Uncovering the Binding Specificities of Lectins with Cells for Precision Colorectal Cancer Diagnosis Based on Multimodal Imaging

Abstract There is a high desire for novel targets/biomarkers to diagnose and treat colorectal cancer (CRC). Here, an approach starting from a polyacrylamide hydrogel–based lectin microarray is presented to screen the high expression of glycans on the CRC cell surface and to identify new lectin biomarkers for CRC. Three common CRC cell lines (SW480, SW620, and HCT116) and one normal colon cell line (NCM460) are profiled on the microarray with 27 lectins. The experimental results reveal that CRC cells highly express the glycans with d‐galactose, d‐glucose, and/or sialic acid residues, and Uelx Europaeus Agglutinin‐I (UEA‐I) exhibits reasonable specificity with SW480 cells. After conjugation of UEA‐I with silica‐coated NaGdF4:Yb3+, Er3+@NaGdF4 upconversion nanoparticles, the follow‐up in vitro and in vivo experiments provide further evidence on that UEA‐I can serve as tumor‐targeting molecule to diagnose SW480 tumor by multimodal imaging including upconversion luminescence imaging, T 1‐weighted magnetic resonance imaging, and X‐ray computed tomography imaging.

The ORCID identification number(s) for the author(s) of this article can be found under https://doi.org/10.1002/advs.201800214.

DOI: 10.1002/advs.201800214
is strongly dependent on the detection of tumor at an earlier stage (early diagnosis), the correct identification of the subtype of cancer (precision treatment) and an accurate prediction of its likely course after suitable treatment (prognosis). For instance, the five year survival rate of patients with advanced CRC is significantly lower than that of patients with CRC at an early stage. Therefore, there is an urgent demand for novel biomarkers to diagnose CRC with high accuracy and sensitivity.
Glycans such as glycolipids and glycoproteins involve in many physiological processes including cell differentiation and proliferation, cell-cell communication, immune response, and tumor growth and metastasis. [4][5][6][7] Numerous cancers including CRC have been known to relate with structural abnormalities of N-and/or O-linked glycoproteins. [8,9] Several glycoproteins are reported as candidate biomarkers for cancers, and a part of them have been approved for clinical diagnosis of cancers by the US Food and Drug Administration. [10,11] Currently, various methods/assays have been developed for glycan detection both on cell surface and in body fluids, such as high performance liquid chromatography, capillary electrophoresis, mass spectrometry (MS), enzyme-linked immunosorbent assay, and microarrays. [12][13][14][15] Lectins, the carbohydrate binding proteins, exhibit high specificities for saccharide moieties, and can perform specific glycan recognition on the cellular and molecular level. Like antibodies, some lectins are known to play important roles in the immune system including defense against invading microorganisms and modulation of inflammatory and autoreactive processes through specific recognition of cellular and/or bacterial surface saccharide moieties. These characteristics make lectins very useful for glycoanalysis and medicine development. [16][17][18][19][20][21] Due to their minimization of sample consumption and high throughput format for analyzing multiple targets simultaneously, lectin microarrays have been employed in the glycomics study. [22][23][24][25][26] However, the interactions of lectins with glycans (dissociation constant, K d = 10 −4 -10 −7 m) are much weaker than the interactions of antigens with antibodies (K d = 10 −8 -10 −12 m). The cells are easily washed off from lectin microarray during the subsequent cleaning and drying process since the binding affinity between a single glycan molecule and lectin on a

