Single-cell fucosylation breakdown: Switching fucose to europium.

Fucosylation and its fucosidic linkage-specific motifs are believed to be essential to understand their distinct roles in cellular behavior, but their quantitative information has not yet been fully disclosed due to the requirements of ultra-sensitivity and selectivity. Herein, we report an approach that converts fucose (Fuc) to stable europium (Eu) isotopic mass signal on hard ionization inductively coupled plasma mass spectrometry (ICP-MS). Metabolically assembled azido-fucose on the cell surface allows us to tag them with an alkyne-customized Eu-crafted bacteriophage MS2 capsid nanoparticle for Eu signal multiplication, resulting in an ever lowest detection limit of 4.2 zmol Fuc. Quantitative breakdown of the linkage-specific fucosylation motifs in situ preserved on single cancerous HepG2 and paracancerous HL7702 cells can thus be realized on a single-cell ICP-MS platform, specifying their roles during the cancering process. This approach was further applied to the discrimination of normal hepatocellular cells and highly, moderately, and poorly differentiated hepatoma cells collected from real hepatocellular carcinoma tissues.

Introduction

Cellular behavior is significantly governed by the biomacromolecules on the cell surface that carry meaningful information and serve biological functions, making the cells to be heterogeneous (Altschuler and Wu, 2010). During which, the linkage-specific multielement-composed motifs within the biomacromolecules direct their roles and eventually influence the cell behavior. Glycosylation is such a typical paradigm with diverse glycosidic linkages between monosaccharides attached covalently to their aglycone such as a protein and/or a lipid displaying various biological consequences (Ohtsubo and Marth, 2006; Pinho and Reis, 2015). For instance, fucosyltransferase (FUT)-mediated fucosylation on the cell surface has profound biological implications that depend heavily on its fucosidic linkages, such as α1,2-, α1,3/4-, and α1,6-linked fucosylation at the terminal, sub-terminal, and innermost core position of N-glycan chains, as well as the other types of fucosylation including O-fucosylation where fucose (Fuc) is added directly to Ser and Thr residue of, for example, epidermal growth factor-like repeats and thrombospondin type 1 repeats (Li et al., 2018; Ma et al., 2006; Schneider et al., 2017). These different fucosidic linkages donate diverse functions of fucosylation, saying that the α1,2-linkage Fuc is mainly responsible for A, B, and H Lewis blood antigens in blood type decision (Marionneau et al., 2001), while the α1,3/4 are believed to participate in the interactions between selectins and the α1,3/4-bearing sialyl-Lewis structures during many biological processes such as immune modulation and tumor metastasis (Barthel et al., 2009); as for the core α1,6, it has been deemed as a biomarker of cancer cells because its abnormal assembly always occurred during tumorigenesis (Aoyagi et al., 1991); as well as O-linked fucosylation promotes the metastasis of cancer by influencing protein folding, secretion, and in the regulation of Notch signaling (Taylor et al., 2014). Quantification of Fuc and more about its diversely fucosidic linkages have been placed at one of the central positions to unveil the causality between fucosylation and diseases (Li et al., 2018). Such diverse fucosidic linkage assemblies are catalyzed by their corresponding FUTs, therefore supposed in general from the perspective of biologists to be reflected by their corresponding FUTs' activity and abundance that could be detectably speculated by optical-based western blots electrophoretic methods and microarrays, but not the exact fucosidic linkage-specific motifs themselves (Chen et al., 2013; Mathieu et al., 2004; Yin et al., 2010). On the other hand, with the aid of the globally chemoselective and/or targeted biospecific enrichment procedures as well as related bioinformatics methods, information regarding monosaccharide composition and site- and glycan-specific glycosylation could be obtained using soft ionization tandem mass spectrometry (MS/MS), typically electrospray ionization Electrospray ionization-MS/MS and matrix-assisted laser desorption/ionization matrix-assisted laser desorption/ionization-MS/MS, together with the sophisticated fragmentation strategies such as collision-induced dissociation (CID), electron capture dissociation, and electron transfer dissociation (Palaniappan and Bertozzi, 2016; Ruhaak et al., 2018). Definitely, collective knowledge contributed by the scientific forerunners built the contemporary comprehension about glycomics and glycoproteomics (Alley et al., 2013; Pinho and Reis, 2015). The state-of-the-art methodology developed, however, generally followed the apparently changeless steps toward the difficulties of (1) intrinsic poor and diverse ionization efficiency of the labile plus low abundance glycosylation and (2) awful heterogeneity raised from the branch-structured tree-shaped sugar chains with different glycosidic linkages of structural isomers. In-depth accurate information on the sequence and glycosidic linkages of a glycan is yet waiting to be fully disclosed, lying greatly on further improvements of MS tools and novel ion fragmentation strategies, in addition to the creation of richer structural library-based bioinformatics. Not just fucosylation exemplified here, it is a universally basic prerequisite to develop a methodology capable to distinguish and quantify not only total glycosylation but also the motifs assembled at different sites via different glycosidic linkages for exploring the structure-dependent roles of the glycosylated biomacromolecules on the cell surface. This still remains a challenge because an unfortunate problem is the unforeseeable cross-reactivity of the lectins used, besides the generally weak binding characteristics of the lectins toward specific glycosidic linkages for their recognition and enrichment (Cecioni et al., 2015; Ghazarian et al., 2011; Lis and Sharon, 1998). For example, the seemingly α1,6-selective Aleuria aurantia lectin displayed cross-reactivity toward multiple glycan epitopes including α1,2- and α1,3-linked Fuc (Noda et al., 2003). A more serious concern, on the other hand, is the facile Fuc ion migration and rearrangement behavior during the analysis of soft ionization MS/MS that unavoidably caused ambiguous even wrong results, despite chemical derivatization with easily ionized groups could overcome the issue of relative low ionization efficiency of Fuc (Harvey et al., 2002; Nwosu et al., 2015). Chemical modifications (Lattova et al., 2019) and fine controlling of the CID energy (Acs et al., 2018; Yuan et al., 2019b), for example, could minimize the consecutive processes of the facile Fuc ion migration and neutral Fuc residue loss during the soft ionization MS/MS enabling the discrimination of core and antenna fucosylation. But the sensitivity was reluctantly compromised, and still there were several (the core fucosylation) to a dozen (the antenna fucosylation) percentage of interferences, merely offering their relative abundance ratios if reference standard compounds were not available. Chemoenzymatic labeling strategies provided a selective approach but just limited to α1,2 Fuc linkage (Chaubard et al., 2012) being not able to provide a full spectrum of fucosylation on the cell surface.

