Labeling cell surface glycosylphosphatidylinositol-anchored proteins through metabolic engineering using an azide-modified phosphatidylinositol

Abstract

Glycosylphosphatidylinositol (GPI) anchorage is one of the most common mechanisms to attach proteins to the plasma membrane of eukaryotic cells. GPI-anchored proteins (GPI-APs) play a critical role in many biological processes but are difficult to study. Here, a new method was developed for the effective and selective metabolic engineering and labeling of cell surface GPI-APs with an azide-modified phosphatidylinositol (PI) as the biosynthetic precursor of GPIs. It was demonstrated that this azido-PI derivative was taken up by HeLa cells and incorporated into the biosynthetic pathway of GPIs to present azide-labeled GPI-APs on the live cell surface. The azido group was used as a molecular handle to install other labels through a biocompatible click reaction to enable various biological studies, e.g., fluorescent imaging and protein pull-down, which can help explore the functions of GPI-APs and discover new GPI-APs.

Graphical Abstract

Introduction

Glycosylphosphatidylinositol-anchored proteins (GPI-APs) are unique surface proteins having the lipids of their GPI anchors at the polypeptide C-terminus embedded in the cell membrane [1]. GPI attachment is one of the most common posttranslational protein modifications in eukaryotes. GPI-APs are responsible for many biological activities, e.g., cell recognition and signal transduction [2-5]. Usually, GPIs are associated with the lipid rafts in the cell membrane to modulate membrane dynamics and GPI-AP aggregation, trafficking, and function [6-11]. Abnormal GPI-AP metabolism has been correlated with human diseases, such as cancer [12-14].

Despite the general acknowledgement of the importance and widespread existence of GPI-APs revealed by in silico analysis [15-17], two-dimensional gel electrophoresis [18], solubility assay [19], immunochemistry [20], mass spectrometry [21-23], metabolic engineering- and toxin/lectin-based GPI-AP enrichment [24, 25] and combinations of different isolating and analytical methods [26-31], GPI-AP discovery remains challenging. First, GPI-APs are difficult to isolate and purify. Second, natural GPI-APs are diverse, and each GPI-AP may have multiple GPI forms [32]. Third, GPI-APs typically exist in low concentrations on the cell surface. Therefore, although >0.5% of all proteins are GPI-modified [33], only ~150 mammalian GPI-APs have been characterized [5], since the characterization of the first GPI anchor over 30 years ago [34].

Irrespective of their origins, GPI-APs identified so far share some conserved features [2, 35]. All GPI anchors have the same core structure (Figure 1), with a tetrasaccharide linked to the myo-inositol 6-O-position of phosphatidylinositol (PI), and GPI-APs always have the polypeptide C-terminus attached to the 6-O-position of the mannose residue at the core glycan non-reducing end. The structural diversity of GPIs is reflected by further modifications of the glycan and lipids.

Figure 1.

Representative GPI-AP containing the conserved core structure of GPI anchors.

The conserved GPI core and GPI-protein linkage are determined by the conserved biosynthetic pathways of GPI-APs in various species [5, 36]. GPI biosynthesis and attachment to proteins are realized on/in the endoplasmic reticulum (ER) [37, 38], as depicted in Figure 2A, starting with the transfer of an N-acetylglucosamine (GlcNAc) to PI on the cytoplasmic side of the ER membrane, followed by GlcNAc de-N-acetylation. Thereafter, GlcNH₂-PI is translocated onto the luminal side of the ER membrane to furnish GPI assembly and protein attachment in the ER [37, 38]. The lipid and glycan structures of GPI anchors are further remodelled in the ER, during transport in Golgi, and on the cell surface [39, 40].

Figure 2.

(A) General schemes to depict GPI-AP labelling through metabolic engineering of GPI anchors: inositol and PI derivatives 1, 2, and 3 are incorporated into the biosynthetic pathway of GPIs and GPI-APs, and the resulting GPI-APs on the cell surface contain an azido group that enables further functionalization by click chemistry. (B) Structures of probes 1, 2, and 3 used for metabolic engineering of GPI-APs.