Introduction
Colorectal cancer (CRC) is one of the commonest alimentary system malignancy, which has relatively high incidence and mortality. [1][2][3] Because of the commonality of malignancies including metastasis and recurrence, successful CRC therapy planar surface is relatively low, leading to reducing of the accuracy and reproducibility of analysis results. The drawback limits the application of conventional 2D lectin microarray for directly profiling glycan expression on cell surface. Comparing with 2D planar substrates, 3D substrates have many advantages including high probe loading capacity and optimal reaction space for adjusting biomolecular distribution. [27][28][29][30][31] In particular, the carbohydrate-lectin interactions can be strengthened on the 3D microarray through formation of multivalent binding among the immobilized probe molecules. However, there are a few examples of 3D lectin microarrays for profiling cellular glycan expression and screening lectin biomarker.
Multimodal imaging has many advantages such as improving diagnostics or guidance through the analysis of complementary, data-rich, coregistered images to address weaknesses in individual imaging modalities. [32,33] Because of their unique upconversion luminescence (UCL) and strong contrast enhancement of magnetic resonance imaging (MRI) and X-ray computed tomography (CT) imaging, NaGdF 4 -based upconversion nanoparticles (UCNPs) have been demonstrated as one of the most appealing contrast agents for multimodal imaging (UCL/MRI/CT). [34][35][36][37][38] Due to undesired accumulation of nanoparticles (NPs) in the liver, spleen, and kidneys, only a small proportion (less 10% of the injected dose (ID) g −1 ) of passive tumor-targeting NPs can reach the tumor sites through the enhanced permeability and retention effect. [35,39] The tumor accumulation efficiency can be improved through conjugation of NPs with tumor-targeting ligand which can specifically bind with the receptor overexpressed on the membrane of tumor cell or tumor vasculature cell. [40][41][42] Because aberrant glycan patterns have already been considered as a hallmark of cancers, lectins may serve as tumor-targeting agents through multivalent binding of glycans on the cell surface. For example, concanavalin A-modified UCNPs have been successfully used to label cell surface glycan labeling and differentiate between HCCHM3 and CL cells. [43] In present study, a polyacrylamide (PAAM) hydrogel-based lectin microarray has been fabricated for profiling glycan expression on CRC cell surface and screening new lectin biomarkers for CRC. The immobilized lectins on PAAM hydrogel may serve as multivalent binding scaffolds to the cellular glycans, resulting in increased binding affinity and selectivity. Uelx Europaeus Agglutinin I (UEA-I) has been demonstrated to have high affinity and specificity with SW480 cells through the interactions of 27 different lectin species with 4 distinct cell types. Using silicacoated NaGdF 4 :Yb 3+ , Er 3+ @NaGdF 4 UCNP as nanoprobe, the UEA-I has been successfully applied to differentiate SW480 cells and mouse-bearing SW480 tumor by multimodal imaging including UCL imaging, T 1weighted MRI, and CT imaging.

PAAM Hydrogel-Based Lectin Microarray Fabrication
The PAAM hydrogel microarray was prepared by our previously reported method with slight modifications (as shown in Figure 1). [44] The morphology of PAAM hydrogel spot was characterized by X-ray photoelectron spectroscopy (XPS), scanning electron microscopy (SEM), and atomic force microscopy (AFM). After gelation, the N1s peak has been clearly observed on the surface of PAAM hydrogel spot (as shown in Figure S1 in the Supporting Information). [45,46] The result demonstrates that the PAAM hydrogel microarray has been successfully fabricated. Representative SEM and AFM images reveal that the PAAM hydrogel spot has a relatively rough surface and the size of PAAM hydrogel spot is about 500 µm in diameter (as shown in Figures S2 and S3 in the Supporting Information).
The results indicate that the PAAM hydrogel spot can provide relatively large surface area for immobilization of lectin molecules. The carboxyl groups of PAAM hydrogel were activated by traditional 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride (EDC)/N-hydroxysulfosuccinimide sodium salt (sulfo-NHS) reaction, which leads to the formation of an outer layer of NHS ester group on PAAM hydrogel. The PAAM hydrogel-based lectin microarray was achieved by noncontact spraying lectin droplets to activate PAAM hydrogel microarray. The amino residues of lectin can react with NHS ester group to form stable amide bond.

www.advancedsciencenews.com
to evaluate the binding performance of PAAM hydrogel-based lectin microarray with living cells since HeLa cells have high affinities with these lectins. [47][48][49] As shown in Figure S4 (Supporting Information), the relative high signal intensities can be achieved with 1 mg mL −1 lectins in spotting solution and 1 × 10 6 cells mL −1 HeLa cells in cell buffer. The high signal-tobackground ratio (S/N) is obtained with incubation of HeLa cells with lectins microarray for 40 min. Taking into account the sensitivity and specificity of the cell-lectin interaction, the following experiments were carried out under the optimized conditions: 1 mg mL −1 lectins in spotting solution, 1 × 10 6 cells mL −1 cells in cell buffer, and 40 min for incubation.