Either in-source ionization and/or post-source fragmentation during the soft ionization MS/MS is a good manner that we initially expected making structure-dependent ionization and identification possible. However, in the same time, the soft ionization source is a gas-phase reactor that produces many unexpected fragment-ion rearrangement outcomes, causing a difficult situation that we do not want to see in the case of fucosylation analysis. Furthermore, the in situ scene of fucosylation and its linkage-specific motifs on the cell surface has been ever more desired to date, considering the fact that the glycosylated biomacromolecules integrated on a single cell is the most front line governing the heterogeneity of cellular behavior of cell-cell communication and immune modulation during normal and abnormal life processes, namely, the important cell profiling and cell-to-cell variability as we have already known (Pelkmans, 2012; Teichmann, 2019). Nonetheless, these not just qualitative but also quantitative information about the glycosidic linkage-specific motifs involved on a single cell are hard to be obtained not just because of the requirement of an unimaginable sensitivity but also because of the loss of the much desired in situ information on the cell surface when employing the current mainstream soft ionization MS/MS tools that generally analyze the cell lysates.

By contrast, Ar-based inductively coupled plasma (ICP) is a hard ionization source that converts most elements into their isotopic ions, especially the metallic elements that compose or are labeled to the biomacromolecules on an intact cell. When coupling to MS, inductively coupled plasma mass spectrometry (ICP-MS) is a universal tool not only for the quantification of the elements regardless of their chemical forms in which they exist (Montaser, 1998) but also for the biomacromolecules and cells via determination of the elements tagged (Bandura et al., 2009; Bendall et al., 2011; Li et al., 2005; Liang et al., 2015; Liu et al., 2016; Sanz-Medel et al., 2012; Virani and Tanner, 2015; Wei et al., 2020; Yan et al., 2013). If a monosaccharide, such as Fuc exemplified here, can be specifically tagged by a metallic element complex according to a certain stoichiometry, quantification of the Fuc would be switched to a genuine and credible isotopic mass signal of the element. The inevitable problem of the fragment ion rearrangements encountered during the soft ionization MS/MS would be fundamentally avoided. Moreover, ICP-MS is capable of directly analyzing intact cells and thus is able to directly readout the element-tagged Fuc in situ preserved on the cell surface upon a single-cell ICP-MS platform. In this study, we explored this capability for the quantitative breakdown of fucosylation and its linkage-specific motifs on a single cell (Scheme 1). Fortunately, post-modifiable Fuc analog such as per-acetylated azide-fucose (Ac4FucAz) has been previously shown to be able to metabolically incorporate into the glycan chains and accumulate on the cell surface catalyzed by the cell's own in vivo FUTs via cell's essential salvage pathway (Rabuka et al., 2006; Sawa et al., 2006). This globally installed FucAz on the cell surface provides an opportunity to be bioorthogonally tagged by either an alkyne (ALK)-DOTA-europium (Eu) complex or ALK-modified and Eu-decorated bacteriophage MS2 capsid nanoparticle (ALK-PEG-MS2-DOTA-Eu) for Eu mass signal multiplication to meet the ultra-sensitivity requirement of a single-cell analysis. In addition to the total FucAz that enables better understanding of the difference between the different types of cells, selective quantification of trace amounts of the fucosidic linkage-specific motifs on a single cell can be achieved to elucidate heterogeneity among the same type of cells and thus the relationship between the linkage-specific motifs and their distinct biological functions, benefitted from linkage-specific fucosidase that can selectively detach its corresponding fucosidic linkage-specific motif from the cell surface glycan chain, while the wanted linkage-specific motif remains to be quantified. Human hepatocellular carcinoma HepG2 and paracancerous HL7702 cell lines were selected as the models of cancerous and normal cells to validate this proposed approach. Moreover, this approach was applied to the discrimination of highly, moderately, and poorly differentiated hepatoma cells from normal hepatocellular cells sampled from real hepatocellular carcinoma (HCC) and healthy liver tissues.

Scheme 1

Quantitative breakdown of single-cell fucosylation and its fucosidic linkage-specific motifs via a Eu-encoding single-cell ICP-MS approach mediated by globally metabolic assembly of FucAz via cell's essential salvage pathway and fucosidase-specific release strategy

For a Figure360 author presentation of this figure, see https://doi.org/10.1016/j.isci.2021.102397.

