1,3-and 1,4-linked polysaccharides uptake in intestinal cells relies on clathrin/dynamin 1/Rab5-dependent endocytosis

Abstract

It is thought that polysaccharides cannot penetrate the intestinal mucosa into the circulatory system due to their high hydrophilicity and large molecular size. However, we show that different linked and charged polysaccharides can penetrate Caco-2 cell monolayers, and find β−1,3-linked glucan and α−1,4-linked glucan is detectable in rats (male) and mice plasma and liver after oral administration the isotope (^99mTc, ³H) and fluorescein labeled polysaccharides. Using gene-knockdown strategies and inhibitors, we further show polysaccharide uptake requires clathrin heavy chain (CLTC) and its associated factors Rab5 and dynamin1 in intestinal cells. Strikingly, polysaccharide absorption is attenuated in both CLTC intestine deficient mice and RAB5A, DNM1conventional-knockout mice. Importantly, membrane receptors bone morphogenetic protein receptor type IA (BMPRIA), Dectin-1 and epidermal growth factor receptor (EGFR) are also critical for specific structural polysaccharide internalization. These findings provide novel insight to understand polysaccharide absorption mechanism and lay foundation for oral polysaccharide-based new drugs development.

The study demonstrated that some exogenous natural polysaccharides could be absorbed in plasma with little degradation after orally administration. Polysaccharide uptake in the intestinal cells relied on clathrin/dynamin 1/Rab5-dependent endocytosis.

Introduction

As one of the four fundamental biomolecules in nature, carbohydrates mediate various biological processes such as signal recognition, pathogenesis prevention, immune modulation, and development¹. Carbohydrate can be found as simple monosaccharide, oligosaccharide, and polysaccharide or forming more complex glycoconjugates, such structural variability makes carbohydrate hot spots for biomedical intervention^2,3. Early in 1940, carbohydrate-based antibiotics, including streptomycin and gentamicin (both are aminoglycosides), were discovered as anti-infection drugs. Then, the ganglioside GM1 was developed for acute stroke treatment, hyaluronic acid and chondroitin sulfate (both are polysaccharides) were investigated as anti-arthritis drug⁴. Since 2000, tremendous progresses in glycoscience fields bring vast opportunities for novel carbohydrate-based drug discovery⁵. In 2019, the conditional launch of sodium oligomannate (GV-971) in China further dramatically boosts the innovative carbohydrate drug development to a new climax⁶. Surprisingly, polysaccharides represent the largest percentage of the recorded carbohydrate drugs⁷. To date, more than 40 commercial polysaccharides-based drugs have been launched including heparin, which is a high sulfated heterogenous glycosaminoglycan, was approved as an anticoagulant drug for more than 80 years⁸. Nevertheless, the biological roles of exogenous polysaccharides go further beyond due to their extensive therapeutic properties in antibiotic, antioxidant, antitumor, anticoagulant, antiaging, and immuno-stimulation fields, etc⁹. Notably, about 17 approved or clinically developed polysaccharides-based drugs are administrated orally according to the records in Cortellis^TM (Drug Discovery Intelligence) database⁸. However, it’s largely unknown whether these bioactive polysaccharides can reach the lesion site after oral administration. Thus, firstly, we are not sure whether compound degradation occurs after intestinal uptake; Secondly, may those polysaccharides enter blood circulation system? More critically, studies on the molecular mechanism of polysaccharide absorption have been limited to date. The main reason is that most researchers believe polysaccharides cannot be absorbed by the gastrointestinal tract due to their large molecular size and high hydrophilicity. However, clinical studies demonstrated the absorption of exogenous high molecular mass chondroitin sulfate in the intestinal lining¹⁰. Nevertheless, it is challenge for clinically used heparin to pass through the gastrointestinal epithelium¹¹. Hence, polysaccharides absorption by the intestinal lining is still a controversial topic.