Screening Specific Lectin for CRC Cells
To screen specific binding of lectin with CRC, we examined the interactions of 27 lectins with 4 cell lines including three CRC cell lines (SW480, SW620, and HCT116) and one normal colon cell line (NCM460). The three CRC cell lines are derived from three kinds of common CRC. Generally, SW480 and HCT116 cells are derived from the stage 2 primary CRC and stage 4 primary CRC, respectively, while SW620 cells are derived from the stage 3 CRC with lymph node metastasis. [50] Because RCA 120 has strong binding affinity with almost all mammalian cells, the binding of RCA 120 with cells is used as the internal control in this study. After incubation with selected cells, the fluorescence intensities of lectin-cell pairs (the lectin arrangement is shown in Table S3 in the Supporting Information) were recorded. Then, the relative fluorescence intensity (F R = (F Lectin − F 0 )/ (F RCA 120 − F 0 )) was calculated and employed to evaluate the binding affinity of lectin with living cells, where F Lectin , F RCA 120 , and F 0 are the average fluorescence intensity of a certain lectincell pair, the average fluorescence intensity of RCA 120-cell pair, and the average fluorescence intensity of control, respectively. The fluorescence images of microarray and the corresponding F R are illustrated in Figure 2. The binding affinities of lectins with cells were divided into 4 grades, i.e., strong binding (F R ≥ 0.8, indicated as: +++), medium binding (0.5 ≤ F R < 0.8, indicated as: ++), weak binding (0.2 ≤ F R < 0.5, indicated as: +), and nonbinding (F R < 0.2, indicated as: −), which were summarized in Table 1. As expected, all of these cells have reasonable binding affinities with RCA 120, which suggests that the cells express high level of d-galactose (Gal) residues in their surface glycans. This phenomenon is consistent with previous studies. [51][52][53] All 4 cell lines herein have strongly binding behaviors with 9 lectins including wheat germ agglutinin (WGA), maackia amurensis lectin I (MAL I), erythrina cristagalli lectin (ECL), phaseolus vulgaris agglutinin (PHA-E+L), LCA, datura stramonium lectin (DSL), maackia amurensis lectin II (MAL II), PSA, and RCA 120, which indicates that the carbohydrate residues on these cellular surfaces have partial consistency. For instance, these cells may express high levels of Gal, d-glucose, and/or sialic acid residues on their cellular surfaces. The difference in the lectin binding patterns of CRC cells and the normal colon cells has also been demonstrated, suggesting differences in carbohydrate motifs on the cellular surfaces. For example, three lectins (narcissus pseudonarcissus (daffodil) lectin (NPL), galanthus nivalis lectin (GNL), and elderberry bark lectin (EBL)) exhibit reasonable binding affinities with SW480 cells and HCT116 cells, while these lectins have poor binding affinities with SW620 cells and NCM460 cells. The result indicates that SW480 cells and HCT116 cells strongly express mannose residue complexes and/or Neu5Acα6Gal glycosyl complexes. In particular, only SW480 cells show strong binding affinity with UEA-I (as shown in Figure S5 in the Supporting Information), suggesting that SW480 cells express high level of α-1,2-fucosylation complexes on the surface. [54] This phenomenon is consistent with the high expression level of H α-2fucosyltransferase messenger RNA (mRNA) in SW480 cells. [55] Therefore, UEA-I could be used as a lectin biomarker to discriminate SW480 CRC subtype.

