Cancer Malignancy Is Correlated with Upregulation of PCYT2-Mediated Glycerol Phosphate Modification of α-Dystroglycan

The dystrophin–glycoprotein complex connects the cytoskeleton with base membrane components such as laminin through unique O-glycans displayed on α-dystroglycan (α-DG). Genetic impairment of elongation of these glycans causes congenital muscular dystrophies. We previously identified that glycerol phosphate (GroP) can cap the core part of the α-DG O-glycans and terminate their further elongation. This study examined the possible roles of the GroP modification in cancer malignancy, focusing on colorectal cancer. We found that the GroP modification critically depends on PCYT2, which serves as cytidine 5′-diphosphate-glycerol (CDP-Gro) synthase. Furthermore, we identified a significant positive correlation between cancer progression and GroP modification, which also correlated positively with PCYT2 expression. Moreover, we demonstrate that GroP modification promotes the migration of cancer cells. Based on these findings, we propose that the GroP modification by PCYT2 disrupts the glycan-mediated cell adhesion to the extracellular matrix and thereby enhances cancer metastasis. Thus, the present study suggests the possibility of novel approaches for cancer treatment by targeting the PCYT2-mediated GroP modification.


Introduction
Epithelial cells interact with the basement membrane, a sheet-like type of extracellular matrix (ECM), through cell adhesion molecules, which regulate cellular growth, motility, and differentiation by integrating signals from ECM or soluble factors [1,2]. One of the most extensively studied cell adhesion molecules is α-dystroglycan (α-DG), which is displayed 2 of 15 on epithelial cell surfaces as a component of the dystrophin-glycoprotein complex [3,4]. This complex connects the cytoskeleton with basement membrane components such as laminin, agrin, and perlecan and plays an essential role in epithelium development and tissue organization. The connection critically depends on unique O-glycans, termed matriglycans, attached to the N-terminal region of the mucin-like domain of α-DG [5][6][7].
We recently found that glycerol phosphate (GroP) modifies the 6-phosphorylated core M3 structure without further elongation [20]. The dual action of FKTN catalyzes this novel modification and, therefore, competes for the first Rbo5P conjugation [20,21]. Furthermore, the GroP modification is enhanced by elevation of intracellular CDP-Gro level, which can be artificially realized by overexpression of TagD, a CDP-Gro synthase from Gram-positive bacteria [22,23], resulting in a concomitant decrease of the matriglycans [24]. These findings raise the possibility that GroP serves as a terminator of the matriglycan elongation. However, the physiological and pathological functions of this modification remain largely unknown.
Mutations in causative genes such as LARGE and FKRP often result in defect or decreased expression of matriglycans [9][10][11][12][13][14][15][16]. On the other hand, matriglycans have been shown to be reduced or abolished in primary human cancer cells and cell lines, including prostate, breast, and colorectal cancers [25][26][27][28][29]. Interestingly, the defects in the matriglycans of α-DG are correlated with poor prognosis [30,31]. Since overexpression of LARGE significantly inhibits the cancer cell migration ability [32,33] and low expression of B4GAT1 results in increased cancer cell invasion, altered glycosylation of α-DG is considered to be associated with cancer development and progression [34]. However, the regulation mechanisms of matriglycan formation in human cancers remain to be uncovered.
In this study, we attempt to determine the possible roles of the GroP modification in cancer malignancy. We reveal that the GroP modification was promoted in several cancer tissues and address the molecular mechanism and pathological significance of this modification.