The per-acetylated azide-fucose (Ac4FucAz) was synthesized and used for incorporating FucAz into cell surface glycan chain via cell's essential fucose salvage pathway. Afterward, the globally incorporated FucAz on the single-cell surface can be recognized and conjugated bioorthogonally by an ALK-modified and Eu-decorated MS2 capsid nanoparticle element tag (ALK-PEG-MS2-DOTA-Eu) via click chemistry. In addition to the total fucosylation on a single cell, its linkage-specific motifs including terminal α1,2-, sub-terminal α1,3/4-, and innermost α1,6-linked fucosylation, as well as other types of fucosylation, can be accurately quantified via the determination of the tagged Eu using 153Eu-species-unspecific isotope dilution ICP-MS on a single-cell ICP-MS platform with aid of the corresponding linkage-specific fucosidases.

Results and discussion

Synthesis and characterization of Ac4FucAz, ALK-DOTA-Eu, MAL-DOTA-Eu, and ALK-PEG-MS2-DOTA-Eu

We first synthesized the unnatural Ac4FucAz according to the previous reports (Burkart et al., 2000; Laughlin and Bertozzi, 2007) with minor modifications (see transparent methods and Scheme S1). The intermediates and final product were characterized using ESI-qTOF-MS ([M + Na]+ at m/z 396.0915, [2M + Na]+ 769.1940) and 1H-/13C-NMR (Figures S1–S10), confirming the synthesis of Ac4FucAz with a final yield of 47.5%. We next evaluated how the treatment with exogenous Ac4FucAz affects the cell growth status. As shown in Figure S11, the CCK-8 cytotoxicity assay indicated that the IC50 of Ac4FucAz is 252.1 μM for HepG2 and 267.9 μM HL7702. Cell viability gradually decreases along with the increase in the concentration of Ac4FucAz in general with a plateau fluctuating around (79.3 ± 0.7) % of HL7702 and (82.6 ± 2.3) % HepG2 (n = 7) from 75 to 200 μM (Figure S11), which is in agreement with the phenomenon observed in other previous reports (Sawa et al., 2006; Hsu et al., 2007). In order to increase the cell uptake, thus for more efficacious incorporation of FucAz into the glycan chain on the cell surface, as well as to maintain stable cell growth status and obtain reproducible results, 200 μM Ac4FucAz was used in the following cell culture experiments. Then, a Eu-loaded 1,4,7,10-tetraazacyclododecane-1,4,7-tris(acetic acid)-10-(3-butynylacetamide) (ALK-DOTA-Eu) complex (Figure S12, ESI-qTOF-MS [M + H]+ at m/z 604.1601 and 606.1616) was synthesized and used to tag the FucAz metabolically incorporated on the cell surface for a bulk cell mode ICP-MS analysis. Meanwhile, a stoichiometrically Eu-decorated bacteriophage MS2 capsid nanoparticle was fabricated for Eu mass signal multiplication (Yuan et al., 2019a) to meet the ultra-sensitivity requirement for single-cell breakdown of fucosylation and its linkage-specific motifs on a direct infusion of lined up single-cell ICP-MS platform established in our lab (see Instrumentation of transparent methods) (Zhou et al., 2020). Different from our previous preparation route via the acylation between NHS-PEGn-N and the amino group (-NH2) on the MS2 capsid nanoparticle and then clickable conjugation with a Eu-loaded 1,4,7,10-tetraazacyclododecane-1,4,7-tris(acetic acid)-10-dibenzocyclooctyne (DBCO-DOTA-Eu) (Yuan et al., 2019a), the active -NH2 groups were firstly converted to sulfhydryl (-SH) groups using 2-iminothiolane hydrochloride (Traut's Reagent) for not only evading the possible hydrolysis of NHS under the experimental conditions but also adding an 8.1 Å spacer arm making -SH more flexible to be effortlessly modified (Figures 1A and 1B). The results obtained from ESI-qTOF-MS characterization before (MS2-NH2 monomer with six modifiable -NH2 groups at m/z 13,734.1438) and after modification (MS2-SH at m/z 14,339.9441) designated that the 1080 -NH2 groups on the MS2 capsid nanoparticle were completely converted into 1080 -SH groups (Figure 1C). The observed transmission electron microscopy (TEM) size and morphology of the obtained MS2-SH did not noticeably change after the modification compared to those of MS2-NH2 (27-nm spherical shell in diameter) (Figure 1D). Subsequently, the pre-synthetic Eu-loaded 1,4,7,10-tetraazacyclododecane-1,4,7-trisacetic acid-10-maleimidoethylacetamide (MAL-DOTA-Eu) (Luo et al., 2013; Yan et al., 2010, 2011) (Figure S13, ESI-qTOF-MS [M + H]+ m/z 675.1367 and 677.1385) was employed to chemically craft MS2-SH via the Michael addition reaction between MAL and -SH. The obtained results using 153Eu-species-unspecific isotope dilution ICP-MS (153Eu-SUID-ICP-MS) coupled with size-exclusion chromatography (SEC) (Figure S14A) indicated that all the -SHs can be quantitatively modified, suggesting that 1080 Eu atoms were chemically decorated on one MS2 capsid nanoparticle (MS2-DOTA-Eu). Moreover, in order to impart the tagging ability of MS2-DOTA-Eu toward the FucAz on the cell surface, a 45-ethylene glycol unit containing PEG-linked (PEG2000, 157.5 Å in length) MAL-PEG2000-ALK (171.6 Å) was used to chemically modify MS2-SH for (1) merely reacting with the outer surface of -SHs because the capsid pores of 18 Å prevent the MAL-PEG2000-ALK from entering into the interior of MS2 capsid and (2) protruding MAL-PEG2000-ALK outside the surface of the MS2-DOTA-Eu far beyond MAL-DOTA-Eu (13.5 Å). Such a design could avoid the electrostatic repulsion and steric hindrance effects, thus benefiting the bioorthogonally clickable tagging of the FucAz on the cell surface. The number of MAL-DOTA-Eu and MAL-PEG2000-ALK that have the same reactive group MAL toward -SH could be tuned via changing their molar ratio from 1/2 to 0/1, resulting in (471 ± 16) to (1080 ± 27) of MAL-DOTA-Eu and (609 ± 16) to (0 ± 27) of MAL-PEG2000-ALK (n = 7) on one MS2 determined using SEC-153Eu-SUID-ICP-MS (Figures S14B–S14H). Compared with PEG600 (14 ethylene glycol units, 49.0 Å), PEG1000 (23, 80.5 Å), PEG3400 (77, 269.5 Å), and PEG5000 (114, 399 Å) tested, ALK101-PEG2000-MS2-DOTA-Eu979, which was obtained at the molar ratio of 1/50 of ALK-PEG2000-MAL to MAL-DOTA-Eu, not only guaranteed the tagging ability toward the FucAz but also demonstrated the highest signal multiplication up to (979 ± 15) times (n = 7). In this way, nearly three orders of magnitude higher 153Eu isotopic mass signal intensity was improved than merely using ALK-DOTA-Eu (Figures S15 and S16). The limit of detection (LOD) of FucAz is down to 16.1 amol (corresponding to 9.7 × 106 FucAz) compared with 15.6 fmol (9.4 × 109 FucAz) using ALK-DOTA-Eu alone (Figure 1E), which was calculated by 3 × SD dividing the slope of the corresponding calibration curves, given the fact that the clickable 1:1 conjugation between -ALK and -N3 (Figure S17) via the bis[(tertbutyltriazoyl)methyl]-[(2-carboxymethyltriazoyl)methyl]-amine (BTTAA) mediated Cu(I)-catalyzed azide-alkyne cycloaddition during which BTTAA and aminoguanidine were used for stabilizing and detoxicating the cytotoxicity of Cu(I) (Besanceney-Webler et al., 2011; Uttamapinant et al., 2012).

