N-Glycans: Phenotypic Homology and Structural Differences between Myocardial Cells and Induced Pluripotent Stem Cell-Derived Cardiomyocytes

Cell surface glycans vary widely, depending on cell properties. We hypothesized that glycan expression on induced pluripotent stem cells (iPSCs) might change during cardiomyogenic differentiation toward the myocardial phenotype. N-glycans were isolated from iPSCs, iPSC-derived cardiomyocytes (iPSC-CM), and original C57BL/6 mouse myocardium (Heart). Their structures were analyzed by a mapping technique based on HPLC elution times and MALDI-TOF/MS spectra. Sixty-eight different N-glycans were isolated; the structures of 60 of these N-glycans were identified. The quantity of high-mannose type (immature) N-glycans on the iPSCs decreased with cardiomyogenic differentiation, but did not reach the low levels observed in the heart. We observed a similar reduction in neutral N-glycans and an increase in fucosylated or sialyl N-glycans. Some structural differences were detected between iPSC-CM and Heart. No N-glycolyl neuraminic acid (NeuGc) structures were detected in iPSC-CM, whereas the heart contained numerous NeuGc structures, corresponding to the expression of cytidine monophosphate-N-acetylneuraminic acid hydroxylase. Furthermore, several glycans containing Galα1-6 Gal, rarely identified in the other cells, were detected in the iPSC-CM. The expression of N-glycan on murine iPSCs changed toward the myocardial phenotype during cardiomyogenic differentiation, leaving the structural differences of NeuGc content or Galα1-6 Gal structures. Further studies will be warranted to reveal the meaning of the difference of N-glycans between the iPSC-CM and the myocardium.


Introduction
In vitro generation of cardiac myocytes by reprogramming is a promising technology in developing cell-transplant therapy for advanced cardiac failure [1] and drug discovery for a variety of cardiac diseases [2]. For both purposes, induced pluripotent stem cells (iPSCs) are most useful, since generation and cardiomyogenic differentiation of iPSCs has been standardized in human and a number of animals [3,4]. In fact, derivatives of iPSCs have been developed to the pre-clinical stage for cell transplantation therapy [5], while cardiac myocytes generated from patient-specific iPSCs have been studied to explore pathologic mechanisms and guide drug discovery [6,7]. However, cardiac myocyte preparations from iPSCs contain immature phenotypes, observed by electrophysiology, electron microscopy, and immunohistochemistry [8,9]; this may limit the safety and efficacy of cell transplantation therapy or reduce the accuracy and efficiency of drug discovery. The maturity of iPSC-derived cardiac myocytes (iPSC-CMs) has not been comprehensively or quantitatively evaluated.
Cell surface glycans have several important functions interacting with numerous proteins, including growth factors, morphogens and adhesion molecules, modulating dynamic cellular mechanisms such as cell-cell adhesion, cell activation, and malignant alterations [10][11][12]. In early mammalian embryos, associated with fertilization, some N-glycans play important roles of cell-cell adhesion [13][14][15]. In addition, cellular responsiveness to growth or arrest depends on total N-glycan number and the degree of branching of cell surface glycoproteins [16]. Furthermore, heparan sulfate, a kind of glycans, is required for embryonic stem cell (ESC) pluripotency, in particular lineage specification into mesoderm through facilitation of FGF and BMP signaling by stabilizing BMP ligand [17], leading the evidence that the expression patterns of cell surface glycans on ESCs changes during differentiation [18]. Thus, we hypothesized that cell surface glycan expression may change during the course of cardiomyogenic differentiation of iPSCs in vitro. We analyzed N-glycan expression in undifferentiated iPSCs, iPSC-CMs, and adult murine myocardium by HPLC, to identify potential indicators of the maturity of differentiating cardiomyocytes from iPS cells in vitro.

Materials and Methods
Animal care procedures were consistent with the ''Guide for the Care and Use of Laboratory Animals'' (National Institutes of Health publication). Experimental protocols were approved by the Ethics Review Committee for Animal Experimentation of Osaka University Graduate School of Medicine.

