Introduction

The over expression of specific cell-surface glycans correlated with the development and progression of many cancers1,2,3,4,5 and their changes are known to affect the ability of cancer cells to grow, divide and metastasize6,7. For example, sialyl Lewis X (sLex) and sialyl Lewis A (sLea) tetrasaccharides are over-expressed in gastrointestinal, pancreatic, breast and hepatic cancers and the increased expression of sLex is known to enhance tumor metastasis7. The development of sensors to rapidly detect cancer-associated glycans is of great importance for cancer diagnosis or biomarker-mediated delivery of therapeutic agents. It is extremely difficult, if not impossible, to develope specific sensor for saccharide detection since saccharide contains only one kind of recognition unit, i.e. hydroxyl group and lacks chromophore or fluorophore to afford signal readouts. Although some biomolecules, such as antibodies and natural lectins which can recognize saccharides with high affinity, have been used to construct saccharide biosensors8,9,10,11, application in cancer diagnosis and therapy is much restricted due to the difficulty in synthesis, high cost, poor stability and immunogenecity12,13,14. Arising from the unique capacity of boronic acids to form boronic esters reversibly with the 1,2 and 1,3 cis-diols presenting on many saccharides, boronic acid-based chemosensors are now proposed for saccharide detection15,16,17,18,19,20,21, due to ease of synthesis, flexibility in molecular design and inherent stability toward rigorous use. However, owing to the complexity of glycans, most of reported sensors focused on recognition of monosaccharides, not cell-surface glycans22,23,24.

For in situ recognition of cancer-associated cell-surface glycans, the chemosensors should satisfy the criteria including ease to synthesize, good biocompatibility, ability to realize recognition at constant physiological pH in aqueous media and glycan targeting ability with high selectivity. In this study, we tried to design a series of boronic acid-functionalized peptide-based fluorescent sensors (BPFSs). Peptides are the most versatile natural molecules with high biocompatibility and good water-solubility25. More importantly, since many of the receptors of bioactive peptide sequences such as arginine-glycine-aspartic acid (RGD) sequence and its receptors (integrins of αvβ3 and αvβ5)26 are over-expressed on cancer cells, the BPFSs containing bioactive peptide sequences can simultaneously target two or more cancer biomarkers to improve the accuracy in cancer cell detection and cancer diagnosis. In fact, owing to the good water-solubility of peptides and flexibility in structure designing, boronic acid-functionalized peptides have recently become most promising agents for recognition of saccharides, including monosaccharides, oligosaccharides and cancer-associated glycans14,27,28,29,30,31,32,33,34. However, there is no report on BPFSs capable of in situ recognizing cancer-associated glycans in live cells and targeting imaging of cancer cells.

In this report, by screening a series of water-soluble and biocompatible BPFSs, we demonstrated that BPFS1 with a peptide sequence of FRGDF is able to in situ recognize cancer-associated glycan of sLex with high specificity. Through the cooperation with the specific recognition between RGD sequence and its receptors, BPFS1 can targetedly label and discriminate cancer cells, presenting a great potential for cancer diagnosis.

Results

Design, synthesis and screening of fluorescent sensors

Figure 1 shows the design principle of the fluorescent sensors for in situ recognition of cancer-associated glycans in live cells and targeted imaging of cancer cells. To endow the sensors with fluorescence and ability to bind with cell-surface glycans, the architecture of anthracene-phenylboronic acid was adopted15,23,24,35,36,37. In this structural motif, the anthracene fluorescence is quenched by nitrogen lone pair electrons on an amino group. However, the binding reaction between boronic acid and saccharides facilitates the formation of B-N bond, which can confine the nitrogen lone pair electrons and lead to the increase in anthracene fluorescence15,35. To improve the water-solubility and biocompatibility, the peptide-based linkers were employed to link two anthracene-phenylboronic acid moieties to obtain the BPFSs. Through altering the peptide sequence and length, we expect to screen and obtain the BPFSs with proper spatial arrangement of the two phenylboronic acid groups that have the potential for in situ recognition of cell-surface glycans with high specificity and targeted imaging of cancer cells.

Figure 1
figure 1

The schematic illustration of BPFSs for in situ cancer cell recognition and imaging.