To facilitate GPI-AP study, our group developed inositol derivatives, e.g., 1 and 2 (Figure 2B), as probes for metabolically engineered labelling of GPI-APs. These probes were incorporated by cells into GPI biosynthesis to result in GPI-APs carrying an azido group (Figure 2A). The azide-tagged GPI-APs on the cell surface were further modified by a click reaction to install various labels, e.g., fluorescent and affinity labels [41, 42]. However, these probes have limitations. First, to pass through the cell membrane, they need to be O-acetylated. As a result, after entering the cell, these probes must be processed into free inositol derivative 4 before being useful for GPI biosynthesis (Figure 2A). Second, several steps are required before 4 is converted into PI–the key intermediate in GPI biosynthesis [43-45]. Third, more importantly, these probes are not specific for GPIs. For example, 4 may be used by cells to create derivatives of inositol phosphates [46]. The purpose of this work is to develop and verify a new probe that can overcome the drawbacks for more efficient and selective metabolic engineering of cell surface GPI-APs.

Materials and Methods

General methods:

Commercial chemicals of analytical grade were used without further purification unless noted otherwise. Fetal bovine serum (FBS), Dulbecco’s modified Eagle’s medium (DMEM), and penicillin-streptomycin solution were purchased from ATCC. Dulbecco’s phosphate buffer saline (DPBS), 4′,6-diamidino-2-phenylindole (DAPI), Alexa Fluor (AF) 647-CD55 antibody, AF488-streptavidin, and EZ-Link Dibenzocyclooctyne-PEG12-Biotin (DBCO-biotin) were from ThermoFisher. DPBS (1x) was used for cell washing. Bacillus cereus PI-specific phospholipase C (PI-PLC, EC 3.1.4.10) was from Invitrogen. Cy5-streptavidin was from Abcam. Paraformaldehyde (PFA), glutaraldehyde, and poly-l-lysine were from Sigma Aldrich. Poly-l-lysine solution (0.5%, w/v) was prepared in 1x DPBS. Streptavidin magnetic beads, Magnesphere^® technology magnetic separation stand, and mammalian cell lysis buffer were from Click Chemistry Tools, Promega, and Goldbio, respectively. Staining buffer for fluorescent imaging was DPBS (1x, pH 7.4) with 10% FBS and 0.02% NaN₃. DNA staining buffer was 10 mM Tris-HCl (pH 7.4) with 10 mM EDTA, 100 mM NaCl, and 0.02% NaN₃. Binding solution for streptavidin-mediated protein pull-down was 50 mM Tris-HCl (pH 7.5) with 150 mM NaCl; washing buffer was 50 mM Tris-HCl with 150 mM NaCl and 1 mM biotin; elution buffer was phosphate buffered saline (PBS, pH 7.5) with 100 mM glycine, 2 M urea, 2% sodium dodecyl sulfate (SDS), and 100 mM biotin. Fluorescent images were obtained with an Olympus IX71 inverted system equipped with LED light source (Cool LED, PE-300), 20X 0.8 and 40X 1.4 NA plan apochromatic objectives, DAPI, GFP, and Cy5 fluorescence channels, and Olympus DP23M color camera. Image analysis was performed with Olympus Cellsens Standard 3 software and FIJI/ImageJ software. Flow cytometry was performed using an Attune NxT flow cytometer equipped with blue and red lasers, and data analyses were performed with Attune NxT flow cytometer software. The synthesis of 3 was reported [47].

Cell culture.

HeLa cells were cultured in DMEM supplemented with 10% FBS and 100 U/mL penicillin–streptomycin at 37 °C in 5% CO₂ and 95% air for 3–4 d. Passage three HeLa cells were used for all the experiments.

Metabolic engineering of HeLa cells:

Poly-l-lysine solution in DPBS (0.5%, 1.0 mL) was added in each tissue culture dish (60 mm). After incubation at rt for 1 h, the dishes were washed (DPBS, 2.0 mL, x3) and air-dried. A pellet of 10⁶ cells in 5.0 mL of 10% FBS/DMEM was seeded in each dish, and the dishes were incubated at 37 °C in 5% CO₂ for 12 h. Cells were washed (2.0 mL, x3) and incubated in serum free media (5.0 mL) supplemented with or without 3 (50 or 200 μM) at 37 °C in 5% CO₂ for 12 h. Then, 10% FBS/DMEM (5.0 mL) was added, and the dishes were incubated for another 36 h. Finally, the cells were washed (2.0 mL, x3) and harvested with a cell scraper for various analyses below.