Figure 1

Preparation, characterization, and application of Eu-decorated bacteriophage MS2 capsid signal multiplication nanoparticle tag

(A) Bacteriophage MS2 is composed of 180 identical protein monomers, and each monomer has six modifiable amino sites, thus totally 180 × 6 = 1080 amino sites can be chemically modified;

(B) The preparation routes of ALK-PEG-MS2-DOTA-Eu via two key steps including the conversion of -NH2 to -SH using Traut's Reagent and -SH conjugation with MAL-DOTA-Eu and MAL-PEG-ALK for Eu mass signal amplification and targeting the FucAz metabolically assembled in the glycan chains on the cell surface;

(C) ESI-qTOF-MS of MS2-NH2 capsid monomer at m/z 13,734.1438 and MS2-SH 14339.9441;

(D) TEM images of MS2-NH2, MS2-SH, and ALK101-PEG2000-MS2-DOTA-Eu979 nanoparticles, scale bar, 100 nm;

(E) Calibration curves of 153Eu-ICP-MS intensity (153Euspike × 99.8% + 153EuSample × 52.19%) against the FucAz amount (left y and bottom x) and cell number of HL7702 and HepG2 (right y and top x) tagged with ALK-DOTA-Eu and ALK101-PEG2000-MS2-DOTA-Eu979 (n = 7, data are represented as mean ± SD).

Metabolic incorporation and quantification of the total FucAz on single cell surface