Full size image

Along the above design principle, five rationally designed BPFSs were synthesized (Supplementary Fig. S1–S9) and their molecular structures are shown in Figure 2A. The peptide chains of BPFSs1-3 are comprised of five natural amino acid residues. The incorporated RGD sequences, which are capable of targeting the integrins of ανβ3 and ανβ5 over-expressed on cancer cells26, are expected to cooperate with the phenylboronic acid groups to strengthen the targeting ability of BPFSs to cancer cells. To study the influence of peptide sequence and peptide length on the spatial arrangement of the two phenylboronic acid groups, BPFS4 with a peptide sequence of FAGDF and BPFS5 with a relatively long peptide sequence of FGRGDGF were also prepared together with BPFSs1-3 to form a BPFS library.

Figure 2
figure 2

(A) Molecular structures of BPFSs; (B) The maximum fluorescence intensity changes of BPFSs (1 μM) upon the addition of cancer-associated glycans of Lex, Ley, sLea and sLex (* means the maximum fluorescence intensity change was not observed within the used glycan concentration range from 0 to 300 μM); (C) Fluorescence intensity change of BPFS1 upon the addition of different amounts of sLex. λex = 370 nm, λem = 424 nm. (D) Geometry optimization of the molecular structure of BPFS1 with low energy simulated by Materials studio (MS) under Dreiding force-field.

Full size image

We first evaluated the binding affinity of BPFSs with the cancer-associated glycans. Four structurally similar glycans of Lewis X (Lex), Lewis Y (Ley), sLea and sLex were respectively added to the PBS solution of BPFSs (1 μM, pH 7.4) and the fluorescence intensity changes were determined. These four glycans were chosen because they represent the common saccharide motifs over-expressed on cancer cells6. The addition of the glycans leads to the increase in the fluorescence intensity of BPFSs (Supplementary Fig. S10) and the maximum fluorescence intensity change profiles are displayed in Figure 2B. BPFSs show various fluorescence intensity changes upon the addition of the glycans, indicating different binding affinities with the glycans. In comparison with other two fluorescent sensors, BPFS1, BPFS4 and BPFS5 having phenylalanine (Phe) residues in their peptide backbones show a relatively high affinity with one or more glycans, implying that certain interactions between Phe residues and glycans, such as the well-known CH-π interaction between the electron-deficient hydrogens on the sugar rings and π donor of electron-rich Phe31,38, play an important role in promoting the binding of BPFSs with glycans. To compare the ability of each BPFS to differentially bind with the glycans, a selectivity factor was defined as the fluorescence intensity divided by the weakest binder induced fluorescence intensity change33. As presented in Table 1, BPFS1 shows the optimal selectivity and a particular preference for binding with sLex, exhibiting more than 7-fold selectivity for sLex over Lex, around 4-fold selectivity over Ley and 3-fold selectivity over sLea. Here, assuming the formation of a 1:1 complex39, the association constant (Ka) between BPFS1 and sLex is calculated as around 4.39 × 104 M−1 from Figure 2C (the Kas between BPFS1 and other glycans are summary in Supplementary Table S1). This high selectivity of BPFS1 for sLex indicates its rational structural and conformational matching with sLex. Figure 2D displays the optimized molecular structure of BPFS1. It can be found that the stereo distances between the phenyl of Phe residues and phenylboronic acid groups on BPFS1 are shorter than other BPFSs (Supplementary Fig. S11). More importantly, owing to the extremely small dihedral (~9.8°) between the plane of anthracene-methylene linked to the phenyl of Phe residue and the plane of anthracene-methylene linked to the phenylboronic acid group, the phenyl and phenylboronic acid with a stereo distance of around 16 Å are nearly stretched along the same side of anthracene. This unique conformation will facilitate the specific binding of BPFS1 with the glycan with a proper size to form stable interaction with phenyl. From the molecular size of the glycans (Supplementary Fig. S11), the size of Ley (~14.4 Å) and sLex (~20.3 Å) is close to the distance described above (~16 Å). Although the molecular size of Ley is smaller than that of sLex, its molecular structure is more rigid because all the structural units are 6-membered rigid sugar rings (Supplementary Fig. S12), which is not favor of the stereo rotation of the bound Ley to form interaction with phenyl. However, if BPFS1 binds with the sialic acid structural unit of sLex (Supplementary Fig. S12), the bound sLex could rotate relatively easily to form interaction with the phenyl of Phe residue. The selectivity mechanism based on structural simulation and optimization is further proved by the addition of another four tetrasccharides (maltotetraose, neocarratetraose, lacto-N-tetraose and tetra-N-acetyl chitotetraose, Supplementary Fig. S13). In comparison with binding with sLex, BPFS1 shows no obvious selectivity for these four glycans (Supplementary Fig. S14).