GroP Modification Is Enhanced in Cancer Tissues
We previously established a monoclonal antibody DG2, reactive with GroP-modified α-DG [24]. Using this antibody, we probed the GroP modification in various normal and cancerous tissues. Several cancerous tissues, including the bladder, uterus, ovary, and colon, tested positive for immunostaining compared to their normal counterparts (Table S1). Focusing on colorectal cancer, we investigated a relationship between degrees of GroP modification and malignancy ( Figure 1). GroP expression in human colorectal cancer (CRC) tissues was analyzed using immunohistochemistry with DG2 for primary tumor tissues with surgical resection. Representative images of DG2 staining are shown in Figure 1A. The 100 CRC patients consisted of 14 patients in stage 0, 20 patients in stage I, 21 patients in stage II, 22 patients in stage III, and 23 patients in stage IV. The stage IV CRC showed a significantly higher rate of positive GroP expression than the other stages (stage 0, 14% (2/12); stage I, 25% (5/20); stage II, 43% (9/21); stage III, 36% (8/22); stage IV, 74% (17/23)) ( Figure 1B). Notably, no immunoreactivity to DG2 was observed in normal colorectal tissues. These data indicate that GroP modification is promoted as cancer progresses. (17/23)) ( Figure 1B). Notably, no immunoreactivity to DG2 was observed in normal colorectal tissues. These data indicate that GroP modification is promoted as cancer progresses.

GroP Modification Promotes Cancer Migration
To address the GroP modification function in cancer, we examined the possible effects of the GroP modification enhanced by TagD overexpression on the proliferation and migration of HCT116 cells as a model of colorectal cancer cells. While the CCK-8 assay showed virtually no effect on cell proliferation, the migration ability was enhanced upon TagD overexpression, as demonstrated by wound-healing and transwell assays ( Figure 2). These data indicated that enhancement of the GroP modification promoted cancer cell migration, which could be caused by an increase in CDP-Gro amount in cells.

GroP Modification Promotes Cancer Migration
To address the GroP modification function in cancer, we examined the possible effects of the GroP modification enhanced by TagD overexpression on the proliferation and migration of HCT116 cells as a model of colorectal cancer cells. While the CCK-8 assay showed virtually no effect on cell proliferation, the migration ability was enhanced upon TagD overexpression, as demonstrated by wound-healing and transwell assays ( Figure 2). These data indicated that enhancement of the GroP modification promoted cancer cell migration, which could be caused by an increase in CDP-Gro amount in cells. (17/23)) ( Figure 1B). Notably, no immunoreactivity to DG2 was observed in normal colorectal tissues. These data indicate that GroP modification is promoted as cancer progresses.

GroP Modification Promotes Cancer Migration
To address the GroP modification function in cancer, we examined the possible effects of the GroP modification enhanced by TagD overexpression on the proliferation and migration of HCT116 cells as a model of colorectal cancer cells. While the CCK-8 assay showed virtually no effect on cell proliferation, the migration ability was enhanced upon TagD overexpression, as demonstrated by wound-healing and transwell assays ( Figure 2). These data indicated that enhancement of the GroP modification promoted cancer cell migration, which could be caused by an increase in CDP-Gro amount in cells.

Subsubsection PCYT2 Synthesizes CDP-Gro
We next attempted to identify mammalian enzymes that catalyze the CDP-Gro synthesis, assuming their homology with bacterial TagD. From the human genome database, we found five genes, CDS1, CDS2, PCYT1A, PCYT1B, and PCYT2, that share >30% homology with the TagD gene (Table 1). We produced recombinant proteins encoded by these genes in HCT116 cells and evaluated their CDP-Gro synthetic activity in vitro ( Figure 3A). Consequently, we found that only PCYT2 could synthesize CDP-Gro, while the remaining four proteins had no activity. To date, PCYT2 has been characterized as ethanolamine-phosphate cytidylyltransferase, catalyzing the formation of CDP-ethanolamine (CDP-Etn) [35,36]. This enzyme has two major splicing isoforms, PCYT2α and PCYT2β, which were shown to be comparable in terms of the enzymatic activities to CDP-Gro synthase as well as CDP-Etn synthase ( Figure 3B and Table 2).