PI-PLC treatment of cells:

Cells obtained above were suspended in PBS supplemented with CaCl₂ (5 mM), MgCl₂ (5 mM) and PI-PLC, and incubated at 20 °C for 1 h. Thereafter, the cells were washed with DPBS three times and subjected to downstream treatments.

Protein extraction:

Protein precipitation using methanol and chloroform was performed following the reported protocol [48]. In brief, methanol, chloroform, and DI water (v/v/v 4:1.5:3) were added to a sample. The mixture was incubated at rt for 5 min with occasional vortex and then centrifuged (21,000x g) at 4 °C for 20 min. The top layer was removed carefully. Methanol (4 volumes) was added to the protein layer, followed by vortex and centrifugation under the same conditions, and then removal of the top layer. This procedure was repeated another time. Proteins were pelleted, air-dried at rt overnight, and resuspended in the protein solubilization buffer (1% SDS and 100 mM NaCl in 1x DPBS). Protein concentrations were determined with a bicinchoninic acid (BCA) assay kit following the manufacturer’s instruction.

Fluorescence microscopy imaging of cells:

Square glass microscopic coverslips were etched in 1 N HCl at rt for 30 min and then in 1 N NaOH for 30 min under sterile conditions. The coverslips were washed with DPBS (x3), treated with 100% ethanol for 30 min, and placed in a 6-well plate. After DPBS washing (1.0 mL), poly-l-lysine solution (0.5%, 1.0 mL) was added in each well, and the plate was incubated at rt for 1 h. After the wells were washed (1.0 mL x3), 50,000 cells were seeded on each coverslip and were allowed to grow in complete growth media (2.0 mL) containing 10% FBS. After reaching ~40% confluence, the cells were washed (1.0 mL) and treated with PBS (control) or 3 (200 μM) as described. Cells on the coverslips were washed (1.0 μL) and incubated with DBCO-biotin in DPBS (200 μM, 0.5 ml) at 4 °C for 45 min, followed by DPBS washing (1.0 μL). The coverslips were incubated with DPBS (1x) containing 4% PFA and 0.1% glutaraldehyde at 4 °C for 20 min, washed with DPBS (1.0 μL x3) and staining buffer (1.0 μL x3), and stored in staining buffer (1.0 mL) at 4 °C. Cy5-streptavidin (1:2000 dilution) was added to cells in staining buffer (0.5 mL), and the plate was sealed with Parafilm, wrapped in aluminum foil, and incubated at 4 °C for 45 min. After DPBS washing (1.0 mL), DAPI (250 ng/mL) in DNA staining buffer (0.5 mL) was added, and the plate was incubated at 4 °C in the dark for 10 min. Coverslips were washed with staining buffer (1.0 mL, x3) followed by adding fixative solution (0.5 mL) and incubation at rt for 5 min. Coverslips were washed with staining buffer (1.0 mL, x3), cleaned with the mounting medium (Fluor mount G, 100 μL), and dried with the cell face downward at rt in the dark overnight. The coverslips were imaged with a fluorescence microscope using 375 and 490 nm lights to excite DAPI and Cy5, respectively.

Flow cytometry:

After treatment with PBS or 3 (200 μM), cells were pelleted by centrifugation (600x g) at 4 °C for 8 min, resuspended in DPBS (1x, 200 μL) containing 200 μM DBCO-biotin, and incubated on ice for 30 min. Cells were washed with staining buffer (1.0 mL) and pelleted by centrifugation, which was repeated three times. The cell pellets were resuspended and incubated with staining buffer (100 μL) containing Cy5-streptavidin (1:2,000 dilution) on ice in the dark for 30 min. The cells were centrifuged and washed with ice cold DPBS (500 μL), which was repeated three times. Finally, the pellet was resuspended in FACS buffer (200 μL) and subjected to FACS analysis using a red (638 nm) excitation laser and 670 nm emission filter. For experiments involving PI-PLC, the only procedural difference was an additional step to treat cells with PI-PLC before DBCO-biotin incubation.