Human HCC and paracancerous cell lines HepG2 and HL7702 were selected as the models of cells. They were cultivated in the presence of 200 μM Ac4FucAz. Afterward, the cells were tagged using ALK-DOTA-Eu. We found that only the cells cultured with Ac4FucAz were determined by ICP-MS when monitoring 151/153Eu, while those without Ac4FucAz cultivation and/or ALK-DOTA-Eu tagging detected with almost negligible background signal (Figure S18). These results suggested the success of the metabolic incorporation of FucAz on the cell surface and the specificity of the clickable ALK-DOTA-Eu tagging. In parallel, ALK-Cy5 with the same reactive ALK group as ALK-DOTA-Eu was used to label the cells. The confocal laser scanning microscope observation confirmed the same results (Figure S19) but could hardly offer the quantitative content of FucAz on the cell surface. The total FucAz can be quantified using 153Eu-SUID-ICP-MS under a bulk cell mode using a dwell time of 100 ms with a cell population of 1.0 × 106 HepG2 and/or HL7702. The average contents of the FucAz per HepG2 and HL7702 that were calculated by the total FucAz determined dividing the cell number used are (8.3 ± 0.3) × 10−17 mol (corresponding to 5.0 × 107 FucAz) and (8.4 ± 0.4) × 10−18 mol (5.1 × 106 FucAz) (n = 7) according to the calibration curves (Figure 1E), indicating that more than one order of magnitude higher FucAz incorporated on HepG2 than HL7702. These results also implied that at least 187 HepG2 and 1848 HL7702 cells are needed under the bulk cell mode ICP-MS analysis using ALK-DOTA-Eu tag for direct readout of the FucAz considering the LOD of 15.6 fmol FucAz and the average FucAz contents on the cells, being far away from the ultra-sensitivity requirement for an exact single-cell analysis. When ALK101-PEG2000-MS2-DOTA-Eu979 was used to tag the cells, by contrast, one HepG2 and two HL7702 cells could be determined owing to the signal-multiplicated LOD of 16.1 amol. It should be noted that when the cells were analyzed on the single-cell ICP-MS platform with the 10-fold shorter dwell time of 10 ms than that used in the bulk cell mode, the signal to noise ratio could be further improved, resulting in the LOD down to 4.1 amol Eu (Figure S20). Taking the advantage of signal multiplication effect of ALK101-PEG2000-MS2-DOTA-Eu979, 4.2 × 10−21 mol (zmol) FucAz (corresponding to 2.5 × 103 FucAz) can be detected, guaranteeing the direct readout of a real single-cell event regarding not only the total FucAz but also its linkage-specific motifs on the single-cell ICP-MS platform (Figures 2A, S21, and S22), which could line up single cells with a controllable adjacent interval time and directly infuse them one by one into ICP-MS with the almost quantitative transport efficiency and detection efficiency of 86%. The average FucAz of (8.3 ± 0.6) × 10−18 mol (corresponding to 5.0 × 106 FucAz) per HL7702 cell and (8.4 ± 0.5) × 10−17 mol (5.1 × 107 FucAz) HepG2 determined from the 3000 independent single-cell events of each run (n = 21) are comparable to those obtained under the bulk cell mode analysis (Figure 2B). The signal multiplication ALK101-PEG2000-MS2-DOTA-Eu979 nanoparticle tag allowed us to determine the single-cell events and distinguish either HepG2 from HL7702 cells or cell-to-cell heterogeneity among the same HepG2 and/or HL7702 cells regarding fucosylation (Figures 2B and 2C). The statistical results obtained (Figure 2C) indicated that the FucAz content per HL7702 cell ranges from (4.8 ± 0.7) × 10−18 (10% limit) to (1.3 ± 0.4) × 10−17 mol/cell (90% limit), being difference of 2.7 times with the average and median values of (8.3 ± 0.6) × 10−18 mol/cell and (7.8 ± 0.7) × 10−18 mol/cell; while on HepG2, the FucAz content ranges from (2.8 ± 0.6) × 10−17 (10%) to (1.5 ± 0.3) × 10−16 mol/cell (90%) of 5.4 times difference with the average and median values of (8.4 ± 0.5) × 10−17 mol/cell and (6.6 ± 0.4) × 10−17 mol/cell. Not only more than 10-fold higher FucAz content on HepG2 than HL7702 according to the average value of the total FucAz (p ˂ 0.001, n = 21) (Figure 2B), but also 2-fold larger heterogeneity of HepG2 than HL7702 in general (Figure 2C). These results also revealed that not all HL7702 cells incorporated lower levels of Fuc. Around 2% of the paracancerous HL7702 cells did express abnormally high levels of FucAz (>1.8 × 10−17 mol/cell) that overlapped with some of HepG2, implying that these paracancerous cells might have the tendency to be tumorigenesis. Among the more heterogeneous HepG2 cells, there are merely 1% hepatoma carcinoma cells which incorporated extremely higher amount of FucAz, 1.9-fold higher than the average amount of FucAz on the cell colony based on the 99% confidence, providing experimental evidence for the speculation that only a few cancerous cells play more invasive notorious roles during cancer metastasis even though they are all belonging to the same HepG2 cell line. Clearly, such information uncovered by ALK101-PEG2000-MS2-DOTA-Eu979 signal multiplication single-cell ICP-MS analysis would be hidden in the bulk cell ensemble analysis.

Figure 2

Quantification of the total FucAz on the single-cell surface via single-cell ICP-MS

(A) Single-cell analysis of HL7702 and HepG2 tagged with ALK-DOTA-Eu and ALK101-PEG2000-MS2-DOTA-Eu979 from 3000 independent single-cell events on the single-cell ICP-MS platform;

(B) Statistical significance of FucAz between HL7702 and HepG2 evaluated from 3000 independent single-cell events of each run (n = 21) under the bulk cell analysis and single-cell ICP-MS based on the average FucAz values with t test, p ˂ 0.001 (n = 21, data are represented as mean ± SD).

(C) Box chart for illustrating heterogeneity regarding the total FucAz on single HL7702 and HepG2 cells derived from 3000 single-cell events, in which the percentile line of the box chart was set at the segments of 10%, 25%, 50%, 75%, and 90%; the square inside denotes the average value and the line, the medium value; statistical significance of the FucAz between HL7702 and HepG2 with t test, p ˂ 0.001 (n = 21).