Cardiomyogenic differentiation of murine iPSCs in vitro
We used the murine iPSC lines, 959A2-1, 959C1-1, 956F-1 (generous gifts from Dr. Okita and Professor Yamanaka of the Center for iPS Cell Research and Application, Kyoto University, Kyoto, Japan). The cell lines were generated from C57BL/6 (B6) (CLEA) mouse embryonic fibroblasts by introducing Oct3/4, Sox2, Klf4, and c-Myc without viral vectors as described [19]. The iPSCs were cultured in the absence of serum and feeder cells by using ESGRO Complete PLUS Clonal Grade Medium (Millipore).
Adult cardiac tissue from B6 mice (CLEA) was used as a control. Male B6 mice (8 weeks old) were sacrificed by intravenous administration of potassium chloride under inhalation anesthesia of isoflurane, and heart tissue from the left ventricle was harvested for further studies and labeled ''Heart''.

Immunohistochemistry analysis
IPSC-CMs were dissociated with 0.25% trypsin-EDTA and then fixed with 4% paraformaldehyde. The cells were stained with the following primary antibodies: mouse anti-a-actinin antibody (Sigma-Aldrich) and rabbit anti-troponin I antibody (Abcam), and then visualized by the following secondary antibodies: Alexa Fluor 488 donkey anti-mouse IgG (Invitrogen) and Alexa Fluor 594 goat anti-rabbit IgG (Invitrogen). The nucleus of the cells were stained with 49, 6-Diamidino-2-phenylindole dihydrochloride (DAPI) and then observed with a confocal laser scanning microscopy FV1200 (Olympus).

Ca 2+ transient measurement and pharmacological analysis
5 mM Fluo-8 regents (AAT Bioquest, Inc.) in serum-free MEM was added to iPSC-CMs after the cells were washed with phosphate buffered saline. The cells were incubated at 37uC for 30 min and then observed with a fluorescence microscopy. Fluorescence intensity of Fluo-8 dye was sequentially measured using iQ2 software (ANDOR) pre and post the administration of 1 mM isoproterenol.

Flow cytometry
IPSC-CMs were dissociated with 0.25% trypsin-EDTA and then fixed with CytoFix fixation buffer (BD) for 20 min. The cells were permeabilized with Perm/Wash buffer (BD) at room temperature for 10 min and then incubated with mouse antitroponin T antibody (Thermo) for 30 min. Cells were washed with Perm/Wash buffer prior to incubation with the Alexa Fluor 488 rabbit anti-mouse IgG secondary antibody (Invitrogen) at room temperature for 30 min. These cells were analyzed on a FACS Canto II (BD).

Characterization of N-glycans derived from iPSCs, iPSC-CM, and Heart
All experimental procedures, including chromatography conditions and glycosidase treatments, have been described previously [23]. Cultured undifferentiated iPSCs, iPSC-CMs, and the heart tissue were treated with chloroform-methanol, then subjected to proteolysis with chymotrypsin and trypsin, followed by glycoamidase A digestion to release N-glycans. After removal of peptides, the reducing ends of the N-glycans were derivatized with 2aminopyridine (Wako). This mixture was applied to a diethylaminoethyl (DEAE) column (Tosoh) or a TSK-gel Amide-80 column (Tosoh); each fraction from the amide column was then applied to a Shim-pack HRC-octadecyl silane (ODS) column (Shimadzu). The elution times of individual peaks from the amide-silica and ODS columns were normalized to a pyridylamino (PA)-derivatized isomalto-oligosaccharide with a known degree of polymerization, and are represented as glucose units (GU). Thus, each compound from these two columns provided a unique set of GU values, which corresponded to the coordinates of the 2D HPLC map. The PA-oligosaccharides were identified by comparison to the coordinates of ,500 reference PA-oligosaccharides in a homemade web application, GALAXY (http://www.glycoanalysis.info/galaxy2/ ENG/index.jsp) [24]. The calculated HPLC map based on the unit contribution values was used to estimate some high-mannose type PA-oligosaccharides. The PA-oligosaccharides were cochromatographed with the reference to PA-oligosaccharides on the columns to confirm their identities. PA-glycans that did not correspond to any of the N-glycans registered in GALAXY were trimmed by exoglycosidase to produce a series of known glycans [25].

