Characterization, selenylation, and antineoplastic effects on HepG2 cells in vitro and in vivo of an arabinofuranan from the fruits of Akebia quinata

2.1 Materials and chemicals

The detailed information of materials and reagents used in presented study can be found in the Supplementary information.

2.2 Isolation and purification of polysaccharide

The crude polysaccharide was prepared by extracting the fruits of A. quinata according to our previous report (Wang, et al., 2021). Firstly, the dried medicinal herbs (5.0 kg) were extracted with hot water (90 ℃) three times at 1:6 g/mL for two hours each time. As the solid residue was filtered out, the filtrate was concentrated under reduced pressure to one tenth of its original volume, and then centrifuged to collect the supernatant. Anhydrous ethanol was added into the concentrated extract to adjust the final concentration to 50 % and 70 %, respectively. Then, the 70 % precipitation was deproteinized with Sevag reagent (dichloromethane: n-butanol = 4: 1, v/v) (Hoseiniyan Benvidi and Jahanbin, 2020), followed by dialysis (3500 Da cut-off) and freeze drying to obtain crude polysaccharide named AQ70.

AQ70 was applied to DEAE-Sepharose Fast Flow (DEAE-FF) column (Ø 3.5 × 40 cm) and eluted first with distilled water, and then with gradient concentration NaCl solution (0.05, 0.1, and 0.2 M) at 1.0 mL/min flow rate. The fraction eluted with 0.05 M NaCl was collected and named AQ70-2. Furthermore, AQ70-2 was purified by Sephadex G-75 gel filtration column (Ø 1.5 × 100 cm) at 0.8 mL/min flow rate. The first elution band was collected and AQP70-2A was harvested successively by concentration and freeze-drying.

2.3 Purity, homogeneity and molecular weight analysis

The UV–visible spectrophotometer was employed to perform band scanning at 200–400 nm to detect the presence of nucleic acids and proteins in the polysaccharides sample. Furthermore, the molecular parameters of AQP70-2A were determined by high performance size exclusion chromatography (HPSEC) in line with previous report (Wang et al., 2022). Detailed information about the operation process can be found in the Supplementary information.

2.4 Identification of monosaccharide composition

The monosaccharide composition of AQP70-2A was analyzed according to the method previously reported (Zhang et al., 2020a). The Supplementary information shows the detailed description of the experiment procedures.

2.5 Fourier transform infrared spectrum and methylation analysis

The dried AQP70-2A sample (1 mg) was ground with potassium bromide powder before pressed into sheets, and analyzed by Fourier transform infrared (FT-IR) spectrometer under a wave number range of 4000 to 400 cm⁻¹.

Glycosidic linkages of the polysaccharide were determined by methylation analysis as described previously (Zhang et al., 2020). The detailed steps of methylation analysis are given in the Supplementary information.

2.6 NMR analysis

The ¹H NMR, ¹³C NMR, DEPT-135 NMR, HSQC, and HMBC were recorded using a 400 MHz Bruker AV-400 spectrometer (Bremen, Germany), and analyzed by MestReNova software (Version 11.0.1, Mestrelab Research SL, Spain).

2.7 Congo red assay

The solution of polysaccharide was mixed with Congo red reagent, and NaOH was added to adjust the final concentration to 0, 0.1, 0.2, 0.3, 0.4, 0.5, and 0.6 M, respectively. The maximum absorption wavelength (λmax) of Congo red-polysaccharide complexes at various concentrations of NaOH was observed.

2.8 Scanning electron microscope (SEM) analysis

The dried AQP70-2A sample (1 mg) was coated with a conductive layer of gold–palladium and applied to FE-SEM (JOEL, JSM-7800F, Tokyo, Japan) at an accelerating voltage of 15.0 kV.