Quantitative breakdown of the fucosidic linkage-specific motifs on a single cell and their bioclinical implications

In addition to the total FucAz, quantitative information on its fucosidic linkage-specific motifs might provide in-depth insights for the evaluations of the different biological behavior between HepG2 and HL7702. Employing the linkage-specific fucosidases (Gotz et al., 2014; Yan et al., 2018), the fucosylation linkage-specific motifs on a single cell can be uncovered on the single-cell ICP-MS platform via ALK101-PEG2000-MS2-DOTA-Eu979 tagging (Figure S22). Firstly, the other types of fucosylation including O-fucosylation on HL7702 and HepG2 were quantified after removing the α1,2, α1,3/4 and α1,6 linkage motifs using their corresponding fucosidases together. The absolute content on HL7702 ranging from (1.0 ± 0.3) × 10−19 (10%) to (3.4 ± 0.4) × 10−19 mol/cell (90%) was determined from the 3000 valid cell events of each run (n = 21) with the average and medium values of (1.9 ± 0.5) × 10−19 mol/cell and (1.7 ± 0.4) × 10−19 mol/cell, while those on HepG2 from (8.9 ± 1.5) × 10−19 (10%) to (2.7 ± 0.3) × 10−18 mol/cell (90%) with average and medium of (1.3 ± 0.4) × 10−18 mol/cell and (1.1 ± 0.3) × 10−18 mol/cell. Average amounts of the other types account for (2.3 ± 0.6)% on HL7702 and (1.6 ± 0.5)% on HepG2 of the total FucAz (Figures 3A and 3B). Although the relative proportion on HepG2 is 0.7% lower than that on HL7702, the absolute content on HepG2 is 6.8-fold higher. This result implies the other types of fucosylation, including the O-linked fucosylation that is in charge of Notch signaling, might play certain roles during carcinogenesis processes (Taylor et al., 2014). Next, we used α1,3/4- and α1,6-specific fucosidases together to cleave α1,3/4- and α1,6-linked Fuc out of the cell-surface, the remained α1,2-linked FucAz was quantified ranging from (4.5 ± 1.1) × 10−19 mol (10%) to (1.5 ± 0.4) × 10−18 mol (90%) per HL7702 cell with the average and medium values of (8.8 ± 1.6) × 10−19 mol/cell and (8.5 ± 1.3) × 10−19 mol/cell by subtracting that of the other types of FucAz linkage-specific motifs, while (1.4 ± 0.3) × 10−18 (10%) to (5.7 ± 0.4) × 10−18 mol (90%) per HepG2 cell with average and medium values of (2.5 ± 0.2) × 10−18 mol/cell and (2.1 ± 0.3) × 10−18 mol/cell were determined, indicating that 1.2 times more heterogeneous and 2.8-fold higher α1,2-linkage FucAz metabolically incorporated on HepG2 than HL7702 (Figure 3A). It was also interesting to find that the relative percentage of (10.6 ± 1.9)% of α1,2-linkage FucAz on HL7702 is inversely 3.5-fold higher than (3.0 ± 0.3)% on HepG2 (Figure 3B). These high absolute contents but low relative percentages of α1,2-linkage FucAz on HepG2 compared with HL7702 might be one of the possible reasons for the contradictory results reported so far about its speculated functions on different kinds of cells that need to be further investigated (Belo et al., 2015). Anyway, unlike the previously reported relative quantification either by using lectin-immobilized columns and then with enzyme-linked immunosorbent assay or via the detection of fucosyltransferase activity to indirectly reflect α1,2-linkage Fuc from the cell lysates (Mathieu et al., 2004), the absolute quantity of α1,2-linkage FucAz was exactly determined on the intact single HepG2 and HL7702 cells, providing a much more solid evidence of α1,2-linked fucosylation playing important roles during tumorigenesis, besides its well-known role in the decision of A, B, and H blood type (Fukushima et al., 2009; Marionneau et al., 2001). Similarly, the absolute contents of α1,3/4-linked FucAz on HL7702 and HepG2 were quantified being from (1.6 ± 0.3) × 10−18 (10%) to (4.0 ± 0.5) × 10−18 mol per HL7702 cell (90%) and (1.0 ± 0.3) × 10−17 to (3.2 ± 0.4) × 10−17 per HepG2 cell with the average and medium values of (2.1 ± 0.2) × 10−18 mol and (2.0 ± 0.4) × 10−18 mol for HL7702 and (2.0 ± 0.3) × 10−17 mol HepG2 and (1.9 ± 0.3) × 10−17 mol for HepG2, respectively. Although 10-fold lower α1,3/4-linked FucAz content on HL7702 than HepG2 with 3.2 times heterogeneity of HepG2 cells and 2.5 HL7702 (Figure 3A) was found, the relative proportion of the average contents of (24.9 ± 2.5)% on HL7702 and (24.1 ± 3.4)% HepG2 was almost the same (Figure 3B). These results suggested again that the absolute content of α1,3/4-linked FucAz on the cell surface dominates more significant roles as thinking about the fact that some sialyl Lewis antigens, especially the overexpressed α1,3/4-linkage fucosylation-bearing sialyl Lewisx and sialyl Lewisa, interact more tightly with the selectin molecules, one possible reason leading to the slower rolling speed of cancer cells along the vascular endothelium, thus enhancing the ability of the so-called circulating tumor cells to extravasate from the vasculature into surrounding tissues (Barthel et al., 2009; Yin et al., 2010). About 2% of HL7702 cells overlapped with HepG2 (> 5.4 × 10−18 mol/cell) and, 1% HepG2 was 1.7-fold higher than the average FucAz of the HepG2 cell colony. Such a small portion with ultrahigh α1,3/4 FucAz on HL7702 and HepG2 might be an implication of either a few of the paracancerous HL7702 cells being prone to tumorigenesis or only a few of the cancerous HepG2 playing notorious roles during cancer metastasis like cancer stem cells (Barkeer et al., 2018). As for the innermost α1,6-linked FucAz, its absolute content was quantified being (2.9 ± 0.3) × 10−18 (10%) to (9.1 ± 0.6) × 10−18 (90%) with the average and medium values of (5.2 ± 0.3) × 10−18 mol and (5.1 ± 0.3) × 10−18 mol per HL7702, accounting for (62.2 ± 3.3)% of the total FucAz; while (2.5 ± 0.4) × 10−17 to (1.1 ± 0.3) × 10−16 with the average and medium values (6.0 ± 0.4) × 10−17 mol/cell and (5.8 ± 0.5) × 10−17 mol/cell accounting for (71.4 ± 5.0)% on HepG2. The absolute value on HepG2 is nearly 12-fold higher than that on HL7702 and the relative proportion is 9.2% higher than that of HL7702 (p < 0.001, n = 21) (Figures 3A and 3B). These straightforward results showed that not only the absolute amount but also relative proportion of α1,6-linked FucAz is higher on HepG2, convincingly indicating that α1,6-linked Fuc plays a critical role in cancer tumorigenesis, invasion, and cancer metastasis processes. The quantitative outcomes are well in accordance with the consensus that extremely high incorporation of the α1,6-linked fucosylation always occurred during tumorigenesis, recognizing as a biomarker of cancer (Taylor et al., 2014; Wang et al., 2015). Furthermore, 1% of the cancerous HepG2 displayed 1.9-fold higher FucAz than the average FucAz content of the cell colony, suggesting again that just a small portion of the cancerous cells playing efficacious roles in the process of invasion and metastasis. Taken together, the relative proportions and especially the absolute contents of the different linkage-specific FucAz motifs discovered on the intact single cell of HepG2 and HL7702 dig up more concealed fucosylation linkage-specific motifs information compared to the previously employed techniques (such as ESI-MS/MS and MALDI-MS/MS and lectin-based glycan array) that analyze the cell lysates of an ensemble population of cells (Acs et al., 2018; Moriwaki and Miyoshi, 2010). However, it is worthy of pointing out that the different fucosidic linkage-specific motifs must not execute independently but synergistically carry out a cellular action. More detailed investigation by professional biologists will be absolutely needed using the single-cell fucosylation breakdown methodology we developed as well as the single-cell fucosylation and its fucosidic linkage-specific motifs information we provided here.