Mass spectrometry
PA-oligosaccharides were analyzed by matrix-assisted laser desorption/ionization time-of-flight mass spectrometric (MALDI-TOF/MS). The matrix solution was prepared as follows: 10 mg of 2,5-Dihydroxybenzoic acid (Sigma) was dissolved in 1:1 (v/v) acetonitrile/water (1 mL). Stock solutions of PA-glycans were prepared by dissolving them in pure water. One microliter of a sample solution was mixed on the target spot of a plate with 1 mL matrix solution and then allowed to air-dry. MALDI-TOF/MS data were acquired in the positive mode on an AXIMA-CFR (Shimadzu) operated in linear mode.

Materials
Glycoamidase A from sweet almond, a-mannosidase, bgalactosidase, and b-N-acetylhexosaminidase from jack bean were purchased from Seikagaku Kogyo (Tokyo, Japan). a-Galactosidase from coffee bean was purchased from Oxford GlycoSciences (Oxford, UK). Trypsin and chymotrypsin were obtained from Sigma (St. Louis, MO). Pronase protease from Streptomyces griseus was from Calbiochem (San Diego, CA). The pyridylamino (PA) derivatives of isomalto-oligosaccharides 4-20 (indicating the degree of polymerization of glucose residues) and reference PAoligosaccharides were purchased from Seikagaku Kogyo.

Semi-quantitative PCR
DNA-free total RNA was extracted with the RNeasy RNA isolation Kit (Qiagen) and reverse-transcribed into cDNA using Omniscript reverse transcriptase (Qiagen), then analyzed by quantitative real-time PCR on an ABI PRISM 7700 thermocycler (Applied Biosystems) with the following TaqMan gene expression assays (Applied Biosystems): ST3Gal-III (Gal b1-3(4) GlcNAc a-2,

Highly purified cardiomyocytes derived from iPSCs
Cardiomyogenic differentiation was induced in murine iPSCs by using a slightly modified culture protocol (Figure 1a). The iPSC-CMs showed significantly higher expressions of Nkx2.5, aMHC, ANP and Isl1 than undifferentiated iPSCs by semiquantitative real-time PCR (Figure 2a), and showed sarcomere structures observed by immunohistological staining of a-actinin and troponin I (Figure 1b). The iPSC-CMs were functional with Ca 2+ transient measurement (Figure 3a, b) and their beating rates were increased by the administration of isoproterenol (Figure 3c), meaning they had b-adrenergic receptors. Nearly all of the iPSC-CMs exhibited spontaneous and regular beating at room temperature (Video S1). The differentiation efficiency of murine iPSC was evaluated by flow cytometry analysis. More than 95% of the 959A2-1 CMs, 92% of the 959C1-1 CMs and 90% of the 956F-1 CMs were positive for troponin T (Figure 2b), while the undifferentiated iPSCs rarely expressed troponin T (Figure 2b).

Structures of N-Glycans isolated from iPSCs, iPSC-CM, and Heart
The isolated N-glycans were analyzed by means of a mapping technique based on their HPLC elution positions and MALDI-TOF/ were not identified in this study because they did not correspond to GALAXY references even after a-galactosidase digestions. They are described in Figure 8 and Table S1-S5 with their proposed formulas based on MALDI-TOF/MS data.
Expression of glycosyl transferase, ST3Gal-III, ST3Gal-IV, ST6Gal-I, and CMAH in the iPSCs, iPSC-CMs, and Heart was assessed by RT-PCR to explore the glycan structures responsible for the differences between groups. The Heart expressed high levels of CMAH (0.9160.13/GAPDH); levels in the iPSCs and iPSC-CMs were markedly lower (iPSCs:  (Figure 10b).