Table 1 Selectivity factors for each BPFS screened against four different cancer-associated glycansa
Full size table

Cell selection and expression of cancer-associated glycans

The results of BPFS screening show that BPFS1 can specifically recognize sLex with high selectivity. Owing to the presence of bioactive RGD sequence on the peptide backbone of BPFS1, it theoretically shows dual-targeting functions for cancer cells, i.e. phenylboronic acid group for specifically binding with the cancer-associated glycan of sLex and RGD sequence for specifically binding with the cancer-associated integrins of αvβ3 and αvβ5. Therefore, the human hepatic cancer cell line of HepG2 with over-expressed glycan of sLex and integrins was chosen to examine the ability of BPFS1 for in situ cell recognition and targeted imaging40,41. For comparison, another three cell lines, Hep3B, HT-29 and COS7, were also studied. As the hepatoma-derived cell line, Hep3B cells have been reported to express high level of glycan of Ley 42. HT-29 cells, as human colon cancer cells, were chosen due to their reported high expression of the cancer-associated glycan of sLea 4. The COS7 cells were used because of their extremely low expression of cancer-associated glycans and integrins24.

Flow cytometry was employed to quantitatively determine the expression of the cancer-associated glycans and integrins on these chosen cell lines. Five monoclonal antibodies of anti-sLex (CSLEX-1 and KM93), anti-sLea (CSLEA-1), anti-Ley, anti-SSEA-1 (anti-Lex) and anti-CD61 (labelling integrin β3 subunit) were respectively used to stain the cells. Another monoclonal antibody of anti-CD18 was used as negative control since there is low level of corresponding antigen (integrin β2 subunit) expressed on the chosen cell lines4,24. Figure 3 shows the detailed expression of the cancer-associated glycans and integrin β3 subunit on the chosen cell lines. Except HT-29 cells that reported to express low level of integrin β3 subunit but high level of integrin β5 subunit43,44, other two cancer cell lines of HepG2 and Hep3B show a much higher expression level of integrin β5 subunit stained by anti-CD61 as compared with the normal cell line of COS7. In case of the expression of cancer-associated glycans, HepG2 cells are found to express high level of sLex stained by CSLEX-1 and KM93 and very low expressions of sLea, Ley and Lex. For the control cell lines, the antigen of Ley recognized by anti-Ley is highly expressed on Hep3B cells while the antigen of sLea recognized by the CSLEA-1 is highly expressed on HT-29 cells. The expressions of sLex, sLea, Ley and Lex are extremely low on COS7 cells.

Figure 3
figure 3

Expression of the cancer-associated glycans (sLex, Ley, sLea and Lex) and integrin β3 subunit on HepG2, Hep3B, HT-29 and COS7 cells determined by flow cytometry quantitative analysis.

Full size image

In situ recognition of cell-surface sLex and integrins for fluorescent imaging of target cells