Enzyme-linked immunosorbent assay (ELISA) of labeled GPI-APs:

Cells treated with PBS or 3 (50 μM) were harvested and divided into two groups (ca. 0.61 million/group). To one group, PI-PLC (0.8 U) in PBS was added, and to the other, only PBS was added (control). Both groups were incubated at 20 °C for 1 h. The cells were separated from the supernatants by centrifugation (600x g) at 4 °C for 8 min. The cells were resuspended in mammalian cell lysis buffer (0.5 mL) and incubated on ice for 30 min. The cell lysates and supernatants were separately treated with DBCO-biotin in DPBS (0.4 mM, 0.2 mL) at rt for 4 h, followed by protein extraction using chloroform and methanol. The proteins were resuspended in protein solubilization buffer. Protein samples (2 μg) in bicarbonate buffer (100 μL, pH 9.6) were added to an ELISA plate, followed by incubation at 4 °C overnight and 37 °C for 1 h. The plate was washed (200 μL, x3), blocked (10% BSA in PBST containing 0.5% Tween-20, 100 μL) at rt for 1 h, washed (200 μL, x3), treated with alkaline phosphatase (AP)-streptavidin in DPBS (100 μL, 1:500 dilution) at rt for 2 h, washed (200 μL, x3), and then incubated with p-nitrophenylphosphate (PNPP) in DPBS (0.25 mg/mL, 100 μL) at rt for 30 min. The optical densities (ODs) were measured with a microplate reader at 405 nm.

Western blot analysis of labeled GPI-APs.

Cells treated with PBS or 3 (200 μM) were harvested, washed, resuspended in PBS (0.5 mL) supplemented with 5 mM MgCl₂ and 5 mM CaCl₂, and then divided into two groups (~10⁷ cells/group). One group was treated with PI-PLC (0.8 U), and the other with PBS. After incubation at 20 °C for 1 h, the cells were separated from the supernatant by centrifugation (600x g) at 4 °C for 8 min. The supernatant was condensed to 50 μL via ultrafiltration using a 5 KDa cut-off membrane, diluted with 50 mM Tris-HCl buffer (pH 7.5, 100 μL), and then concentrated again via ultrafiltration. This procedure was repeated another time. The supernatant was concentrated to 20 μL with a SpeedVac concentrator, and an aliquot (10 μL) was subjected to DBCO-biotin treatment in DPBS (100 μM, 20 μL total volume), followed by protein extraction. In the meantime, the cells were washed (1.0 mL, x3), resuspended in mammalian cell lysis buffer (500 μL), and incubated on ice for 45 min. The resultant lysates were centrifuged at 4500x g for 5 min to remove cell debris and concentrated to 100 μL by the above protocol. Protein concentration was estimated using a BCA assay kit. About 25 μg of proteins from each sample was heated with loading dye (1x) at 95 °C for 5 min and then subjected to SDS-PAGE. The proteins in gel were transferred onto PDVF membranes that were washed with TBST, incubated with blocking solution (1 h), washed with TBST (x3) again, and then treated with AP-streptavidin (1:1000 dilution) at 4 °C for 45 min. The membranes were washed, and the protein bands were detected by incubating with BCIP/NBT chromogenic substrate before being photographed.

Western blot analysis of labeled CD55 antigen:

After cells were treated with 3 (200 μM) and then PBS or PI-PLC, the supernatant was separated from the cells and concentrated to 20 μL. An aliquot (10 μL) of the concentrated supernatant was subjected to DBCO-biotin (100 μM, 20 μL) treatment as above. Proteins were extracted, with concentrations estimated by the BCA assay. About 50 μg of proteins from each sample were resuspended in the binding buffer (1.0 mL), and the suspension was incubated with streptavidin magnetic beads (1.5 mg) at 4 °C overnight with end-to-end rotation. The beads were separated with the magnetic separation stand, washed with DPBS (1.0 mL, x2) and washing buffer (1.0 mL, x2), incubated with the elution buffer (30 μL) at 95 °C for 15 min, and then centrifuged (21,000x g) at rt for 10 min. The supernatant was subjected to Western blot by the described protocol using AF647-CD55 antibody to stain.