2.9 Preparation of selenium-modified polysaccharide

The dried polysaccharide (20 mg) was dissolved in 5 mL HNO₃ (0.6 %, v/v) and stirred at room temperature for 30 min. Then Na₂SeO₃ and BaCl₂ were added to the mixture, and the reaction lasted for 8 h with stirring under 65 ℃. The pH value of the reaction mixture was adjusted to 7–8 with NaOH, and the Ba²⁺ was removed with sodium sulfate. After dialysis, the reaction product was precipitated with ethanol, and the selenium-modified polysaccharide named Se-AQP70-2A was recovered by freeze-drying. Chemical characterization of Se-AQP70-2A was performed using the methods consistent with natural polysaccharides. Meanwhile, SEM equipped with energy dispersive spectrometer (EDS) was used to semi-quantitatively measure Se element content of Se-AQP70-2A. Accurate quantification of Se element was performed by applying the samples to the inductively couple plasma mass spectrometry (ICP-MS) platform.

2.10 Effects of polysaccharides on HepG2 cells viability

Human hepatocellular carcinoma cell line HepG2 was selected as a model and MTT assay was used to evaluate the antitumor activity of polysaccharides in vitro. The detailed procedure of cell viability assessment is summarized in the Supplementary information.

2.11 Evaluation of antitumor activity in vivo

Zebrafish, a vertebrate model, has been widely used in pharmacological and toxicological studies due to its high homology with human and its transparent body at the early developmental stages, which is easy to observe the changes of tissue structure (Gu et al., 2021; Gu et al., 2022). According to the reported methods (Liang et al., 2020), the in vivo activity of polysaccharides against tumor cell invasion and metastasis was evaluated in a zebrafish tumor xenograft model. Meanwhile, the anti-angiogenic activity of Se-AQP70-2A was evaluated using the transgenic zebrafish Tg (fli1: EGFP) and compared with the positive control treated with 1 μg/mL sunitinib malate (Wang et al., 2021). The detailed procedure of in vivo experimental are presented in the Supplementary information.

2.12 Statistical analysis

The experimental data were statistically analyzed using GraphPad Prism 7.0 software and presented as mean ± standard deviation. Significant differences between different groups were analyzed by one-way ANOVA followed by Turkey's multiple range tests.

3.1 Isolation and purification of AQP70-2A

The percentage yield of crude polysaccharide (AQ70) separated by alcohol precipitation was 1.3 % (polysaccharides weight/ dried fruits weight). The elution profiles of AQ70 was shown in Fig S1A. There were two elution bands within the elution range of 0.05 M NaCl, and the first peak was collected with the yield of 0.2 % (polysaccharides weight/ dried fruits weight) and named AQP70-2. Anion-exchange chromatographic elution products were further purified, and AQP70-2 was divided into two components named AQP70-2A and AQP70-2B by Sephadex G-75 column (Fig. S1B). The percentage yield of the two fractions AQP70-2A and AQP70-2B purified by molecular sieve both amounted to 0.02 % (polysaccharides weight/ dried fruits weight).

Close

Fig. 1

NMR analysis of AQP70-2A. (A) The ¹H NMR spectrum of AQP70-2A. (B) The ¹³C NMR spectrum of AQP70-2A. (C) The HSQC spectrum of AQP70-2A. (D) The HMBC spectrum of AQP70-2A. (E) The DEPT-135 NMR spectrum of AQP70-2A.

