugated rGAA uptake by fibroblasts of Pompe disease patient

The uptakes of M6PgP-conjugated rGAA and unmodified rGAA by Pompe disease patient fibroblasts were compared by measuring intracellar GAA activity after incubation with increasing concentrations of the enzyme (Fig. 5A). M6PgP-conjugated rGAA showed much higher activities at all of the measured concentrations. Moreover, the highest activity displayed by 100 nM of rGAA was similar to that by 10 nM of M6PgP-conjugated rGAA. The intracellular GAA activities enhanced by 20 nM M6PgP-conjugated rGAA treatment were greatly reduced by addition of free M6P (5 mM) and not by mannan (2 mg/ml) (Fig. 5B), indicating that the uptake was achieved by MPR-mediated endocytosis.

The deficiency of lysosomal enzyme GAA in Pompe disease patient led to glycogen accumulation in many tissues including skeletal muscle and heart. However, most of the stored glycogen is exocytosed from the cultured fibroblasts of Pompe patient²⁹. When the Pompe fibroblasts were examined by using periodic acid-schiff (PAS) staining, cell surfaces were stained and only a few purple spots were observed inside cells (Fig. 5C). However, addition of colchicine, which was reported to inhibit glycogen exocytosis³⁰, greatly increased the number and intensity of strong purple spots in Pompe fibroblasts (Fig. 5C), which indicates that exocytosis inhibition induced glycogen accumulation. The areas of strong purples spots were quantitatively analyzed from the images of the PAS-stained cells in 12-well plates (Supplementary Fig. S6), followed by calculation of relative fold change in comparison with the strong purple spot area of control cells (Fig. 5D). The cells cultured in the presence of colchicine for five days showed the highest fold change of area (29.0 ± 2.3), which indicated the accumulation of glycogen. Further culture in the medium without colchicine for two days resulted in the slightly decreased fold change value (22.3 ± 5.3), which would be caused by spontaneous exocytosis of glycogen. When the cells were further cultured in the medium containing 100 nM rGAA, the fold change value of strong purple spot area was 20.2 ± 3.6, which is a statistically insignificant decrease. In contrast, the further culture in the medium containing 100 nM M6PgP-conjugated rGAA showed the significantly decreased value (7.7 ± 2.3), clearly indicating that M6PgP-conjugated rGAA was much more efficient in removing the accumulated glycogen in the Pompe fibroblasts. The value dropped by M6PgP-conjugated rGAA treatment was reversed up to 19.5 ± 4.8 by addition of 5 mM free M6P, indicating that the clearance of accumulated glycogen is achieved by MPR-mediated endocytosis of M6PgP-conjugated rGAA.

Preparation of M6PgPs from glyco-engineered yeast

M6PgPs were prepared from the cell wall mannoproteins of the S. cerevisiae och1Δmnn1Δ strain transformed with the YEP352-YlMpo1 plasmid¹⁸. First, the ~20 mg of yeast cell wall mannoproteins (wet cell weight) were extracted with hot citrate buffer, as described previously¹⁸. The obtained mannoproteins (~1 mg) were digested with 0.1 mg of pronase at 55 °C for 48 h. From the 1 ml of resulting reaction mixture, short glycopeptides were purified by using solid-phase extraction using graphitized carbon columns (Alltech, Lexington, MA) as previously described³⁴. The mannosylphosphorylated glycans of glycopeptides were uncapped by MAH; 0.5 M formic acid at 80 °C for 1 h^19,22. After drying in a speedvac, the uncapped glycopeptides (~0.2 mg) were treated with 10 mU of α(1,2)-mannosidase in 20 mM ammnonium acetate (pH 5.0) at 37 °C for 16 h. Finally, the M6PgPs were selectively purified by using Titanosphere (TiO2) beads (5 µm chromatographic material; GL Sciences, Japan), as previously described²⁴ with a slight modification. Briefly, the whole glycopeptide fraction was loaded on TiO₂ beads in loading buffer (1 M glycolic acid in 80% ACN, 5% TFA), followed by incubation at room temperature (RT) for 30 min. After the removal of supernatant, the beads were thoroughly washed by using washing buffer 1 (80% ACN, 1% TFA) and 2 (80% ACN, 0.1% TFA). The M6PgPs were eluted from the TiO₂ beads by using elution buffer (28% ammonium hydroxide solution). The concentrations of the glycopeptides were measured by using a fluorescamine fluorescence assay³⁵ and monosaccharide composition analysis³⁶. For the fluorescamine assay, the glycopeptides eluted from the TiO₂ beads were dried in a speedvac and then were redissolved in 160 μl of 0.5 M sodium borate buffer (pH 8.0). While the solutions were vortexed vigorously, 40 μl of 0.3 mg/ml fluorescamine (Sigma-Aldrich, St. Louis, MO) in acetone was rapidly added. The fluorescence was analyzed on a SpectraMax i3 plate reader (Molecular Devices, San Jose, CA) with 390 nm excitation and 470 nm emission (10 nm bandwidth). For the monosaccharide composition analysis, the carbohydrates released from glycopeptides by acid hydrolysis were labeled with 2-AA, and then analyzed by using reverse phase HPLC with fluorometric detection as described previously³⁶. Monosaccharide standard solutions (Prozyme, Hayward, CA) were used for peak identification and quantification.