Results and Discussion

The first dedicated step to GPI/GPI-AP biosynthesis is the enzymatic addition of GlcNAc to the inositol 6-O-position of PI, which occurs on the cytoplasmic side of the ER membrane (Figure 2A). Since targeting the cytoplasmic side of the ER is relatively easy, PI is an excellent entry for metabolic intervention of GPI biosynthesis. Therefore, we have designed a PI derivative 3 (Figure 2B) having the inositol 4-hydroxyl group replaced with an azido group as the probe for GPI-AP metabolic engineering. Compared to 1 and 2, probe 3 also has several other useful features. First, it can save a few metabolic steps via direct involvement in GPI biosynthesis (Figure 2A). Second, compared to the azidoethoxyl group in 1 and 2, the azido group at the inositol 4-C-poistion of 3 is small and a better mimic of the hydroxyl group, thus 3 can imitate natural PI closely. Third, the amphiphilic property of 3 would facilitate its membrane passage and cell uptake, as observed with various glycolipids [49-51], and its targeting to the ER. Fourth, 3 should be more selective than 1 and 2 for GPI engineering, because 3 will not be engaged in the biosynthesis of PI phosphates. In addition, similar to 1 and 2, modifications at the inositol 4-C-position, which is the furthest from 1-, 2- and 6-C-positions involved in GPI synthesis, may maximize the incorporation of 3 in GPI biosynthesis. Therefore, 3 is anticipated to be more efficient and selective than 1 and 2 for the metabolic engineering of cell surface GPI-APs.

Accordingly, we developed a method to synthesize 3 [47] and evaluated its capacity to engineer surface GPI-APs on HeLa cells, an ovarian cancer cell line. First, we investigated whether 3 was incorporated by cells to express azide-labelled GPI-APs. To this end, we incubated cells separately with 3 (200 μM) and PBS (the control), subjected the engineered cells to biotinylation by strain-promoted azide-alkyne cycloaddition (SPAAC) with DBCO-biotin [52, 53], and then treated the cells with fluorophore-streptavidin conjugates. Finally, the cells were analysed by fluorescence microscopic imaging and flow cytometry (FACS).

Fluorescence imaging of 3-treated cells clearly showed their labelling by the green fluorophore AF488, in sharp contrast to cells in the control group (Figure 3A). In addition, the fluorescence was especially concentrated on the cell surface. The strong fluorescence intensity of 3-treated cells also indicated the high efficiency of metabolic engineering. The FACS results (Figures 3B and 3C) of cells treated by the same protocol using Cy5 as the dye gave the same conclusion. For example, the mean fluorescence intensity (MFI) of 3-treated cells was almost 100 folds higher (Figure S2 of Supporting Materials, SM) than that of cells in the control group.

Figure 3.

(A) Fluorescent images of HeLa cells treated with PBS (a to d) or 3 (200 μM, e to h) for 48 h, and then DBCO-biotin (200 μM), AF488-streptavidin, and DAPI. (a)/(e) bright field images; (b)/(f) DAPI-stained cells; (c)/(g) A488-stained cells; (d)/(h) overlay of DAPI and AF488 images. The scale bars are 20 μm. Single parameter FACS histograms (B) and density plots (C) of cells treated with PBS (orange line and dots in B and C) or 3 (black solid line and dots in B and C), and then DBCO-biotin and Cy5-streptavidin under the same conditions as above. The black dotted line in B shows the FACS histogram of cells treated with 3, PI-PLC, DBCO-biotin, and Cy5-streptavidin.