Clearly, there is a critical need to understand polysaccharide pharmacokinetics to further fully utilize bioactive polysaccharides and develop novel polysaccharide-based drugs. However, owing to their large molecular weight and deficiency of UV-absorption groups, it is challenging to detect plasma polysaccharides, which may be not distinguishable when the exogenous and endogenous polysaccharide consisting of same or similar monosaccharides. During the few last decades, many attempts have been made to address such challenge. The prior method is radioactive isotope labeling (such as ¹⁴C, ¹²⁵I, ³⁵S, ³H and ^99mTc), which were widely used in the pharmacokinetics investigations of polysaccharide^12–17. Another common method used in pharmacokinetic studies is liquid chromatography coupled with tandem mass spectrometry (LC-MS/MS). However, the current chromatographic and mass spectrometric techniques have limited application due to large molecular size and lacking of chromophoric groups of polysaccharides. Luckily, there has been some progress with the development of spectrometric techniques, such as high-performance liquid chromatography (HPLC) followed by post-column fluorescence derivatization or pre-column fluorescein labeled^18,19.

Except studying the ability of polysaccharides to permeate the intestinal mucosa, it is critical to study the absorption mechanism. Several pathways were reported involving in extracellular macromolecules uptake, such as macropinocytosis^20,21, phagocytosis²² and endocytosis²³. Endocytosis, the process by which cells internalize substances via the formation of vesicles from the plasma membrane, is critical for antigen presentation and cell migration, and also regulates many intracellular signaling cascades²⁴. Eukaryotic cells utilize two major endocytic pathways: clathrin-dependent and clathrin-independent. The mechanism of clathrin-mediated endocytosis (CME) has been extensively characterized^25,26. In contrast, clathrin-independent pathways comprise a diverse group of processes, including caveolin- and flotillin-dependent endocytosis, as well as phagocytosis^27–29. Additionally, CME is required for the uptake of large molecules such as proteins, pathogenic bacteria, and viruses in mammalian cells^30,31. Building upon these findings, we sought to determine the endocytic mechanism responsible for intestinal polysaccharide uptake, specifically whether it is primarily mediated by clathrin or occurs via an alternative pathway.

In this study, we explored  the ability of various polysaccharides to permeate intestinal cell layers in vitro. Moreover, GFPBW1 (β-1,3-linked glucan from Grifola frodosa³²) and WGE (α-1,4-linked glucan from Gastrodia elata Bl³³), as two model polysaccharides (two main types of polysaccharides in plant, fungus, and herbal medicine), were employed for the in vivo and mechanism study.

Results

Polysaccharides with different linkage types and charge could be absorbed and internalized into intestinal cells in vitro

Caco-2 cell monolayer is widely used as an in vitro model that mimics human small intestinal mucosa to predict orally administered drugs absorption³⁴. To test the potential whether polysaccharide can be absorbed by intestine epithelial cells, transport studies for GFPBW1 and WGE were firstly performed. The results showed that the retention times of the transported GFPBW1 and WGE were not changed significantly compared with the prototypical polysaccharides in the HPLC chromatograms (Fig. 1a), indicating the molecular sizes of the penetrated polysaccharides did not change much, further suggesting that polysaccharides might be transported by Caco-2 cell monolayer without significant degradation. This result surprised and encouraged us to ask whether this phenomenon occurs widely and to further test the absorption situation of other linkage types and different charged polysaccharides. Hence, the transport experiments of lentinan (clinically used branched 1,3-linked glucan), MDG-1 (1,2-linked fructan), dextran (commercially available 1,6-linked glucan), GFPBW1-NH₃ (amino-branched 1,3-linked glucan), WGE-NH₃ (amino-branched 1,4-linked glucan), GFPBW1-SO₄ (sulfated-branched 1,3-linked glucan), and WGE-SO₄ (sulfated-branched 1,4-linked glucan) were then conducted. Surprisingly, the results showed that diversely linked and differently charged polysaccharides could penetrate Caco-2 cell monolayers. In terms of the penetrated amounts, the value for GFPBW1 was close to that for dextran but significantly greater than that for WGE. Moreover, the penetrated amount of MDG-1 (with the smallest molecule size) was the largest, and was similar as that for lentinan (crude polysaccharide). Furthermore, the charged polysaccharides also displayed variation in the penetration values. The amount of GFPBW1-NH₃ was dramatically more than that of GFPBW1 and GFPBW1-SO₄. Oppositely, the amount of both WGE-NH₃ and WGE-SO₄ was higher than that of WGE (Fig. 1b). Surprisingly, the apparent permeability coefficient (P_app) values of the tested polysaccharides were more than 1 × 10⁻⁶ cm/s, indicating that they might be absorbed by the cells (Supplementary Table 1).