Export to PPT

3.2 Homogeneity and molecular weight of AQP70-2A

The UV–vis spectrum showed that there were no characteristic absorption peaks at 260 nm and 280 nm (Fig. S2B), indicating no proteins and nucleic acids present in AQP70-2A. Moreover, AQP70-2A was eluted as a single peak in HPGPC (Fig. S2A), suggesting it was a homogenous polysaccharide with a narrow molecular weight distribution. In addition, the sugar content test showed that the carbohydrate content in AQP70-2A was 95.41 %. The calibration curve was obtained by plotting the retention time of the serial Dextran standards versus the logarithm of the molecular weight (Fig. S3; Table S1). The average molecular weight of AQP70-2A was calculated to be 1.49 × 10⁴ Da according to the standard calibration curve. Furthermore, HPSEC-MALLS-RI system was used to determine the molecular parameters of polysaccharide (Fig. S4; Table 1). The number-average molecular weight (M_n), weight-average molecular weight (M_w) and z-average molecular weight (M_z) of AQP70-2A were calculated to be 1.49 × 10⁴, 1.53 × 10⁴, and 1.56 × 10⁴ g/mol, respectively. The M_w/M_n and M_z/M_n as polydispersity indexes for evaluating the molecular weight distribution width of AQP70-2A were 1.023 and 1.047, respectively, which suggested the molecular weight distribution of AQP70-2A was narrow. The data were consistent with the results of HPGPC analysis, confirming that AQP70-2A was a homogeneous polysaccharide.

Close

Fig. 2

(A) The putative structure of AQP70-2A. (B) The possible stretching state of AQP70-2A in 0.1 M NaCl solution.

Export to PPT

3.3 Monosaccharide composition of AQP70-2A

Compared with the retention time of the standard monosaccharide, the PMP pre-derivatization method showed that the AQP70-2A was composed only of arabinose (Fig. S5). AQP70-2A was a homopolysaccharide different from AQP70-2B, which was also eluted by 0.05 M NaCl, and AQP70-2B was a heteropolysaccharide composed of rhamnose, glucose, galactose and arabinose (Wang et al., 2022).

3.4 Glycosidic linkages analysis of AQP70-2A

The disappearance of the O—H band (3100–3700 cm⁻¹) in the IR spectrum of the methylated AQP70-2A indicated that hydroxyl (−OH) was replaced with methoxy groups (−OCH₃), which confirmed that the methylation reaction of the polysaccharides was complete (Fig. S6). By analyzing mass spectrometry of PMAAs (Fig. S7), the linkage patterns of monosaccharide residues were shown in Table S2. Methylation analysis showed that the AQP70-2A was composed of 1-substituted, 1,3-disubstituted, 1,5-disubstituted, 1,2,5-trisubstituted, 1,3,5-trisubstituted, and 1,2,3,5-tetrasubstituted-l-arabinose residues. The detection of tetra-substituted monosaccharide residues suggested that AQP70-2A was a novel arabinan different from that previously reported (Wang et al., 2021).

3.5 NMR spectroscopy analysis of AQP70-2A

The primary structure of AQP70-2A was further derived comprehensively by NMR spectra. From the ¹H NMR spectrum (Fig. 2A), six anomeric proton signals were shown clearly at 5.00, 5.03, 5.06, 5.09, 5.12, and 5.14 ppm. The ratio of the integral areas of these anomeric proton signals was appropriately calculated as 5.80: 1.40: 1.27: 2.44: 1.91: 1, respectively. In the spectral region 3.60–4.30 ppm, the signals were non-anomeric protons of sugar ring (H-2 − H-5). In the anomeric carbon region (90–110 ppm), the ¹³C NMR spectrum of AQP70-2A exhibited seven anomeric carbon signals (Fig. 2B). Among these resonance peaks of anomeric carbon, the signals at 107.4, 107.2, 106.9, 106.8, and 106.3 ppm could be assigned to five different α-l-arabinose residues, while signals at 101.6 and 101.3 ppm were attributed to β-l-arabinose residues (Cardoso et al., 2002; Dourado et al., 2006). According to the literature, methylene carbon atom (C-5) of arabinose contributed to the six inverted carbon signals (61.0, 63.0, 66.1, 66.3, 66.8, and 66.9 ppm) in DEPT-135 spectrum (Fig. 2E) (Das et al., 2013).