Subsubsection PCYT2 Synthesizes CDP-Gro
We next attempted to identify mammalian enzymes that catalyze the CDP-Gro synthesis, assuming their homology with bacterial TagD. From the human genome database, we found five genes, CDS1, CDS2, PCYT1A, PCYT1B, and PCYT2, that share >30% homology with the TagD gene (Table 1). We produced recombinant proteins encoded by these genes in HCT116 cells and evaluated their CDP-Gro synthetic activity in vitro (Figure 3A). Consequently, we found that only PCYT2 could synthesize CDP-Gro, while the remaining four proteins had no activity. To date, PCYT2 has been characterized as ethanolamine-phosphate cytidylyltransferase, catalyzing the formation of CDP-ethanolamine (CDP-Etn) [35,36]. This enzyme has two major splicing isoforms, PCYT2α and PCYT2β, which were shown to be comparable in terms of the enzymatic activities to CDP-Gro synthase as well as CDP-Etn synthase ( Figure 3B and Table 2).
Furthermore, to investigate whether PCYT2 actually synthesizes CDP-Gro in cells, we established PCYT2-KO cells (Supplementary Figure S1) and quantified intracellular CDP-Gro and CDP-Etn in the cells based on their extraction ion chromatograms ( Figure  4A). As expected, CDP-Etn was markedly reduced in the PCYT2-KO cells ( Figure 4B). Moreover, CDP-Gro was undetectable in the PCYT2-KO cells. Furthermore, the CDP-Gro level was recovered in the PCYT2-KO cells rescued by overexpressing PCYT2 ( Figure 4B). These results indicate that PCYT2 is the enzyme responsible for the synthesis of CDP-Gro in mammalian cells. Ethanolamine-phosphate cytidylyltransferase 30% 89%  Table 1 and positive (TagD) and negative controls. The peak corresponding to CDP-Gro is indicated with the arrow. (B) The enzymatic activities of PCYT2α and PCYT2β with various glycerol-3-phosphate or ethanolamine   Figure S1) and quantified intracellular CDP-Gro and CDP-Etn in the cells based on their extraction ion chromatograms ( Figure 4A). As expected, CDP-Etn was markedly reduced in the PCYT2-KO cells ( Figure 4B). Moreover, CDP-Gro was undetectable in the PCYT2-KO cells. Furthermore, the CDP-Gro level was recovered in the PCYT2-KO cells rescued by overexpressing PCYT2 ( Figure 4B). These results indicate that PCYT2 is the enzyme responsible for the synthesis of CDP-Gro in mammalian cells.

GroP Modification of α-DG Critically Depends on PCYT2
We overexpressed the Fc-fused α-DG recombinant protein with amino acid substitution and deletion of α-DG (α-DG373(T322R)-Fc) in the wild-type (WT) and PCYT2-KO cells. We analyzed their GroP modifications by nanospray liquid chromatography-tandem mass spectrometry (LC-MS/MS) ( Figure 5). The data demonstrated that the GroP modification of matriglycans was not detected in the PCYT2-KO cells in contrast to the WT cells. In addition, the overexpression of PCYT2 resulted in a significant reduction in the matriglycans on α-DG ( Figure 6). These data suggested the critical involvement of PCYT2 in the GroP modification of α-DG by supplying CDP-Gro as a donor substrate.
We overexpressed the Fc-fused α-DG recombinant protein with amino tion and deletion of α-DG (α-DG373(T322R)-Fc) in the wild-type (WT) an cells. We analyzed their GroP modifications by nanospray liquid chromat dem mass spectrometry (LC-MS/MS) ( Figure 5). The data demonstrated t modification of matriglycans was not detected in the PCYT2-KO cells in c WT cells. In addition, the overexpression of PCYT2 resulted in a significan the matriglycans on α-DG ( Figure 6). These data suggested the critical in PCYT2 in the GroP modification of α-DG by supplying CDP-Gro as a donor We detected the expression of PCYT2 in serial sections of stage IV CRC the anti-PCYT2 antibody. Of 23 stage III CRCs, 13 CRCs revealed positive PCYT2. Interestingly, the immunostaining score was significantly correla GroP and PCYT2 expressions in the CRC tissues (r = 0.614, p = 0.002) (Figure