After the determination of the expression of cancer-associated glycans and integrin, BPFS1 was incubated with HepG2 cells to evaluate its ability to in situ recognize cell-surface sLex and integrins for fluorescent imaging of the target cells. The concentration of BPFS1 used here was 20 μM because more than 85% cells were live according to the cytotoxicity assay (Supplementary Fig. S15). Figure 4 shows the confocal laser scanning microscopy (CLSM) images of the cells incubated with BPFS1 for different time intervals. No fluorescence can be found for the cells incubated without BPFS1 (Figure 4A1). After incubation with BPFS1 for 1 min, besides few of internalized molecules, most of the BPFS1 molecules are bound on the cell surface to form a uniform ring-shaped fluorescence pattern (Figure 4B1). Increasing the incubation time to 3 min, the cellular fluorescence is enhanced and the vast majority of BPFS1 molecules still remain on the cell surface (Figure 4C1). However, after incubation for 5 min, most of BPFS1 molecules have been internalized to form clusters with bright fluorescence (Figure 4D1). Further elongating the incubation time to 10 min does not change the distribution pattern of BPFS1 molecules (Figure 4E1). The results of flow cytometry quantitative analysis in Figures 5A and 5C are consistent with the above CLSM observation. The mean fluorescence intensity (MFI) of the cells rapidly increases from around 480 to 1430 as the incubation time increasing from 1 to 5 min. However, there is only a slight increase in the MFI as the incubation time elongating to 10 min (MFI, around 1540). Both the CLSM observation and flow cytometry analysis indicate that BPFS1 can rapidly bind with HepG2 cells and then traffic into cells, presenting the fluorescently labelling behavior. To demonstrate that the labelling behavior of BPFS1 is built on the in situ recognition of cell-surface sLex and integrins, the antibody of CSLEX-1 was first incubated with HepG2 cells for 15 min and BPFS1 (20 μM) was then added. As shown in Figure 4F1, although the recognition between CSLEX-1 and cell-surface sLex inhibits the binding of phenylboronic acid with sLex, BPFS1 can still label HepG2 cells to form a ring-shaped fluorescence pattern through the recognition between RGD and its receptors. However, owing to the relative low expression level of integrin β3 subunit compared to sLex as shown in Figure 3, the cellular fluorescence is rather weak (Figure 5C). In addition, the analogue of BPFS1 without phenylboronic acid groups (Supplementary Fig. S16) was also incubated with HepG2 cells. From Figure 4G1, without the phenylboronic acid groups to recognize cell-surface sLex, BPFS1 can also bind with cells with weak cellular fluorescence (Figure 5C). If replacing the RGD sequence of BPFS1 with AGD (BPFS4), the cellular fluorescence slightly decreases as displayed in Figures 4H1 and 5C. All these results strongly demonstrate that the specific recognition between phenylboronic acid groups and cell-surface sLex dominates the fluorescently labelling behavior of BPFS1 while the RGD sequence could cooperate with phenylboronic acid groups to strengthen the labelling ability of BPFS1.

Figure 4
figure 4

(A): CLSM images of HepG2 cells incubated in free culture medium; (B–E): CLSM images of HepG2 cells incubated with BPFS1 (20 μM) for 1 (B), 3 (C), 5 (D) and 10 min (E); (F): CLSM images of HepG2 cells incubated with the antibody of CSLEX-1 for 15 min and then further incubated with BPFS1 (20 μM) for another 5 min; (G): CLSM images of HepG2 cells incubated with BPFS1 analogue (20 μM) without phenylboronic acid groups for 10 min; (H): CLSM images of HepG2 cells incubated with BPFS4 (20 μM) with a peptide sequence of FAGDF for 5 min. A1–H1: confocal fluorescence field images; A2–H2: bright field images. (The scale bar is 30 μm).

Full size image
Figure 5
figure 5

Flow cytometry profiles (A), (B) and quantification of the cellular fluorescence shown via MFI (C), (D) of HepG2 cells respectively incubated with BPFS1 (20 μM) for different time (A), (C) and BPFS1 at different concentration for 5 min (B), (D).

Full size image

We also investigated the influence of the solution concentration on the labelling behavior of BPFS1. The HepG2 cells were incubated with BPFS1 at a concentration ranging from 1 to 80 μM for 5 min and then quantitatively analyzed by flow cytometry. As shown Figure 5B and 5D, the cells can be fluorescently labelled by BPFS1 and MFI gradually increases as the concentration increasing from 1 to 40 μM. However, the MFI increase is limited if further increasing the concentration to 80 μM. Based on the cellular fluorescence changes upon the addition of BPFS1 with different concentrations, the association constant (Ka) between BPFS1 and cell-surface sLex can be calculated as around 1.38 × 105 M−1 (Figure 5D). Combining the labelling behavior under different incubation times, the optimal labelling effect of BPFS1 for HepG2 cells can be achieved when incubating the cells with BPFS1 (40 μM) for 5 min. If further increasing the incubation time or the concentration, there is no obvious increase in the cellular fluorescence intensity. This result implies that self-quenching behavior does not appear for the cell bound BPFS1 molecules since the self-quenching of fluorophores generally induces decline in the fluorescence.