Discussion
Sixty-eight different N-glycans were isolated from iPSCs, iPSC-CMs, and the Heart. The structures of 60 N-glycans were identified, based on their HPLC elution peaks (Figure 8, Table S1-S5). Each preparation contained a combination of neutral, monosialyl, and disialyl N-glycans.
The molar ratios of high-mannose, monofucosylated, and difucosylated N-glycans were substantially different between groups (Figure 9), although no clear differences in the abundance of these glycans were found. The decrease in high-mannose Nglycans and increase of fucosylated N-glycans in iPSC-CMs versus iPSCs is consistent with a previous report on a comparison of ESC derived cardiomyocytes to undifferentiated ESCs [18]. Generally, all N-glycans are synthesized from the high-mannose type by a large array of sequentially and competitively acting biosynthetic enzymes located throughout the endoplasmic reticulum and Golgi apparatus [26], indicating that the high-mannose type of Nglycans could be categorized as a marker of immaturity. In this study, the high-mannose N-glycans were highest in the immature iPSC and lowest in the Heart, or mature tissue; thus, the quantity of high-mannose-type N-glycans might be an indicator of maturity in iPSC-derivatives and the iPSC-CMs in our protocol may still be immature in comparison to cardiac tissue.
The proposed spectra-based composition of the D8 glycans in iPSCs was [(Hexose)5(HexNAc)5(NeuGc)2(PA)1], indicating that it contains NeuGc. However, D8 might be quite a rare exception because transcript levels of CMAH, which catalyzes the conversion of NeuAc to NeuGc, was quite low in iPSCs in comparison to the Heart. This data suggests that during the process of reprograming, iPSCs suppress or eliminate CMAH activity. We conclude that iPSCs contain less sialic acid (especially NeuGc) and high-mannose structures are abundant in the N-glycans. In contrast, heart cells produce numerous sialyl-N-glycans, especially NeuGc. Transcript levels of CMAH tended to increase in iPSC-CMs relative to iPSCs, suggesting cardiomyogenic differentiation may induce expression of CMAH. If the iPSC-CMs could be matured more closely to the Heart by some additional methods of culture, the quantity of high mannose type of N-glycans might decrease more closely to the Heart, and might produce N-glycans containing NeuGc, followed by the expression of CMAH.
A terminal NeuGc, the Hanganutziu-Deicher (H-D) epitope [27], is widely distributed in the animal kingdom with the exception of humans and chickens. Expression of NeuGc is controlled by CMAH activity. Irie et al. [28] and Chou et al. [29] cloned the cDNA for human CMAH and reported that the Nterminal truncation of human CMAH is caused by deletion of Exon 6, a 92-base pair segment in the genomic DNA. Expression of this truncation in the heart eliminates NeuGc in sialyl Figure 8. Structures of neutral, monosialyl, and disialyl PA-oligosaccharides in iPSCs, iPSC-CM, and heart cells. Glucose units (GU) were calculated from the peak elution times for the ODS column in Figure 5, 6  structures. If human iPSCs or iPSC-CMs do not express CMAH in the same way as murine iPSCs or iPSC-CMs, there may be no difference between human iPSCs, iPSC-CMs, and the human Heart. Further study on human iPSC-CM will be needed to completely understand the features of the sialyl acid of N-glycans.
It was reported that human iPSCs produced a2,6sialyl glycans but did not contain a2,3sialyl structures, in contrast to human fibroblast, the origin of iPSCs, which produced a2,3sialyl but not a2,6sialyl structures [30,31]. The murine iPSCs in this study contained a2,3sialyl structures in NeuAc, M5, M23, D4-1, D10-1 and D12, and the iPSC-CMs produced a2,3 and a2,6sialyl structures in NeuAc. These differences may be due to variations between species, because mouse Heart cells also contained a2,3 and a2,6sialyl structures in NeuGc. Further studies are needed to characterize the glycome shift in the production and differentiation of iPSCs.