Discussion
The glycan-mediated cell-cell communications provide the basis of the higher functions of our body, which are represented by the immune and nervous systems [37,38]. Therefore, the glycans have to be properly displayed on cell surfaces, and glycosylation abnormalities cause disorders of cellular society, resulting in severe pathological symptoms, as exemplified by cancer malignant transformation [39]. The glycan formation involves a series of glycosyltransferases whose up-and downregulations can control glycan elongation and consequent cellular interactions and, moreover, sophisticate functions [40]. In the case of matriglycan formation on α-DG, at least nine different glycosyltransferases play indispensable roles, and their misfunctions result in CMD. Recently, we found that GroP modification provides an alternative mode of regulation of elongation of  Mapping of correlation between GroP modification and PCYT2 expression levels based on the staining scores. Staining scores of both GroP and PCYT2 were calculated at stage IV colorectal cancer tissues, and data were analyzed using the Spearman rank correlation. Dot size corresponds to the number of colorectal cancer tissue samples.

Discussion
The glycan-mediated cell-cell communications provide the basis of the higher functions of our body, which are represented by the immune and nervous systems [37,38]. Therefore, the glycans have to be properly displayed on cell surfaces, and glycosylation abnormalities cause disorders of cellular society, resulting in severe pathological symptoms, as exemplified by cancer malignant transformation [39]. The glycan formation involves a series of glycosyltransferases whose up-and downregulations can control glycan elongation and consequent cellular interactions and, moreover, sophisticate functions [40]. In the case of matriglycan formation on α-DG, at least nine different glycosyltransferases play indispensable roles, and their misfunctions result in CMD. Recently, we found that GroP modification provides an alternative mode of regulation of elongation of

Discussion
The glycan-mediated cell-cell communications provide the basis of the higher functions of our body, which are represented by the immune and nervous systems [37,38]. Therefore, the glycans have to be properly displayed on cell surfaces, and glycosylation abnormalities cause disorders of cellular society, resulting in severe pathological symptoms, as exemplified by cancer malignant transformation [39]. The glycan formation involves a series of glycosyltransferases whose up-and downregulations can control glycan elon-gation and consequent cellular interactions and, moreover, sophisticate functions [40]. In the case of matriglycan formation on α-DG, at least nine different glycosyltransferases play indispensable roles, and their misfunctions result in CMD. Recently, we found that GroP modification provides an alternative mode of regulation of elongation of the α-DG O-glycans by capping the core M3 structure [20]. This modification is catalyzed by the Janus-like enzyme FKTN, which also transfers Rbo5P and thereby positively contributes to the glycan elongation [21], suggesting some additional mechanism for selectively regulating the modification by GroP as a glycan termination factor.
Our immunohistochemical analysis for GroP modification in human colorectal cancer tissues indicated a significantly positive correlation between colorectal cancer progression and GroP modification. The data are consistent with a reduction in expression levels of matriglycan observed in high-grade tumors in the previous studies [26,30]. Matriglycans are responsible for the formation of basement membranes in tumor invasion through the interaction of ECM proteins such as laminin [41]. For example, an increase in amount of matriglycan by overexpression of LARGE results in suppression of the migration ability of cancer cells [32,33]. In addition, depletion of matriglycans by lowering the expression of B4GAT1 promotes cell migration and thereby increases tumor formation [34]. It has also been pointed out that inadequate expression of the matriglycans activates the integrinmediated AKT/ERK pathway involved in cancer migration [34]. Because integrins are competitive with α-DG in terms of laminin-binding, the loss of the matriglycans of α-DG enhances the integrin-laminin interactions, thereby enhancing the signaling leading to cancer metastasis.
This study identified PCYT2 as a CDP-Gro synthase and that the GroP modification critically depends on this enzyme. We furthermore found a significantly positive correlation between GroP modification and PCYT2 expression. Moreover, we demonstrate that GroP modification promotes the migration of cancer cells. Finally, we integrate these findings into a cancer metastasis model wherein the GroP modification by PCYT2 suppresses formation of matriglycans and thereby abrogates the glycan-mediated cell adhesion to ECM, thereby promoting migration and invasion of cancer cells. Lu et al. showed that matriglycans in the breast cancer cell line MCF7 increased upon culturing with the ribitol as a supplement [27]. The enhancement of matriglycan was supposed to result from increased level of CDP-ribitol (CDP-Rbo). The present study demonstrates that overproduction of CDP-Gro causes depletion of matriglycans by capping the non-reducing end of M3 structure through GroP modification. These data suggest that the balance between CDP-Gro and CDP-Rbo in the cells is an important determining factor of the elongation of matriglycans.
Thus, our findings raise the possibility of novel approaches for cancer treatment by antagonizing GroP modification. Particularly, PCYT2 can be a potential target for developing anticancer drugs. However, the present study indicates that PCYT2 is also a dual-function enzyme that synthesizes both CDP-Gro and CDP-Etn, known to be a vital precursor in the biosynthesis of phosphatidylethanolamine [42]. Therefore, it is important to explore how these two enzymatic activities are regulated from an anticancer drug discovery perspective. An alternative approach will be developing therapeutic antibodies targeting the GroP groups displayed on malignant cancer cells.
The present study highlights the pathological aspect of GroP modification in the context of cancer metastasis. Conversely, it is also important to elucidate the physiological roles of this modification. Lymphocytes need to be freed from the adhesion to the ECM for migrating throughout our body. Indeed, T cells have been reported to lose their matriglycans upon maturation [43]. An intriguing possibility is that the PCYT2-mediated GroP modification is involved in such regulations. Further elucidation of the regulatory mechanisms of this modification system will contribute to a deeper understanding of the physiological control of cell adhesion and also to develop therapeutic approaches for a disorder of the system. While preparing this manuscript, Dr. Endo and his coworkers published a report showing that PCYT2 was a mammalian CDP-Gro synthase [44], consistent with our findings.