To verify that the FACS signals of 3-treated cells were indeed from Cy5-labeled GPI-APs, we exposed the cells to PI-PLC prior to DBCO-biotin and Cy5-streptavidin treatments. PI-PLC is an enzyme that can selectively cleave GPIs and is widely used to release GPI-APs from cells [54]. As expected, PI-PLC treatment almost completely abolished the fluorescent signals of 3-treated cells (dotted line in Figure 3B and Figure S1 of SM) on DBCO-biotin and Cy5-streptavidin treatments, thereby confirming GPI-anchored molecules as the source of FACS signals. These results strongly support that 3 was effectively taken up and incorporated in GPI anchors by HeLa cells.

Next, we examined the influence of the concentration of 3 on GPI-AP metabolic engineering in HeLa cells. To this end, we incubated cells with PBS (the control) and different concentrations (0.05, 0.20, 0.50, 1.00, and 1.50 mM) of 3, and then with DBCO-biotin and AF488-streptavidin for fluorescence imaging or Cy5-streptavidin for FACS. Both imaging and FACS results (Figures 4A and 4B) exhibited the increase of fluorescence intensity in a probe concentration-dependent manner, indicating the increase of azide-modified GPI-AP molecules on the cell surface alone with increased concentrations of 3. At the relatively low concentrations of 3 (e.g., 50 μM), a superficially small increase of the fluorescent signals in the FACS histogram (by only several folds, Figure 4B and Figure S3 of SM) compared to the control was observed but the fluorescent images (Figure 4A) revealed an efficient labeling of GPI-APs on the cell surface. Therefore, metabolic engineering of cells using a low concentration (50 μM) of 3 should be sufficient for GPI-AP study with sensitive analytical techniques, such as mass spectrometry and immunostaining.

Figure 4.

(A) Overlays of AF488 and DAPI images of HeLa cells treated with PBS (control) or different concentrations (0.05, 0.20, 0.50, 1.00, and 1.50 mM) of 3, and then DBCO-biotin, AF488-streptavidin, and DAPI. The scale bars are 20 μm. (B) Single parameter FACS histograms of HeLa cells treated with PBS or different concentrations of 3, and then DBCO-biotin and Cy5-streptavidin.

To further verify the incorporation of 3 in GPI-APs, we studied the proteins isolated from 3-treated cells by ELISA. In this regard, cells were incubated with PBS or 3 (50 μM), and then PBS or PI-PLC. The cells were separated from the supernatants via centrifugation and lysed. The cell lysates and supernatants were separately treated with DBCO-biotin, which was followed by protein precipitation using chloroform and methanol [48]. The extracted proteins were suspended in PBS containing 1% SDS, with protein concentration estimated by the BCA assay. The obtained proteins were applied to ELISA plates. After washing and blocking, the plates were incubated with AP-streptavidin, followed by PNPP (an AP substrate) and analysis with a microplate reader at 405 nm. The lysates of 3-treated cells gave a significantly higher OD than that of the control (Figure 5A). Moreover, PI-PLC treatment before cell lysis and biotin labeling resulted in a significant decrease in the OD value of cell lysates. Both findings support the presence of azide-labeled GPI-APs on 3-treated cells. ELISA results of the cell supernatants (Figure 5B) provided additional supports to this conclusion. For example, the OD value difference of the supernatants of the control (3/PI-PLC −/−) and 3-treated (3/PI-PLC +/−) cells was statistically insignificant, but the OD value difference of the supernatants of 3/PBS- (3/PI-PLC +/−) and 3/PI-PLC-treated (3/PI-PLC +/+) cells was high and statistically significant. These results suggested a PI-PLC-mediated release of azide-labeled GPI-APs into the supernatants.

Figure 5.

ELISA results of the lysates (A) and supernatants (B) of cells treated with PBS only (the control: 3/PI-PLC −/−), 3 but no PI-PLC (+/−), and both 3 and PI-PLC (+/+), followed by DBCO-biotin treatment, protein extraction, and ELISA. Data are reported as mean ± standard deviation of three replicates. ***: P < 0.005; **: P < 0.01; ns: no statistical significance.