Discrimination of cell lines and determination of recognition sites

To examine the ability of BPFS1 to discriminate HepG2 cells from normal cells, COS7 cells were incubated with BPFS1 (40 μM, 5 min). From the CLSM image in Figure 6A1 and flow cytometry profile in Figure 6D, due to the low expression of cancer-associated glycans and integrins, the cellular fluorescence (MFI, around 230) is much weaker than that of HepG2 cells incubated under the same conditions. To prove that the cellular discrimination ability of BPFS1 is mainly built on the specific recognition between phenylboronic acid groups and cell-surface sLex but not tissue-specific, the hepatoma-derived cell line of Hep3B with over-expressed Ley and another cell line of HT-29 with over-expressed sLea were studied. Under the same conditions (40 μM, 5 min), owing to the weak binding affinity of BPFS1 with Ley (Figure 2B) and the presence RGD receptors on Hep3B cell surface (Figure 3), weak cellular fluorescence (MFI, around 300) can be observed in Figure 6B1 and 6E. The similar result can be found for HT-29 cells (Figure 6C1 and 6F).

Figure 6
figure 6

CLSM images (A–C) and flow cytometry profiles (D–F) of COS7 (A), (D), Hep3B (B), (E) and HT-29 (C), (F) cells incubated with BPFS1 (40 μM) for 5 min. A1–C1: confocal fluorescence field images; A2–C2: bright field images. (The scale bar is 30 μm).

Full size image

To determine the binding sites between BPFS1 and cell-surface sLex, neuraminidase specifically catalyzing the hydrolysis of α(2,3) sialic acid linkages and fucosidase specifically catalyzing the hydrolysis of α(1,3)- as well as (1,4)-linked fucose were respectively used to incubate with HepG2 cells for 4 h. Subsequently, BPFS1 (40 μM) was added and the cells were further incubated for another 5 min. From the flow cytometry profile presented in Figure 7, in comparison with the cells only incubated with BPFS1 (40 μM, 5 min), the addition of neuraminidase results in around 72% decrease in MFI while the addition of fucosidase induces around 78% decrease in MFI, indicating that both sialic acid and fucose residues of cell-surface sLex are the binding sites for BPFS1. To prove this statement, fructose (400 μM), one of the strongest 1:1 boronic acid binders45,46,47, was employed to incubate with HepG2 cells labelled by BPFS1. It is noteworthy that the effective competition can be only observed when fructose with a high concentration (at least 10-fold excess compared to BPFS1 concentration) is added. The possible reason is that the monosaccharide of fructose poorly competes with the multi-saccharides of cell-surface sLex since the multivalent interactions are nearly stronger than the sum of monovalent interactions33,48. As shown in Figure 7B, because the competitive reaction between fructose and sLex with BPFS1 leads to the shedding of the cell bound BPFS1 molecules, there is around 85% decrease in MFI after 30 min incubation. With respect to the residual cellular fluorescence, it mainly corresponds to the cell bound BPFS1 molecules via the interaction between RGD segments and the receptor of integrins, which is in agreement with the CLSM observation in Figures 4F1 and 4G1. The results of enzymatic hydrolysis and fructose competition not only provide the information for the binding sites of BPFS1 with cell-surface sLex, but quantitatively indicate the influence of RGD sequence on the binding affinity of BPFS1 with sLex. From the Figure 7B, with addition of neuraminidase to hydrolyze sialic acid linkage of sLex into Lex, the cellular fluorescence shows a 3.6-fold decrease, which is lower than decreased value (7.7-fold) revealed in Figure 2B, implying that the interaction between RGD sequence and its receptors could improve the binding affinity of phenylboronic acid group with cell-surface sLex and thus strengthen the labelling and targeting ability of BPFS1. This interesting finding provides a valuable chance to use BPFS1 as probe for clinical cancer diagnosis since the dual-targeting functions of BPFS1 can facilitate each other and simultaneously recognize two types of cancer-biomarkers.