LC-MS/MS for glycopeptide analysis

The purified glycopeptide sample was dissolved in 20 μl of water containing 0.1% formic acid (Sigma-Aldrich) and analyzed using an LC-MS/MS system consisting of an EASY-nanoLC 1100 System (ThermoFisher Scientific, Waltham, MA) and an LTQ Orbitrap Elite mass spectrometer (ThermoFisher Scientific) equipped with an EASY-Spray source. Briefly, six µl aliquots of the peptide solutions were loaded into Acclaim PepMap 100 (75 µm × 2 cm, nanoViper). Peptides were desalted for 4 min at a flow rate of 4 μl/min. Then, the trapped peptides were separated on a PepMap RSLC column (C₁₈ particle size 2 μm, 100 A, 75 μm × 50 cm). The mobile phases, A and B, were composed of 0 and 100% acetonitrile, respectively, and each contained 0.1% formic acid. The LC gradient began with 5% B for 1 min and was ramped to 8% B over 16 min, to 40% B over 74 min, and to 95% B over 1 min. Then, it remained at 95% B over 8 min and was decreased to 5% B for another 5 min. The column was re-equilibrated with 2% B for 15 min before the next run. The voltage applied to produce an electrospray was 2.2 kV. During the chromatographic separation, the LTQ Orbitrap Elite was operated in a data-dependent mode. The MS data were acquired using the following parameters: five data-dependent collision induced dissociation-high energy collision dissociation (CID-HCD) dual MS/MS scans per full scan; CID scans were acquired in linear trap quadrupole (LTQ) with two-microscan averaging; full and HCD scans were acquired in Orbitrap at resolution 120,000 and 15,000, respectively, with two-microscan averaging; 35% normalized collision energy (NCE) in CID and 35% NCE in HCD; ± 5 Da isolation window. Previously fragmented ions were excluded for 30 sec. In CID-HCD dual scan, each selected parent ion was first fragmented by CID and then by HCD.

All MS and MS/MS spectra were manually identified as follows. First, the glycopeptides were selected from HCD and CID MS/MS spectra based on the presence of unique oxonium ions from glycan fragmentation. There are ions of hexose with phosphate [m/z 243.03 of Hex(1)Pi(1), 405.03 of Hex(2)Pi(1) and 770.22 of HexNAc(1)Hex(3)Pi(1)] and without phosphate [(m/z 325.11 of Hex(2) and 690.24 of HexNAc(1)Hex(3)]. Second, we confirmed the presence of B-type fragment ions by the neutral loss of the peptide with HexNAc(1) from the glycopeptide precursor ions in CID MS/MS spectra. Finally, in order to determine the molecular weight of peptide, at least two Y-type ions were confirmed³⁷.

Biotin labeling of glycopeptides and cellular uptake analyses

The glycopeptides in each of the M6PgP preparation steps were biotinylated with the Biotin Protein Labeling Kit (Roche, Basel, Switzerland); 50 μM of glycopeptides were incubated in phosphate-buffered saline (PBS) with 10-fold molar excess of biotin-7-NHS for 2 h at room temperature (RT) and non-reacted labeling reagents were removed by dialysis over an one kDa cutoff membrane. The 2.5 μM of biotinylated glycopeptides were allowed to bind 10 μg/ml of streptavidin-Alexa633 (ThermoFisher Scientific) in 0.2 ml Dulbecco’s Modified Eagle Medium (DMEM) for 1 h at RT. The resulting solutions were incubated with human fibroblasts which were cultured in DMEM supplemented with 10% fetal bovine serum (FBS), at 3 × 10⁴ cells/well in μ-slide 8-well plates (Ibidi, Verona, WI) in a 5% CO₂ incubator at 37 °C. After 1 h incubation of the glycopeptide-streptavidin complex with human fibroblasts, the cells were extensively washed with PBS and then analyzed on a FACSVerse flow cytometer (BD Biosciences, San Diego, CA).