Antibodies
The following antibodies were used in this study: anti-Flag (

Patients
One hundred colorectal cancer (CRC) patients who underwent surgical and endoscopic resection at Nagoya City University Hospital from March 2018 to December 2019 were enrolled in the present study. The clinical stage was determined based on the final pathological diagnosis after resection, according to the 7th edition of the UICC-TNM classification. The study protocol was approved by the institutional review board (IRB) of Nagoya City University (IRB #. 60-19-0047), and it was conducted following the ethical guidelines of the 1975 Declaration of Helsinki (6th revision, 2008).

Immunohistochemistry
Immunohistochemical staining was performed using serial sections of each sample as follows. All tumor samples were resected from primary CRC tissues, fixed in formalin, and embedded in paraffin. Consecutive sections (4 µm thick) were deparaffinized and dehydrated. After inhibition of endogenous peroxidase activity by immersion in 3% H 2 O 2 /methanol solution, antigen retrieval was achieved by heating the samples in 10 mM citrate buffer (pH 6.0) in a microwave oven for 10 min in at 98 • C. Sections were then incubated with primary antibodies against 20 µg/mL of the anti-GroP DG2 antibody and anti-PCYT2 antibody at dilution of 1:200. After thorough washing, samples were incubated with biotinylated secondary antibodies and then with avidin-biotin horseradish peroxidase complexes. Finally, immune complexes were visualized by incubation in 3,3'diaminobenzidine tetrachloride. Nuclear counterstaining was accomplished using Mayer's hematoxylin. All immunostained specimens were assessed by an independent observer blinded to all clinical information. When more than 10% of all cancer cells in each section were stained, the staining intensity score was classified as negative (0), weak (1), moderate (2), or strong (3). Finally, immunostaining scores of (2) and (3) were defined as positive. In the statistical analysis, the binary data were analyzed using the chi-squared test or Fisher's exact probability test, as appropriate. The nonparametric Spearman's rank correlation coefficient (r) was used as a correlation measure. All statistics were calculated using SPSS Statistics version 25 (IBM Corp., Tokyo, Japan). A two-tailed p-value of less than 0.05 was considered statistically significant.