Figure 7
figure 7

Flow cytometry profiles (A) and quantification of the cellular fluorescence shown via MFI (B) of HepG2 cells incubated with BPFS1 (40 μM) for 5 min (BPFS1), incubated with neuraminidase for 30 min and then further incubated with BPFS1 (40 μM) for 5 min (Neuraminidase), incubated with fucosidase for 30 min and then further incubated with BPFS1 (40 μM) for 5 min (Fucosidase) and incubated with BPFS1 (40 μM) for 5 min and then further incubated with fructose (400 μM) for 30 min (Fructose).

Control represents the cells incubated without BPFS1.

Full size image

Discussion

Through structure-based screening and design, our newly developed BPFS1 in this work represents the first example of water-soluble and biocompatible fluorescent sensors with dual-targeting functions to in situ recognize and discriminate structurally similar cancer-associated glycans and cancer cell lines. Through the cooperation of the phenylboronic acid group for specifically recognizing cell-surface sLex and RGD sequence for recognizing its receptors, BPFS1 can rapidly bind to the surface of HepG2 cells and then traffic into cells, showing the ability to targetedly and fluorescently label HepG2 cells. Our findings agree with the fact that the sialic acid containing glycans such as sLex are organized on the cell membrane and reside in the intracellular organelles along the exocytotic and endocytic pathways49,50. Owing to this sLex-mediated endocytosis and the interaction between RGD and its receptors to enhance the stability of the complex between BPFS1 and sLex, the binding affinity of BPFS1 for sLex on cell surface (Ka = 1.38 × 105 M−1, see Fig. 5D) is much stronger than that in an aqueous medium (Ka = 4.39 × 104 M−1, see Fig. 2C). To fully understand the structural features of BPFS1 that induce the specific recognition of cell-surface sLex, more computational and conformational work is needed. It is known that over-expression of sLex has been found in chronic inflammatory diseases of the liver and thus regarded as one of the most important biomarkers for hepatic cancers40,41. For clinical cancer diagnosis, the diagnostic accuracy will be significantly improved if combining the recognition of sLex with other cancer-associated biomarkers. Owing to the dual-targeting functions of BPFS1, i.e. phenylboronic acid group for recognition of cancer-associated sLex and RGD sequence for recognition of cancer-associated integrins of αvβ3 and αvβ5, our results suggest that it can be considered as useful fluorescent probe with high accuracy for cancer diagnosis. Further development along this line could lead to a number of small molecule fluorescent probes that can be used for cell labeling, drug delivery and selective imaging applications.

Methods

Materials

The cancer-associated glycans (sLex, sLea, Ley and Lex) and monoclonal antibodies (anti-sLex, anti-sLea, anti-Ley, anti-Lex and anti-CD61) stored in PBS buffer were purchased from Sigma-Aldrich. Modified Eagle's Medium (DMEM), 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazoliumbromide (MTT), fetal bovine serum (FBS), trypsin and penicillinestreptomycin were purchased from Invitrogen Corp.

Chemical synthesis

Details of the synthesis of BPFSs are shown in Supplementary data.

Screening of BPFSs

Fluorescence spectroscopy was used to determine the affinity and selectivity of each BPFS for the cancer-associated glycans. The PBS solution of each BPFS at a concentration of 1 μM was prepared and the fluorescence emission spectra with the addition of different amount of glycans were recorded on a LS55 luminescence spectrometry (Perkin-Elmer) with excitation at 370 nm and emission data range between 380 and 500 nm. The BPFSs show different degree of increase in the fluorescence as the addition of the glycans and the maximum fluorescence intensity change at 424 nm were used to evaluate the selectivity of each BPFS for the cancer-associated glycans.