The ¹H and ¹³C chemical shift assignments of AQP70-2A shown in Table 2 were determined according to the spectral information in the 1D and 2D NMR experiments and by comparison with the NMR data from arabinans or arabinan rich-polysaccharides (Zhang et al., 2020a; Zhang et al., 2020b; Cardoso et al., 2002; Dourado et al., 2006; Shakhmatov et al., 2014; Fernandes et al., 2019; Rodrigues et al., 2019; Shakhmatov et al., 2015). Based on the literature (Cardoso et al., 2002; Dourado et al., 2006), the H-1/C-1 cross peaks at 5.00/101.6 and 5.03/101.3 ppm in HSQC spectrum (Fig. 2C) indicated that the polysaccharides contained two types of t-β-l-Araf in different chemical environments, named residues A and B, respectively. Combining with the ¹H, ¹³C, and HSQC spectra, the strongest signal at 5.00/107.4 ppm was assigned to → 5)-α-l-Araf-(1→ (residue F) due to its high molar ratio in methylation analysis. Meanwhile, the correlation in ¹H–¹H COSY spectrum (Fig. S8) allowed the assignments of non-anomeric chemical shifts of residue F (H-2 at δ 5.00, H-3 at δ 4.18, H-4 at δ 3.95, H-5/5′ at δ 3.84/3.72), and these chemical shift characteristics were found to be close to those observed in the literature (Zhang et al., 2020a; Shakhmatov et al., 2014). Moreover, in the HSQC spectrum, the correlations at 5.09/106.8, 5.06/107.2, 5.12/106.9 and 5.14/106.3 ppm were also be attributed to α-l-Araf (Zhang et al., 2020b; Fernandes et al., 2019), and these residues were designated as D, E, G and I, respectively. The other carbon and proton signals in the residues were assigned according to ¹H–¹H COSY and HSQC spectra. The downfield shift of C-3 confirmed the presence of the 3-O substituted (82.8 ppm), which indicated that the residue D was → 3)-α-l-Araf-(1→ (Rodrigues et al., 2019). Compared with 1,5-disubstituted-l-arabinose residue, the HSQC spectrum showed a downfield shift for C-3 (83.9 ppm) in residue E, and a downfield shift for C-2 (84.8 ppm) in residue G, suggesting that the residues E and G were 1,3,5-trisubstituted and 1,2,5-trisubstituted α-l-arabinofuranose, respectively. Similarly, the significant glycosylation shifts of C-2, C-3 and C-5 in residue I indicated that the unit was 1,2,3,5-tetrasubstituted-l-arabinose residue (Shakhmatov et al., 2015).

The HMBC spectrum was applied to further confirm the sequence of arabinose residues in AQP70-2A (Fig. 2D). In detail, the cross peak of δ_H 5.00 (residue F, H-1) and δ_C 66.3 (residue F, C-5) in the HMBC spectrum, indicated that the AQP70-2A contained two residues F connected through 1,5 linkages. Moreover, HMBC correlations of δ_H/C 3.84/107.2 (F H-5/E C-1), 3.79/107.4 (E H-5/F C-1), 5.14/66.3 (I H-1/F C-5), 5.09/66.9 (D H-1/G C-5), and 3.76/107.4 (I H-5/F C-1) supported that the O-5 of residue F was linked to C-1 of residue E, the O-5 of residue E was linked to C-1 of residue F, the O-1 of residue I was linked to C-5 of residue F, the O-1 of residue D was linked to C-5 of residue G, and the O-5 of residue I was linked to C-1 of residue F. Similarly, the bonding information of polysaccharide branch chain residues can be characterized as follows: C-3 signal of → 3,5)-α-l-Araf-(1 → correlated with H-1 signal of → 2,5)-α-l-Araf-(1→, C-2 signal of → 2,5)-α-l-Araf-(1 → correlated with H-1 signal of β-l-Araf-(1→ (residue B), C-5 signal of → 2,5)-α-l-Araf-(1 → correlated with H-1 signal of → 3)-α-l-Araf-(1→, C-2 signal of → 2,3,5)-α-l-Araf-(1 → correlated with H-1 signal of → 3)-α-l-Araf-(1→, and H-3 signal of → 2,3,5)-α-l-Araf-(1 → correlated with C-1 signal of → 2,5)-α-l-Araf-(1 →. Further, residue A linked to the end of the branched chain could be verified by the cross peaks of δ_H 5.00 (A H-1) and δ_C 82.8 (D C-3). The above signal attribution matched with the characteristic signals of several arabinose residues reported in the literature (Makarova and Shakhmatov 2020; Zhang et al., 2019). Based on all the structural evidence, AQP70-2A was an arabinan composed of seven kinds of sugar residues, and its putative structure of AQP70-2A was displayed in Fig. 2.