Preparation and Characterization of UCNP@SiO 2 -UEA-I
To verify its capability, the UEA-I was covalently conjugated with UCNP@SiO 2 COOH through reactions of the carboxyl groups  on the UCNP surfaces with the primary and secondary amines residues of UEA-I. [56] In this case, UCNP@SiO 2 COOH was used as nanoprobe because that carboxyl-terminated SiO 2 shell has weak nonspecific binding with cellular membrane and NaGdF 4 :Yb 3+ , Er 3+ @NaGdF 4 UCNP can serve as multifunction contrast agent for multimodal imaging (UCL/MRI/ CT). UCNP@SiO 2 COOH was synthesized by previously reported strategy. [57,58] The UCL emission of NaGdF 4 :Yb 3+ , Er 3+ @NaGdF 4 is much stronger than that of NaGdF 4 :Yb 3+ , Er 3+ UCNPs since the outer NaGdF 4 shell can protect the migrating energy in Gd sublattice from trapping by surface quenchers. [59] The average size of UCNP@SiO 2 COOH is 21.0 ± 0.5 nm in diameter including a NaGdF 4 :Yb 3+ , Er 3+ core with the size of 10.0 ± 0.5 nm in diameter, a NaGdF 4 inner shell with the thickness of ≈2.5 nm and a SiO 2 outer shell with the thickness of ≈3 nm (as shown in Figure S6 in the Supporting Information). After UEA-I conjugation, the morphology, dispersity, and UCL emission of UCNP@SiO 2 COOH exhibit a negligible change (as shown in Figure S6 in the Supporting Information). The corresponding X-ray diffraction patterns (as shown in Figure S7 in the Supporting Information) of NaGdF 4 :Yb 3+ , Er 3+ UCNPs and NaGdF 4 :Yb 3+ , Er 3+ @NaGdF 4 UCNPs indicate that they are pure hexagonal phase (JCPDS No. 27-0699). The XPS measurements clearly show the element of Si in UCNP@SiO 2 COOH and the elements of N and Si in UCNP@SiO 2 -UEA-I (as show in Figure S8 in the Supporting Information). The Fourier transform infrared (FTIR) spectrum of UCNP@SiO 2 -UEA-I exhibits one peak at 1639 cm −1 , which is associated with stretching vibration of amide bonds (as shown in Figure S9 in the Supporting Information). The XPS and FTIR results confirm that UEA-I has been successfully conjugated on UCNP@SiO 2 COOH surface. After conjugation of UEA-I, the hydrodynamic diameter and zeta potential of UCNP@SiO 2 COOH are changed from 29.4 ± 1.1 to 108.1 ± 2.0 nm, and −43.7 ± 0.5 to −10.5 ± 1.0 mV, respectively, which gives further evidence on successful preparation of UCNP@SiO 2 -UEA-I (as shown in Figure S10 in the Supporting Information).
To evaluate their contrast enhancement capacities for MR/ CT imaging, T 1 -weighted MR and CT imaging of UCNP@ SiO 2 COOH and UCNP@SiO 2 -UEA-I were investigated. Both MR and CT signals are linearly increased with increasing the concentrations of UCNP@SiO 2 COOH and UCNP@SiO 2 -UEA-I (as shown in Figures S11 and S12 in the Supporting Information). The molar longitudinal relaxivities (r 1 = 1/T 1 , the slopes of lines in Figure S12 in the Supporting Information) of UCNP@ SiO 2 COOH and UCNP@SiO 2 -UEA-I were calculated to be 7.05 and 8.73 mm −1 s −1 , respectively. The higher r 1 value of UCNP@ SiO 2 -UEA-I might be due to the polar amino acids in UEA-I and the rigidity of UEA-I, which increase the density of water molecule around gadolinium ions through accelerating the exchange rate of intraspherical water molecule. [60,61] The hounsfield unit (HU) values (200) of 4.3 mg mL −1 UCNP@SiO 2 COOH and UCNP@ SiO 2 -UEA-I are equal to that of 8.0 mg mL −1 iodine in Omnipaque (a contrast agent for clinical CT imaging), indicating that these NPs have high CT contrast enhancement ability.
In addition, the SW480 cells still have more than 80% viability after incubated with as high as 200 µg mL −1 NPs for 24 h (as shown in Figure S13 in the Supporting Information), suggesting that both of UCNP@SiO 2 COOH and UCNP@SiO 2 -UEA-I exhibit low cytotoxicity.

Interactions of UCNP@SiO 2 -UEA-I with Cells
As shown in Figure 3 Figure S14 in the Supporting Information). In addition, the MR and CT signal intensities of UCNP@SiO 2 -UEA-I-stained SW480 cells are much higher than those of UCNP@SiO 2 -UEA-I-stained other cells and that of UCNP@SiO 2 COOH-stained SW480 cells (as shown in Figure S15 in the Supporting Information). The results confirm that UEA-I has high affinity with SW480 cells. including heart, liver, spleen, lung, and kidneys were collected at 2 and 24 h postinjection of NPs. The amounts of Gd element in these tissues were then analyzed by inductively coupled plasma mass spectrometry (ICP-MS) (as shown in Figure 6). After 24 h injection, the accumulation amount of UCNP@ SiO 2 -UEA-I in SW480 tumor (15.0% ID g −1 ) is much higher than those of UCNP@SiO 2 COOH in SW480 tumor (7.8% ID g −1 ), UCNP@SiO 2 -UEA-I in HCT116 tumor (8.5% ID g −1 ), and UCNP@SiO 2 COOH in HCT116 tumor (7.6% ID g −1 ). These experimental results clearly demonstrate that the UEA-I can serve as a biomarker for SW480 tumor diagnosis through specific binding with α-1,2-fucosylation glycan on the cellular surface.

In Vivo Toxicology Investigation
The histochemical analysis indicates that there are no notable evidences of tissue damage and adverse effect of the UCNP@ SiO 2 -UEA-I to major organs of BALB/c mice including heart, liver, spleen, lung, and kidneys at 30 d postinjection (as shown in Figure S16 in the Supporting Information). Comparison with the control group, UCNP@SiO 2 -UEA-I-treated mice exhibit negligible difference on the numbers of blood cells (as shown in Table S3 in the Supporting Information). The results confirm the low toxicity of UCNP@SiO 2 -UEA-I.