Determination of antigens

The cells (HepG2, Hep3B, HT-29 and COS7) were first incubated in DMEM containing 10% FBS and 1% antibiotics (penicillin-streptomycin, 10,000 U/mL) at 37°C in a humidified atmosphere containing 5% CO2. Cells were then harvested and seeded in 24-well plate. After incubation for 24 h, l μL of monoclonal antibodies of anti-sLex (CSLEX-1 and KM93), anti-sLea (CSLEA-1), anti-Ley, anti-SSEA-1 (anti-Lex) and anti-CD61 stored in PBS buffer were respectively used to stain the cells for 15 min. Thereafter, the cells were washed with fresh DMEM (3 × 0.5 mL) and 2 μL of fluorescein isothiocyanate-conjugated goat anti-mouse IgM (0.1 mg/mL stored in PBS buffer) was added to stain the cells for 15 min. As a negative control, the cells were incubated with another monoclonal antibody of anti-CD18 (1 μL, 0.5 mg/mL stored in PBS buffer) for 15 min and then stained with fluorescein isothiocyanate-conjugated mouse IgG1/IgG2 (2 μL, 0.1 mg/mL stored in PBS buffer). After washing with fresh DMEM (3 × 0.5 mL), the cells were collected for flow cytometry analysis (BD FACSAria™ III, USA) with excitation at 488 nm.

Cytotoxicity assay

Based on the results of structure screening, BPFS1 shows a high affinity and optimal selectivity for sLex. Therefore, BPFS1 was chosen to evaluate its cytotoxicity by using MTT assay. In brief, the cells (HepG2, Hep3B, HT-29 and COS7) were seeded in 96-well plate and then incubated in 100 μL of DMEM containing 10% FBS for 24 h. Thereafter, the medium was removed and 100 μL of DMEM containing BPFS1 was added. After incubation for 48 h, the medium was replaced with 200 μL of fresh medium and 20 μL of MTT (5 mg/mL in PBS buffer) solution was added to each well. After incubation for 4 h, the medium was replaced by 200 μL of DMSO and the mixture was shaken at room temperature for several minutes. The optical density (OD) was measured at 570 nm with a microplate reader, model 550 (BIO-RAD, USA). The average value of four independent experiments was collected and the cell viability was calculated as: cell viability (%) = (OD570 (FS)/OD570 (control)) × 100, where OD570 (control) is obtained in the absence of BPFS1 and OD570 (FS) is obtained in the presence of BPFS1.

In situ recognition of cell-surface sLex and integrins for fluorescent imaging of target cells

The cells (HepG2, Hep3B, HT-29 and COS7) were seeded in 24-well plate and then incubated in 100 μL of DMEM containing 10% FBS for 24 h. Thereafter, the medium was removed and 200 μL of DMEM containing BPFS1 was added. After incubation for a fixed period, the medium was removed. After washing the cells with fresh DMEM (3 × 1 mL), the cells were viewed under CLSM with excitation at 400 nm and collected for flow cytometry quantitative analysis with excitation at 375 nm. As a control, the BPFS1 analogue of without phenylboronic acid groups and BPFS4 dissolved in DMEM (20 μM) were also respectively incubated with HepG2 cells and the cells were subsequently viewed under a lasers canning confocal microscope (Nikon C1-si TE2000, Japan) with excitation at 405 nm and collected for flow cytometry quantitative analysis (BD FACSAria™ III, USA) with excitation at 375 nm.

Determination of recognition mechanism

HepG2 cells were seeded in 24-well plate and then incubated in 100 μL of DMEM containing 10% FBS for 24 h. Subsequently, the medium was removed and the cells were respectively treated with neuraminidase specifically catalyzing the hydrolysis of α(2,3) sialic acid linkages and fucosidase specifically catalyzing the hydrolysis of α(1,3)- as well as (1,4)-linked fucose according to modifications of manufacturer's protocols. After incubation for 4 h, the cells were washed with fresh DMEM (3 × 1 mL) and 200 μL of DMEM containing BPFS1 (40 μM) was added. After incubating for 5 min and then washing with fresh DMEM (3 × 1 mL), the cells were collected for flow cytometry quantitatively analysis (BD FACSAria™ III, USA) with excitation at 375 nm. To examine the competition reaction between fructose and sLex for BPFS1, HepG2 cells were first incubated in the DMEM containing BPFS1 (40 μM) for 5 min. After washing with fresh DMEM (3 × 1 mL), the cells were further incubated in 200 μL of DMEM containing fructose (400 μM) for 30 min. Thereafter, the cells were washed with fresh DMEM (3 × 1 mL) and collected for flow cytometry quantitative analysis.