Western Blot
Western blots were performed as described previously [20,24]. Recombinant proteins or cell lysates were subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and subsequently transferred to polyvinylidene difluoride membranes (Merck Millipore, Burlington, MA, USA). After blocking with Blocking One solution (Nacalai Tesque, Tokyo, Japan), the membranes were incubated with primary antibodies, followed by incubation with respective horseradish peroxidase (HRP)-conjugated secondary antibodies. For the detection of α-DG, the lysates were enriched by wheat germ agglutinin (WGA)agarose (Sigma-Aldrich) and subjected to Western blot analysis. The protein bands were developed with Immobilon Western Chemiluminescent HRP substrate solution (Millipore) and were imaged with an Amersham™ Imager 600 (GE Healthcare, Chicago, IL, USA). After removing the antibodies or avidin by Western Blot Stripping Buffer (Takara Bio Inc.), the membrane was reproved using different antibodies.

Cell Proliferation Assay
The cell counting kit-8 (CCK-8 ® ) (Dojindo, Kumamoto, Japan) was used for the cell proliferation assay, according to the manufacturer's instructions. Briefly, at 24 h after transfection with TagD expression or control vectors, 104 cells per well were seeded in 96-well plates and then incubated at 37 • C. After 0, 24, 48, and 72 h, 10 µL of CCK-8 ® solution was added to each well and incubated at 37 • C for 2 h. The formazan formation was assessed using a plate reader (Nivo S, PerkinElmer, Waltham, MA, USA) at 450 nm.

Cell Migration Assay
Migratory ability of HCT116 cells was examined by wound-healing and transwell migration assays. A scratch wound assay evaluated cell migration and motility. At 24 h after transfection of TagD expression or control vectors, the transfected cells overgrown in the 33 mm glass-bottom dishes (Iwaki, Tokyo, Japan) were coated with poly-L-lysin. Then, the monolayer cells were scratched by sterile pipette tips uniformly. The cell mobility was monitored for 24 h using a Keyence BZ-X800 microscope (Keyence, Osaka, Japan) equipped with a time-lapse module. The wound closure areas were measured with ImageJ software (1.48 q, Rayne Rasband, National Institutes of Health, Bethesda, MA, USA). Cell migratory ability of wound-healing was assessed using the following formula: [(wound area at 0 h)-(wound area at indicated 24 h)]/(wound area at 0 h). A higher score indicates a better migratory ability.
In transwell assay, cells were placed onto 24-well transwell chambers with a pore size of 8 µm (FALCON). The cells had migrated to the reverse face of the membrane. A total of 24 h after transfection with TagD expression or control vectors, the cells were seeded at 1.5 × 10 5 cells/well in 200 µL medium (supplemented with 0% FBS) in the upper chamber. At the same time, 500 µL medium (supplemented with 10% FBS) was added in the lower chamber. After incubating for 24 h, the cells on the reverse face were fixed in 4% paraformaldehyde phosphate buffer solution (Wako) and then stained by 1% Crystal Violet (Sigma-Aldrich) and 2% ethanol.

HPLC Analysis of In Vitro Enzymatic Reactions
CDP-Gro synthetic activity was examined using a mixture of 625 µM glycerol-3phosphate, 300 µM CTP, 1 mM MgCl2, 1 mM DTT, and~12 µg/mL PCYT2 or one of the other candidate enzymes in 4 mM Tris-HCl (pH 7.5). The mixture was incubated at 37 • C overnight. Enzymatic activities of PCYT2 were quantified using a mixture of varying concentrations of ethanolamine phosphate or glycerol-3-phosphate, 300 µM CTP, 1 mM MgCl2, 1 mM DTT, and 72 µg/mL of PCYT2α or 24 µg/mL of PCYT2β in 4 mM Tris-HCl (pH 7.5). The reaction mixture was incubated at 37 • C for 3 h. After quenching the reaction by boiling at 94 • C for 5 min, the reaction mixture was twofold diluted with acetonitrile and subjected to HPLC analysis as described previously (17). CDP-Gro was measured by UV absorbance at 269 nm. The elution positions of CDP-Gro and CDP-ethanolamine were identified using authentic standards (Sigma-Aldrich).