Figure 3

Quantification and breakdown of the fucosidic linkage-specific motifs on a single cell

(A) Box chart of the α1,2, α1,3/4, α1,6 and other linkage-specific fucosylation illustrates the difference between HL7702 and HepG2 and the cell-to-cell heterogeneity among the same type of HL7702 and/or HepG2, in which the percentile line was set in the segments of 10%, 25%, 50%, 75%, and 90%, and the inside square denotes the average value and the line, the medium. Statistical significance of the different FucAz linkage-specific motifs between HL7702 and HepG2 with t test from the 3000 valid single-cell events of each run, p ˂ 0.001 (n = 21);

(B) Breakdown of the α1,2, α1,3/4, α1,6 and other linkage-specific FucAz on the single cell of HL7702 and HepG2;

(C) Screening of cancerous HepG2 from paracancerous HL7702 in the mixture of HepG2 and HL7702 cells based on the α1,2; α1,3,4; α1,6 and other linkage-specific fucosylation motifs on the single-cell ICP-MS platform from 3000 valid single-cell events, p ˂ 0.001, (n = 21).

(D) Average FucAz contents of the total and linkage-specific motifs on the single-cell surface of clinically resected liver tissues of 12 patients including 3 normal liver and 3 highly, 3 moderately, and 3 poorly differentiated HCC cells determined on the single-cell ICP-MS platform from 3000 valid single-cell events of five replicated runs (n = 5).