Type I Lactose structures were not detected, although over 98% of glycans in each cell were accounted for in this study. The Nglycans of N9-3, M8, M12, M17, and M23, which were identified after a-galactosidase digestion, contained Gala1-6Gal, not only in the neutral glycans but also in the monosialyl N-glycans of the iPSC-CM preparation. The same structure was not found in iPSCs, but only one structure, M23, was present in Heart cells. Therefore, in iPSC-CMs, Gala1-6Gal enzyme activity appears to be up-regulated in comparison to wild-type myocardium, although enzyme activity was not assessed by RT-PCR because of the limited availability of genetic sequence data.
The D8 was identified in all of three iPSC lines and not in the iPSC-CMs and Heart. This structure, unfortunately not identified in this study, may be useful as markers of undifferentiated iPSCs in the same way as well-known pluripotency biomarkers such as stage-specific embryonic antigens (SSEA)-3, SSEA-4 (glycosphingolipids) [32].
Previous MALDI-TOF/MS and MS/MS studies concluded that many kinds of N-glycans are found in organs and cells. The number of detected N-glycans is attributed to the sensitivity of the MS and HPLC methods employed. That is, MS data are sensitive and can be rapidly obtained, but a glycan structure is identified based only on the calculated molecular weight. Therefore, discriminating between isomeric structures is difficult. On the other hand, it thus appears that the accuracy of the data presented here using HPLC mapping in conjunction with a MALDI-TOF technique provides much more detailed information. Our data were used to identify the representative features of each N-glycan in these three cell types.
There may be a concern that the heart tissue used in this study contains connective tissues, vessels or nerves other than cardiomyocytes. Therefore, some of the N-glycans detected from the Heart sample might be derived from the tissues other than cardiomyocytes. However, heart is majority composed by cardiomyocytes, and furthermore, even if a small amount of N-glycans derived from connective tissues were contaminated in the Heart sample, the main evidences in this study, such as the proportion of the high-mannose type N-glycans, the ratio of the active sialyltransferase genes, the existence of NeuGc, and the uncommonness of Gala1-6 Gal, are essentially not affected.
In summary, murine iPSCs were rich in high-mannose type Nglycans but very poor in sialyl type N-glycans. Murine heart tissue contained a relatively low volume of high-mannose glycans, but was very rich in neuraminic acid, especially NeuGc type sialyl structures. Under these conditions, the volume of each type of glycan was similar for iPSC-CMs and iPSCs. That is, they were rich in high-mannose and relatively poor in sialyl type N-glycans by volume. In addition, most of the sialyl structures of the iPSC-CMs were different from those of the Heart, and the iPSC-CMs expressed no NeuGc. Moreover, the iPSC-CMs produced several unique glycans with the Gala1-6Gal structure. These results provide important data that can be useful in future clinical iPSC studies.
It is quite important to investigate the meaning of N-glycans transitions during the cardiomyogenic differentiation presented in this study, for deeply understanding the relationship between the N-glycan expression and cardiomyogenic differentiation. Knockout or knock-down of the genes related to cardiomyogenic differentiation or glycosylation may be useful for such purpose. However, the N-glycan signature in the cell surface is determined by a variety of the genes. Knock-out or knock-down of a single gene related to cardiomyogenic differentiation would alter an array of gene expressions, such as sarcomere proteins, transcriptional factors, or cell surface proteins, all of which would affect the signature of N-glycans in the cell surface. Therefore, the data interpretation for relationship between expression of a single gene and N-glycan signature would be difficult. Some different experimental approach may be needed to investigate the meaning of change in N-glycan expression during cardiomyogenic differentiation.

Supporting Information
Table S1 Structures and relative quantities of neutral (Table S1, S2) PA-oligosaccharides derived from iPSC, iPSC-CM, and heart cells. a. Glucose units (GU) were calculated from the peak elution times of the peaks obtained from the ODS column in Figure 5   Video S1 (MP4)