3.6 Congo red test of AQP70-2A

Compared with those of Congo red, the maximum absorption wavelengths of AQP70-2A plus Congo red did not change significantly with the change of NaOH concentration, which indicated that AQP70-2A was absent of triple-helix structure (Fig. S9).

3.7 Conformation analysis

The slope of the fitting curve in the conformation plot of AQP70-2A was calculated to be 0.38 (Fig. S4C), indicating that the branching polymer presented a compact coil chain conformation (Wang et al., 2022). The possible stretching state of the AQP70-2A simulated by employed Chem 3D was shown in Fig. 2B, which was consistent with the conformation analysis.

3.8 FT-IR analysis of polysaccharides

The FT-IR infrared spectrum of AQP70-2A was shown in Fig. S10A. Specifically, the absorption bands at 3400 cm⁻¹ and 2932 cm⁻¹ correspond to stretching vibration of O—H and C—H, respectively (Xiang et al., 2019). The absorption bands at 1647 cm⁻¹ can be attributed to the O—H bending vibration. Moreover, the bands in the region of 1419 cm⁻¹, 1363 cm⁻¹, and 1220 cm⁻¹ were assigned to C—H bending vibration (Hu et al., 2019; Li et al., 2021). According to the literature (Zheng et al., 2020), the typical absorption bands between 1200 and 1000 cm⁻¹ were dominated by the sugar ring vibrations and the overlapped stretching vibrations of C−OH side groups and the C—O —C glycosidic bonds. As reported in the literature, the absorption band at 807 cm⁻¹ was caused by furanose rings, suggesting that the AQP70-2A contained furan sugar residues (Jahanbin et al., 2017).

Se-AQP70-2A and AQP70-2A shared the same main characteristic bands, meaning that the basic skeleton remained unchanged (Fig. S10B). Se-AQP70-2A showed two absorption bands at 860 cm⁻¹ and 805 cm⁻¹, corresponding to the stretching vibrations of Se = O and C—O−Se. In accordance with the previous experimental findings, these absorption bands confirmed that selenylation occurred in AQP70-2A (Wang et al., 2022; Zhu et al., 2016; Wei et al., 2015).

3.9 SEM-EDS analysis of polysaccharides

Surface morphology of AQP70-2A was observed with SEM (Fig. S11). As shown in the 957 × image, AQP70-2A had rod-like morphological features. Further SEM image at higher magnifications (2000 × ) clearly showed that the surface of AQP70-2A presented a porous structure with a hollow interior similar to that of a honeycomb.

SEM images of Se-AQP70-2A provided surface morphological information (Fig. 3A). In contrast to the morphology of native polysaccharide (Fig. S11), Se-AQP70-2A exhibited a relatively continuous surface with local undulation and folds, which was much smoother than that before selenylation modification. These changes suggested that the presence of selenium significantly changed the rough and irregular surface characteristics of the polymer, as reported by Ru et al. (2020).

Close

Fig. 3

SEM-EDS analysis of Se-AQP70-2A. (A and B) SEM images and element distribution of Se-AQP70-2A. (C) Surface elemental composition of Se-AQP70-2A. (D−F) Carbon, oxygen, and selenium element distribution of Se-AQP70-2A.