For fluorescent microscopic image analysis, after 1 h incubation with the glycopeptide-streptavidin complex (2.5 μM biotinylated glycopeptides and 10 μg/ml streptavidin- Alexa633), the fibroblasts were washed with PBS and then stained with 50 nM LysoTracker green (ThermoFisher Scientific) for 1 h at 37 °C. After PBS washing, the cells were fixed in 2% paraformaldehyde for 20 min at RT, followed by nuclei staining using 4′-6 diamidino-2-phenylindole (DAPI) (ThermoFisher Scientific). Fluorescence images of the stained cells were analyzed using a Zeiss LSM510 confocal microscope (Carl Zeiss, Oberkochen, Germany).

Conjugation of rGAA with M6PgPs by using two hetero-bifunctional crosslinkers

The rGAA, aglucosidase alfa (Myozyme, Sanofi Genzyme, Boston, MA, USA), was conjugated with NHS-PEG₄-Azide (Pierce, Rockford, IL, USA) or NHS-PEG₄-dibenzocyclooctyne (DBCO) (Click Chemistry Tools, San Diego, CA). Crosslinker stock solutions (200 mM) prepared in dimethyl sulfoxide (DMSO) were diluted to 200 µM of rGAA solution in 100 µL PBS for the conjugation reaction with 5-fold, 10-fold, or 20-fold molar excess. After 3 h incubation at RT, non-reacted crosslinkers were removed by dialysis in PBS over a 30 kDa cutoff membrane. Likewise, the M6PgPs were also conjugated with NHS-PEG₄-DBCO or NHS-PEG₄-Azide. Crosslinker stock solutions (200 mM) prepared in DMSO were diluted to 1 mM of M6PgP solution in 200 µL PBS for the conjugation reaction with 10-fold molar excess. After 3 h incubation at RT, non-reacted crosslinkers were removed by dialysis in PBS over a 1 kDa cutoff membrane (Spectrum Labs). The resulting M6PgP conjugates were dried in a speedvac and made to 50 mM stock solution by re-dissolving it in PBS. In the second reaction, azide-conjugated and DBCO-conjugated rGAA were incubated, respectively, with four-fold molar excess of DBCO-conjugated and azide-conjugated M6PgPs in PBS for 4 h at 37 °C, which generated 1:5:20 (50 µM rGAA: 250 µM NHS-PEG₄-Azide: 1 mM DBCO-M6PgP and 50 µM rGAA: 250 µM NHS-PEG₄-DBCO: 1 mM azide-M6PgP), 1:10:40 (50 µM rGAA: 500 µM NHS-PEG₄-Azide: 2 mM DBCO-M6PgP and 50 µM rGAA: 500 µM NHS-PEG₄-DBCO: 2 mM azide-M6PgP), and 1:20:80 (50 µM rGAA: 1 mM NHS-PEG₄-Azide: 4 mM DBCO-M6PgP and 50 µM rGAA: 1 mM NHS-PEG₄-DBCO: 4 mM azide-M6PgP) conjugation products. Unreacted azide- or DBCO-conjugated M6PgPs were removed by dialysis in PBS using a 30 kDa cutoff membrane.

In vitro GAA activity assay

The GAA enzyme activity was assayed by using the substrate 4-methylumbelliferyl-α-D-glucopyranoside (4-MUG) (Sigma-Aldrich) which generates fluorescence on digestion. The 50 µl of the reaction mixture (0.2 M sodium acetate, 0.4 M potassium chloride, pH 4.3) including one µg of rGAA or its conjugates were incubated for 30 min at 37 °C. The reaction was terminated by adding 150 µl of 100 mM glycine/NaOH buffer (pH 11) and the fluorescence was measured by using a plate reader SpectraMax i3 (Molecular Devices) with 375 nm excitation and 460 nm emission.

Dom9 blotting for detection of M6P glycans

The gene coding domain 9 (Dom9) of the cation-independent M6P receptor was synthesized with codon optimization for recombinant expression in Pichia pastoris and cloned to the pD912 vector by the gene synthesis and cloning service of ATUM (Newark, CA). Three tandem Flag-tags and a His₈-tag were added to the C-terminus of Dom9 (Dom9–3xFlag-His₈); the whole amino acid sequences including the α-mating factor for secretion are represented in Supplementary Fig. S8. The vector pD912-Dom9 was transformed to P. pastoris GS115 strain and the transformants were selected on YPD (1% yeast extract, 2% Bacto peptone, 2% glucose) containing 0.1 mg/ml Zeocin. A single colony was initially cultured in YPD medium containing 0.1 mg/ml Zeocin, which was transferred to flasks containing BMMY (1% yeast extract, 2% peptone, 100 mM potassium phosphate, pH 6.0, 1.34% YNB, 0.00004% biotin, 0.5% MeOH) medium for methanol induction and further cultured for 3 days at 30 °C; Dom9 overexpression was maintained by adding 0.5% methanol every 24 h. Dom9–3xFlag-His₈ protein was purified from the culture supernatant by using Ni-NTA column chromatography as described previously^34,38.