Fig. 1. Different linked and charged polysaccharides could permeate Caco-2 cell monolayer and uptake by intestine cells.

a The transport experiments of GFPBW1 and WGE were performed as described in Methods. The chromatograms of the prototypical polysaccharides (arrowheads indicated) and samples (collected at 8 h) are shown. b The Q-T curves of the polysaccharides with different linkages such as GFPBW1, WGE, dextran, MDG-1, lentinan; GFPBW1 and its derivatives (GFPBW1-NH₃, GFPBW1-SO₄); WGE and its derivatives (WGE-NH₃ and WGE-SO₄) are exhibited in the left, middle and right panel, respectively. c Representative images of internalized polysaccharides in T24 cells. Scale bars, 20 μm. d–g HIEC-6 cells were treated with fluorescein labeled GFPBW1 and WGE for the indicated time intervals (d) or at different concentrations (f), then cells were fixed and photographed by microscopy, mean fluorescence intensity was measured for all the fields of view for each group at three independent experiment and showed in (e) and (g), respectively. Similarly, images of the uptake experiments for GFPBW1 and WGE in Caco-2 cells in various time (h) or at different concentrations (j) are displayed, while (i) and (k) represent the quantified fluorescence intensity of (h) and (j). l IEC-6 cells were treated with GFPBW1 and WGE at various time points and examined by confocal microscopy, measured fluorescence intensity is shown in (m), scare bar, 10 μm. Data are presented as mean values ± SEM from three biologically independent experiments. Source data are provided as a Source Data file.

To trace polysaccharide transport, we fluorescein-labeled the polysaccharides³⁵, the labeling positions were verified using raffinose, mannose via NMR (Supplementary Fig. 1a–c), and confirmed that the labeling procedure did not affect the structure (Supplementary Fig. 1d, e) and bioactivity (Supplementary Fig. 1f) of the saccharides referring to our previous report³². Next, the uptake experiment of different linkages and charged polysaccharides was performed. The results clearly showed that all the tested polysaccharides could be internalized into the cells (Fig. 1c), while uptake of the fluorescein alone did not happen in the cells (Supplementary Fig. 1g, h). We then asked whether time and/or concentration affected polysaccharides uptake. To better understand the details of polysaccharide penetration, GFPBW1 and WGE were employed as the model polysaccharides to do the following exploration. To address the question, human intestinal epithelial cell line (HIEC-6) was treated with GFPBW1 or WGE at different concentrations or for various time points. The results revealed that both GFPBW1 and WGE could be internalized into the cell as early as 5 min, both polysaccharides showed a sharp increase internalization at 10 min, then the internalized polysaccharide decreased at 15 and 30 min. However, the uptake of GFPBW1 had recovered by 1 h (Fig. 1d, e). Moreover, distinct fluorescence could be observed at 1.25 mg/mL for GFPBW1 and 2.5 mg/mL for WGE, respectively (Fig. 1f). The uptake of both polysaccharides exhibited a concentration-dependent manner (Fig. 1g). Subsequently, similar experiment was designed using Caco-2 cells. Similarly, GFPBW1 and WGE showed analogous uptake time dynamics in this cell line (Fig. 1h, i), and both demonstrated a concentration dependent uptake manner (Fig. 1j, k). Furthermore, polysaccharides uptake time dynamics were examined in IEC-6 (rat intestinal cell line), WGE showed a time-dependent pattern. Conversely, GFPBW1 showed a decrease at 15 min and complete recovery at 30 min (Fig. 1l, m). Thus, the uptake dynamics of GPFBW1 and WGE in rat and human cell lines were different. The reason of such difference between the two polysaccharides might be partially due to their structures.