CRISPR/Cas9 Genome Editing
The pSpCas9(BB)-2A-Puro (PX459) vector (#48139) was used for the cloning via BbsI site and the expression of single target gRNAs to generate the KO cells deficient in PCYT2, with target sequences as follows: PCYT2, 5 -CGTTGTCCTTTTCCTAGAGG-3 and 5 -GTACAGGTGAGTCTCCACCG-3 . The two constructed pSpCas9(BB)-2A-Puro (PX459) vectors against individual genes were transiently transfected into HCT116 cells in a six-well plate using polyethyleneimine "Max" (Polysciences, Inc.). At 48 h post-transfection, the cells were treated with puromycin (Thermo Fisher Scientific). Knockout cells were selected with 7.5 µg/mL puromycin for 5 days, then cultured in medium without puromycin for 2-3 weeks. The cells were replated as single cells in 6 cm dishes. Deletions of target genes in the clones were confirmed by PCR, sequencing analysis, and Western blotting ( Figure S1).

Stable Cell Line Generation
PCYT2-overexpressing cells and PCYT2-KO rescue cells were generated by 600 µg/mL G418 (Thermo Fisher Scientific) selection for 14 days after the transfection with a Flagtagged PCYT2 vector. The expression of PCYT2 in G418-resistant cells was checked by Western blot using an anti-PCYT2 antibody ( Figure S3).
For transposon-mediated gene transfer, cells were transfected with the Tol2 transposase expression vector (pCAGGS-T2TP) and the donor vector (pT2K-CAGGS-rtTA-PCYT2) using PEI-Max. One day after the transfections, cells were selected with 7.5 µg/mL puromycin for 5 days, then cultured in a medium without puromycin for 2 weeks.

LC-MS/MS Analysis of Nucleotide Derivatives
The HILIC-ESI-MS/MS method was used [45,46] to determine the abundances of CDP-glycerol and CDP-ethanolamine in cultured cells. Briefly, cultured cells plated on a 6 cm dish were collected in ice-cold 70% ethanol (1.5 mL) and spiked with an unnatural GDP-glucose (1 nmol). The extracts were centrifuged at 16,000× g for 15 min at 4 • C, and the supernatants were lyophilized. As reported previously [46], the samples were dissolved in 2 mL of 10 mM NH 4 HCO 3 and purified on an Envi-Carb column. The 70% ethanol extraction precipitate was dissolved in 2% SDS to measure the protein concentrations, followed by quantifying protein concentration using Pierce BCA Protein Assay Kit (#23227).

LC-MS/MS Analysis of Glycopeptides
Glycopeptides derived from α-DG were identified by LC-MSMS analyses as previously described [20,24]. Briefly, the Fc-fused α-DG [α-DG 373(T322R)-Fc] were expressed in HCT116 and the PCYT2-KO cells. After purification of the recombinant proteins followed by separation using SDS-PAGE, the gel bands containing α-DG373(T322R)-Fc were excised and subjected to reduction, alkylation, and then in-gel digestion by trypsin (Promega, Madison, WI, USA). Digested peptides purified using a ZipTip C18 (Merck Millipore) were analyzed by LC-MS/MS on an Orbitrap Fusion Tribrid (Thermo Fisher Scientific) coupled to an Easy-nLC 1200 (Thermo Fisher Scientific). The identities of glycopeptides were then manually verified based on positive detection of the expected peptide pyrQIHATPTPVR b and y ions, and/or the peptide core ion at m/z 551.804 (z = 2) or 1102.600 (z = 1) in their respective HCD MS/MS spectra, as described previously [20,24].

Institutional Review Board Statement:
The study was conducted in accordance with the Declaration of Helsinki and approved by the institutional review board (IRB) of Nagoya City University (IRB #. 60-19-0047).

Informed Consent Statement:
Written general consent that included research uses of clinical data and tumor tissues had been obtained from all patients.

Data Availability Statement:
The data presented in this study are available on request from the corresponding authors. The mass spectrometric data of LC-MS have been deposited to the Mass Spectrometry Interactive Virtual Environment (MassIVE) with the dataset ID: MSV000089635.