Conclusion
In summary, the present study provides a proof-of-principle demonstration of the screening candidate lectin biomarkers for CRC through direct profiling of binding affinities of living cells with lectins on PAAM hydrogel microarray. UEA-I is   (UCL/MRI/CT) studies. This exemplifies its potential for noninvasive diagnosis of CRC at subtype level. Although only limited interactions of cells with lectins have been studied, the finding would open up possibilities for the future to discover lectins as biomarkers toward a broader biomedical application including cancer diagnosis and therapy.
Screening Specific Lectin for CRC Cells: Three common CRC cell lines (SW480, SW620, and HCT116), one normal colon cell line (NCM460) and one cervical carcinoma cell line (HeLa) were cultured in the desired fresh medium supplemented with 10% fetal bovine serum and 100 U mL −1 penicillin-streptomycin in humidified air with 5% CO 2 at 37 °C. Leibovitz's L-15 was used for culturing SW480 and SW620 cells, McCoy's 5A was used for culturing HCT116 and NCM460 cells, and Dulbecco's modified Eagle medium was used for culturing HeLa, respectively.
In Vitro UCL, MR, and CT Imaging: For UCL imaging, SW480 cells, SW620 cells, HCT116 cells, and NCM460 cells (5 × 10 4 cells per well in 0.5 mL culture medium) were seeded in 48-well culture plates for 24 h.

www.advancedscience.com
The culture medium was discharged and the cells were washed with PBS. Subsequently, 100 µg mL −1 UCNP@SiO 2 COOH or UCNP@ SiO 2 -UEA-I in 0.5 mL fresh culture medium was introduced into each well and incubated at 37 °C for another 0.5 h, respectively. After washed with PBS (3 times), the NP-stained cells were fixed with 4% paraformaldehyde for 20 min and subjected to UCL imaging under an external 980 nm laser (0.5 W cm −2 ). For CT and MR imaging, SW480 cells, SW620 cells, HCT116 cells, and NCM460 cells (1 × 10 6 cells per well in 2.5 mL culture medium) were seeded in 6-well culture plates for 24 h. The culture medium was discharged and the cells were washed with PBS. Then, 100 µg mL −1 UCNP@SiO 2 COOH or UCNP@SiO 2 -UEA-I in 2.5 mL fresh culture medium was introduced into each well and incubated at 37 °C for another 1 h, respectively. The cells were detached by 1 mL trypsin and centrifuged at 1000 rpm for 5 min. The supernatants were discharged. Subsequently, 1 × 10 6 cells were immobilized in the 1.5 mL Eppendorf tubes by 1% agarose, respectively. The corresponding unstained SW480 cells were employed as control samples. T 1 -weighted MR images were collected using a GE Signa 1.5-T MR unit with the following imaging parameters: repetition time (TR), 240 ms; echo time (TE), 15.9 ms; field of view, 120 mm × 72 mm; slice thickness, 2.0 mm. CT imaging was performed as previously described.
In Vivo UCL, MR, and CT Imaging: BALB/c nude mice (six week old, 20 ± 0.2 g, male) were purchased from Beijing HFK Biotechnology Ltd. (Beijing, China). Animal experiments were conformed to the guidelines of the Regional Ethics Committee for Animal Experiments established by the Jilin University Institutional Animal Care and Use. The mice were subcutaneously inoculated with SW480 cells and HCT116 cells, respectively.
For in vivo multimodal (UCL/MR/CT) imaging, the SW480 tumorand HCT116 tumor-bearing nude mice were injected intravenously with NaCl solution (0.9 wt%, 200 µL) containing the desired amounts of UCNP@SiO 2 COOH or UCNP@SiO 2 -UEA-I (Gd 3+ : 1.5 mg mL −1 for UCL, 1.5 mg mL −1 MR imaging, and 15 mg mL −1 for CT imaging), respectively. The images were collected at the appropriate time points after injection. The in vivo UCL images were obtained by a CCD camera with a 980 nm NIR laser at the power density of 1.0 W cm −2 as excitation light. In vivo MR and CT images were collected as previously described, except a 129 mm field of view was used for CT.
In Vivo Biodistribution and Toxicology Investigation: For biodistribution study, the SW480 tumor-and HCT116 tumor-bearing nude mice were sacrificed at 2 and 24 h postinjection of UCNP@SiO 2 COOH or UCNP@SiO 2 -UEA-I (200 µL, 1.5 mg mL −1 Gd 3+ ), respectively. Then, the distribution of UCNP@SiO 2 COOH or UCNP@SiO 2 -UEA-I in tumors and main organs including heart, liver, spleen, lung, and kidney were digested in aqua regia under heat treatment (80 °C) for 2 h, and the as-obtained liquids were subjected to ICP-MS analysis. Histology analysis and blood biochemistry assay were employed to evaluate the biosafety of UCNP@SiO 2 -UEA-I.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.