The discrepancy in the absolute contents of not only the total fucosylation but also different fucosylation linkage-specific motifs on the single cell can be further utilized to distinguish the cancerous HepG2 from the cell mixture of HepG2 and the paracancerous HL7702 cells (p ˂ 0.001, n = 21) on the single-cell ICP-MS platform (Figure 3C), which would be advancing for the early targeted screening of cancer. Furthermore, we applied this developed approach to the real samples of the normal liver and variously differentiated HCC tissues including highly, moderately, and poorly differentiation HCC collected from Xiamen University Affiliated Zhong Shan Hospital. In addition to the clear discrimination between normal [(1.6 ± 0.1) × 10−17 mol/cell] and highly [(3.6 ± 0.5) × 10−17 mol/cell] (p < 0.01), moderately [(5.7 ± 1.0) × 10−17 mol/cell] (p < 0.01), and poorly [(1.1 ± 0.1) × 10−16 mol/cell] (p < 0.001) differentiated HCC cells, using the average total FucAz contents determined on the single cells with p < 0.05 for highly and moderately and p < 0.01 for moderately and poorly (n = 5) differentiated HCC cells (Figure 3D), the linkage-specific motifs as well can distinguish not only the poorly and moderately but also highly differentiated HCC cells from the normal ones. Among them, the average FucAz content of α1,6 linkage-specific motif could discriminate the normal liver cells from the highly differentiated HCC cells with p < 0.05, the moderately (p < 0.01) and the poorly (p < 0.01) as determined being (1.0 ± 0.2) × 10−17 mol per normal liver cell, (2.5 ± 0.6) × 10−17 mol highly, (3.9 ± 0.6) × 10−17 mol moderately, and (7.2 ± 1.6) × 10−17 mol poorly differentiated HCC; and the highly and moderately were distinguished with p < 0.05 and, the moderately and poorly p < 0.05. Based on the average α1,2 FucAz of (1.3 ± 0.6) × 10−18 mol/cell on the normal cell and (3.3 ± 0.6) × 10−18 mol/cell the highly, they could be distinguished with p < 0.05; the 3.6- and 5.9-fold up-regulated α1,2 FucAz on the moderately and poorly differentiated HCC cells compared with the normal tissue, resulting in a statistically significant difference of p < 0.01 between the normal and the moderately and/or the poorly. It is worth noting that the average α1,3/4 FucAz content could discriminate the normal (4.6 ± 1.0) × 10−18 mol/cell from the poorly ((3.3 ± 0.6) × 10−17 mol/cell) with p < 0.01 and the moderately ((1.2 ± 0.4) × 10−17 mol/cell, p < 0.05), but less significant difference of p = 0.084 for classifying the normal from the highly differentiated HCC cells ((7.1 ± 1.7) × 10−18 mol/cell). Such results might be ascribed to that the function of α1,3/4FucAz was supposed to be in charge of the interactions between the α1,3/4-bearing sialyl-Lewis structures and selectins during tumor metastasis, which overexpressed at the poorly differentiated stage of HCC (Barthel et al., 2009). Anyway, compared with biopsies and tissue imaging that examine advanced cancer (Wald et al., 2013) and sera biomarkers (Adamczyk et al., 2012) such as AFP, CEA, CA125, and CA199 which frequently encounter the false-positive or false-negative outcomes (Table S1), the total FucAz, α1,6 and α1,2 fucosydic linkage-specific motifs contents can unambiguously discriminate highly differentiated HCC cells from the normal liver cells, in addition to the discrimination between the normal and the moderately or the poorly differentiated HCC cells. This is very significant supporting the early diagnosis and surveillance of HCC.

Conclusion

Thanks to the cell's essential salvage pathway that mediates FucAz incorporation into the cell surface glycan chain, we are able to use an ALK101-PEG2000-MS2-DOTA-Eu979 signal multiplication nanoparticle for tagging and reporting the FucAz on a single cell. The conversion of FucAz to 153Eu+ mass signal determined using ICP-MS not only avoided the inevitable Fuc ion rearrangements encountered during the soft ionization MS/MS but also provided in situ fucosylation information on an intact cell. Almost 1000 Eu atoms decorated on the ALK101-PEG2000-MS2-DOTA-Eu979 deliver an ever lowest detection limit of 4.2 zmol FucAz, allowing the quantification of the total FucAz and the quantitative breakdown of the linkage-specific fucosylation in situ on a single cell on the single-cell ICP-MS platform, uncovering the cell-to-cell heterogeneity. Even though the total FucAz revealed by the approach developed here just accounts for (11.8 ± 0.3) % on HepG2 and (8.4 ± 0.4) % HL7702 (n = 3) of the “whole fucosylation” that metabolically incorporated via both of the cell's own salvage and de novo pathways, which could be determined using the cell-lysed standard monosaccharide 1-Phenyl-3-methyl-5-pyrazolone (PMP)-derivatization and HPLC method (Figure S23), the new findings discovered on the fucosidic linkage-specific motifs on cancerous HepG2 and paracancerous HL7702 cells do provide more insights that specify their distinct roles on the cellular behavior regarding a cancering process.The capability demonstrated to distinguish the cancerous cells from the normal ones supports the surveillance and diagnosis of cancer. Distinct functions of the biomacromolecules with isomerically linkage-specific glycosylation motifs governing the marked behavior of biomacromolecule-integrated cells are universal phenomena, not limited to the cell's fucosylation demonstrated here. The developed methodology of switching a relatively unstable motif within a biomacromolecule to a stable elemental isotopic mass signal on ICP-MS should find more applications when quantitative information on the linkage-specific motifs within a biomacromolecule or on the cell surface is mandatorily required for the elucidations of its distinct roles and understanding the pathological mechanisms, thus contributing to a more accurate diagnosis. More importantly, such quantitative breakdown information on the linkage-specific motifs within the functional meaningful biomacromolecules on the cell surface has been ever more expected for the development of novel target-based inhibitory drugs and/or corresponding vaccines for targeted clinical treatment of cancer in the future.

Limitations of the study

In this study, we developed a single-cell quantitative breakdown methodology of fucosylation mediated by Ac4FucAz metabolic incorporation and ALK101-PEG2000-MS2-DOTA-Eu979 clickable tagging on a single-cell ICP-MS platform. The limitation is mainly on the correlation between the quantitative fucosylation information obtained on a single cell and their exactly biological implications that need the intensive collaboration of chemists and biologists in the future. It is also worthy of pointing out that, not limited to fucosylation and its linkage-specific motifs, a cellular action should be governed synergistically by all the glycosylation happened on the cell. Investigation on more kinds of glycosylation should be carried out for a more comprehensively multidimensional evaluation of cellular behaviors that is ongoing in our laboratory right now.

Resource availability

Lead contact

Further information and requests for resources should be directed to and will be fulfilled by the lead contact, Qiuquan Wang (qqwang@xmu.edu.cn).

Materials availability

This study did not generate new unique reagents.

Data and code availability

This study did not generate any data sets.

Methods

All methods can be found in the accompanying Transparent methods supplemental file.