We also utilized Western blot to investigate 3-mediated metabolic engineering of HeLa cells. In this experiment, the lysates of 3/PBS (3/PI-PLC +/−)-treated cells (Lane 3, Figure 6A left) and the supernatant of 3/PI-PLC (+/+)-treated cells (Lane 5) were treated with DBCO-biotin to label the azide-modified GPI-APs, followed by analysis with SDS-PAGE and AP-streptavidin staining. As indicated in Figure 6A (left), both lanes showed a series of biotinylated protein bands, which was different for cells without 3 treatment (3/PI-PLC −/+, Lane 2), while all three lanes contained the similar quantity of total proteins shown by Coomassie blue staining (Figure 6A, right). Similar results were obtained in a duplicate experiment (Figure S4 of SM). These studies provide the direct evidence for 3 incorporation into GPI anchors to express azide-labeled GPI-APs.

Figure 6.

(A) Western blot (left) and Coomassie blue staining (right) of proteins in the supernatants of cells treated with PBS/PI-PLC (3/PI-PLC −/+, Lanes 2 and 2’, control) or 3/PI-PLC (3/PI-PLC +/+, Lanes 5 and 5’) and in the lysates of cells treated with 3/PBS (3/PI-PLC +/−, Lanes 3 and 3’), followed by DBCO-biotin incubation, protein precipitation, SDS-PAGE, and staining with AP-streptavidin/PNPP or Coomassie blue. (B) Western blot of CD55 expressed by 3-treated HeLa cell. After cells were treated with 3/PBS (3/PI-PLC +/−) or 3/PI-PLC (+/+), the supernatants were separated from cells and subjected to DBCO-biotin treatment and protein precipitation. The extracted proteins were resuspended and enriched with streptavidin magnetic beads, followed by heating with biotin at 95 °C to release biotinylated proteins. Proteins were analyzed by SDS-PAGE and Western blot using AF647-CD55 antibody. PM: protein marker; BK: blank.

Finally, we examined a specific GPI-AP—the CD55 antigen (70-75 KDa), complement decay accelerating factor (DAF) that regulates the complement system [55]—to further verify 3-mediated metabolic engineering of cell surface GPI-APs. In this context, the supernatants of cells treated with 3 (3/PI-PLC +/−, the control) or 3 and PI-PLC (3/PI-PLC +/+) were applied to reaction with DBCO-biotin. Biotinylated GPI-APs were pulled down from the methanol/chloroform-precipitated protein pools with streptavidin magnetic beads and then released from the beads via competition with free biotin along with heating following the reported protocol [56]. The released proteins were subjected to SDS-PAGE and blotting with AF647-CD55 antibody. A clear band of biotinylated CD55 with the 74 KDa molecular weight (Figure 6B) was observed to directly prove the metabolic incorporation of azido-PI in CD55, and most likely in other cell surface GPI-APs as well.

Conclusion

We have developed a new method for metabolic labeling of cell surface GPI-APs. This method is based on an azido-PI derivative 3, which closely mimics the structure of natural PI and can act as a biosynthetic precursor of GPIs and GPI-APs. Using HeLa cell as the model, we have shown by several biological assays that 3 is effectively taken up by cells and incorporated in GPI anchors to express azide-tagged GPI-APs. This finding is consistent with the recent report that a synthetic PI analog could rescue GPI-AP biosynthesis [57]. The azide-tagged GPI-APs present on the cell surface can be further elaborated by biorthogonal click reactions to introduce other functionalities, e.g., fluorescent and affinity labels, to facilitate the visualization and investigation of GPI-APs on live cells. Compared to our first-generation biosynthetic precursors of GPI anchors (e.g., 1 and 2) for GPI-AP metabolic engineering, 3 has several advantages. Therefore, it can be a powerful tool for the exploration of various issues about GPI-APs, such as GPI-AP trafficking and organization on the cell surface. It also opens a window to label cell surface GPI-APs with an affinity tag (e.g., biotin) to facilitate the pull-down and proteomics analysis of GPI-APs and discovery of new GPI-AP biomarkers, which is currently pursued in our laboratory.

Highlights.

• A new method for metabolic engineering of GPI-anchored proteins (GPI-APs)

• Labeling GPI-APs on live cells with an azide to enable further functionalization

• Labeling GPI-APs on cells with fluorophores to enable fluorescence-based studies

• Labeling GPI-APs on cells with an affinity tag to facilitate proteomics analysis