Export to PPT

Close

Fig. 4

Antineoplastic effects of Se-AQP70-2A cells in zebrafish xenograft. After transplanting CM-DiI stained HepG2 cells into 2 dpf zebrafish embryos via microinjection, tumor-bearing embryos were treated with Se-AQP70-2A (50, 100, and 200 μg/mL). (A) Intensity and distribution of the red fluorescence were observed under a confocal microscope. (B) Fluorescence intensity of the tumor xenografts, representing the number of HepG2 cells. (C) Quantification of the fluorescent area of the tumor xenografts, representing HepG2 cells metastasis. The data expressed as mean ± SD (n = 15/group). *p < 0.05, **p < 0.01, and ***p < 0.001 versus control group.

Export to PPT

Energy dispersive spectrometer (EDS) images indicated a homogeneous distribution of C, O, and Se elements in AQP70-2A (Fig. 3B, D − F), and the contents of these three elements were calculated to be 59.8 %, 39.9 %, and 0.3 % (Fig. 3C), respectively. While, the ICP-MS test was performed to reveal that accurate content of Se element in Se-AQP70-2A was 0.92 %. The content data were consistent with those reported in the literature, which were between 0.11 and 0.91 % in selenylation polysaccharide samples prepared by the method Na₂SeO₃-HNO₃ (Cheng et al., 2018).

3.10 Inhibitory effect of AQP70-2A and Se-AQP70-2A on HepG2 cell viability

MTT assay results indicated that in the tested concentration range (25–400 μg/mL), the natural polysaccharide AQP70-2A had weak effects on the viability of cancer cells in vitro (Fig. S12). After selenylation, the inhibitory activity of polysaccharides on cancer cells was enhanced in a dose-dependent manner, especially in the concentration range of 100–400 μg/mL, the inhibition rate was significantly up-regulated, which was 7 to 10 times than that of AQP70-2A. The half maximal inhibitory concentration (IC₅₀) of Se-AQP70-2A on HepG2 viability was calculated to be 191.4 μg/mL.

3.11 anti-metastasis and anti-angiogenesis effect of Se-AQP70-2A in vivo

Red fluorescent-labeled cancer cells and disseminated foci from tumor mass in the xenograft model were observed under confocal microscopy (Fig. 4). With Se-AQP70-2A administration, the total fluorescence intensity and the number of disseminated foci in vivo model decreased significantly in a dose dependent manner. These two indices were down-regulated by 44.4 % and 30.7 % in 200 μg/mL Se-AQP70-2A-treated group, respectively, which was almost consistent with the anti-tumor invasion and metastasis effects of lentinan (LNT) (200 μg/mL) used as a positive control. These results implied that Se-AQP70-2A played an anticancer role by effectively blocking tumor invasion and metastasis.

Furthermore, the antiangiogenic effects of Se-AQP70-2A were evaluated by the development of intersegmental vessels (ISVs) in transgenic zebrafish model Tg(fli1: EGFP) (Fig. 5). Se-AQP70-2A treatment significantly inhibited the growth of ISVs, in which the ISVs total lengths of zebrafish in the 25, 50, and 100 μg/mL concentration groups were approximately 9.26 %, 31.5 %, and 33.3 % shorter than those of the control group, respectively. The antiangiogenic activity of 100 μg/mL Se-AQP70-2A could reach 85.6 % of sunitinib malate as the positive control.

Close

Fig. 5

Antiangiogenic activity of Se-AQP70-2A. The anti-angiogenic activity of Se-AQP70-2A (25, 50, and 100 μg/mL) was evaluated using the transgenic zebrafish Tg (fli1: EGFP) and compared with positive control treated with 1 μg/mL sunitinib malate. (A) The development of intersegmental vessels (ISVs) and dorsal longitudinal anastomotic vessels (DLAVs) in different groups. (B) The average length of ISVs of zebrafish after treating with different concentrations of Se-AQP70-2A. The data expressed as mean ± SD (n = 15/group). *p < 0.05, ***p < 0.001 versus control group.

Export to PPT