After standard sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) using 8% polyacrylamide gel, rGAA and its conjugates were transferred to a PVDF membrane. After a blocking step using 5% skim milk in Tris-buffered saline (50 mM Tris-Cl, pH 7.5, 150 mM NaCl) containing 0.1% Tween-20 (TBST), the membrane was incubated with 0.5 μg/ml of Dom9–3xFlag-His₈ in TBST for 2 h. After three times repeated washing steps, the membrane was incubated with a horseradish peroxidase-conjugated anti-Flag antibody (Sigma-Aldrich) in TBST for 1 h. After extensive washing steps, the resulting membrane was visualized by using ECL Western blot detection kit (GE Healthcare Life Sciences, Chicago, IL).

N-Glycan analysis

N-Glycans were released from the glycopeptides or rGAA by PNGase F (New England Biolabs, Ipswich, MA) treatment and purified by solid-phase extraction using graphitized carbon columns as described previously^34,39. After labeling with 2-AA, the labeled glycans were purified from unreacted reagents using cyano-SPE cartridge (Agilent Technologies, Santa Clara, CA) as previously described^19,38. The 2-AA-labeled glycans were analyzed on a Shodex Asahipak NH2P-50 4E (4.6 × 250 mm) column (Showa Denko, Tokyo, Japan) using a Waters Alliance system equipped with a Waters 2475 fluorescence detector (Milford, MA) in the conditions optimized for M6P glycan analysis as previously described^19,22. The eluted fractions corresponding to the glycan peaks were collected, and their masses were identified by MALDI-TOF mass spectrometry (Microflex, Bruker Daltonik, Bremen, Germany), as described previously^18,22.

Cellular uptake and intracellular GAA activity assay

The fibroblasts cells (GM00248) of Pompe patient (Pompe fibroblast), who had homozygous mutation in exon 18 of the GAA gene leading to formation of premature stop codon at 854, were obtained from the Coriell Institute (https://catalog.coriell.org/). Pompe fibroblasts were seeded at 3 × 10⁴ cells/well in 24 well culture dishes in DMEM supplemented with 10% FBS and allowed to settle for 24 h at 37 °C. Various concentrations (0–100 nM) of rGAA and its conjugates were added to the cells and incubated for 18 h at 37 °C; for inhibition assay, 5 mM of M6P (GeneChem, Daejeon, Korea) and 2 mg/ml of mannan (Sigma-Aldrich) were added. After incubation, the cells were washed first with PBS containing 1 mM M6P and then with PBS alone. The cells were harvested with 0.25% trypsin and resuspended in GAA assay buffer (0.2 M sodium acetate, 0.4 M potassium chloride, 0.1% triton X-100, pH 4.3). For cell lysis, cell suspensions were frozen at −80��°C for 30 min, followed by thawing at 37 °C. The resulting cell lysates were centrifuged at 14,000 g for 10 min to remove cell debris. The supernatants were assayed by using 4-MUG as described in ‘in vitro GAA activity assay’ subsection.

Glycogen exocytosis inhibition and periodic acid-schiff (PAS) staining

Pompe fibroblasts were seeded at 1 × 10⁵ cells/well in 12-well culture plates and cultured for 1 day in DMEM supplemented with 10% FBS. To inhibit glycogen exocytosis, Pompe fibroblasts were further cultured in the presence of 1 nM colchicine (Sigma-Aldrich) for 5 days, followed by the culture in five different conditions for 2 days; (condition 1) in the presence of 1 nM colchicine, (condition 2) without both colchicine and enzyme treatment (exocytosed), (condition 3) in the presence of 100 nM rGAA, (condition 4) in the presence of 100 nM M6PgP-conjugated rGAA, and (condition 5) in the presence of 100 nM M6PgP-conjugated rGAA with 5 mM M6P. The cells were stained by using PAS staining kit (Merck KGaA, Darmstadt, Germany) according to the manufacturer’s instruction and the resulting images were obtained by light microscopy. The images obtained from each well of 12-well culture plates were analyzed by using NIH Image J software (http://rsb.info.nih.gov/ij/). Strong purple spots over arbitrary threshold values were identified (Supplementary Fig. S6), and their spot areas were quantified by using Image J software. The relative fold changes were obtained by comparing the strong purple spot areas of corresponding cells with those obtained from control cells. Statistical analyses including one-way ANOVA with Tukey’s multiple comparison test were performed by using GrphPad Prism (GraphPad Software, San Diego, CA; http://www.graphpad.com)

Competing Interests

The authors have declared that no competing financial interests exist. Although a patent application has been filed on aspects of this work, there is no other relevant competing interest.