Polysaccharides were absorbed in rat and mice intestinal lining in vivo and could be detected in rodent blood and liver

Next, we sought to ascertain whether the two model polysaccharides could be absorbed in vivo. GFPBW1 and WGE were orally and intravenously administered to rats, while commercially available lentinan was also included as a control. Firstly, the spectrum of blank rat plasma mixed with GFPW (internal standard) or without followed extraction as described in “Methods” were displayed (Supplementary Fig. 2a, b). The HPLC chromatograms showed the retention time and peak pattern of lentinan before and after the pretreatment procedure did not change obviously, indicating good efficiency of the extraction process (Supplementary Fig. 2c). Surprisingly, GFPBW1 was quantifiable in all plasma samples following oral and intravenous administration (Fig. 2a, upper panel). Similar results were obtained for WGE and lentinan, and there were trace changes between the retention time and peak pattern of the plasma polysaccharides and their prototypical ones (Fig. 2a and Supplementary Fig. 2d, e), suggesting that the tested polysaccharides were absorbed at least partially as macromolecular form and underwent minimal degradation in vivo. The pharmacokinetic profiles and parameters of GFPBW1, WGE, and lentinan by oral administration or intravenous injection are shown in Supplementary Fig. 2f–h and Supplementary Tables 2 and 3, respectively. Yet, there were a few differences among them regarding to their pharmacokinetic characteristics. The maximum concentration of GFPBW1 in plasma (T_max) was at 5 h, while the half time (t_1/2) was 1.44 h. However, the clearance rate (CL) was 74.84 mL kg⁻¹ h⁻¹, suggesting that GFPBW1 was absorbed slowly but cleared quickly. The T_max of WGE occurred at 2 h, the t_1/2 was 8.44 h, and the CL was 1.879 mL kg⁻¹ h⁻¹, indicating rapid absorption and slow clearance (Supplementary Table 2).

Fig. 2. Polysaccharides could be absorbed by the rat and mice intestine in vivo and could be detected in rat and mice plasma and liver.

a The chromatograms of GFPBW1 and WGE before and after oral administration or intravenous injection in rat are shown. The solid and thin lines represent the prototypical or plasma polysaccharides, respectively. b The HPLC chromatograms of mice plasma after oral administration of fluorescein labeled WGE are shown, and the chromatogram of fluorescein-WGE spiked into blank C57BL/6J mice plasma is included as embedded panel. Moreover, the Radio HPGPC chromatograms of plasma (left panel) and liver lysate (right panel) after oral administration (c) or intravenous injection (d) of ^99mTc-GFPBW1 to C57BL/6J mice for the indicated time intervals were exhibited, the embedded photograph was the Radio HPGPC chromatograms of the original ^99mTc-GFPBW1. Meanwhile, C57BL/6J mice were also oral administrated or intravenous injection with ^99mTc-WGE, and the Radio HPGPC chromatograms of plasma (left panel) and liver lysate (right panel) were shown in e, f, the chromatogram of original radio labeled WGE was exhibited in the embedded picture. Images of isolated organs (upper panels) and the whole intestine (lower panels) after the BALB/c mice were oral administrated (g, i) or intravenous injection (h, j) of Cy5.5 labeled polysaccharides (g, h for GFPBW1, i, j for WGE) for the indicated time intervals were presented. k Frozen section images of the intestinal tissues were shown after BALB/c mice were orally administered with fluorescein labeled GFPBW1 and WGE for the indicated time points. Scale bars, 100 μm. l Similarly, images of duodenum, jejunum, and ileum of C57BL/6J mice after orally administrated with fluorescein labeled WGE for 1 h or 2 h were presented, white arrowheads indicate the internalized polysaccharides. Scale bar, 100 μm. The mean fluorescence intensity of each group was quantified in (m). Data are presented as mean ± SEM from three biologically independent experiments. Source data are provided as a Source Data file.

To further validate these findings, fluorescein-labeled GFPBW1 and WGE were labeled and detected on HPLC with a fluorescence detector. The results showed that both GFPBW1 and WGE demonstrated a typical peak pattern with measurable signals (Supplementary Fig. 2i). Next, C57BL/6J mice were orally administrated by fluorescein labeled WGE, followed by the polysaccharide extraction in mice serum and analysis. HPLC detection of plasma samples collected 1 h after oral administration showed a typical peak for WGE, with a retention time very close to that of the original fluorescein labeled WGE, indicating that WGE was absorbed in its macromolecular form (Fig. 2b).

To more precisely detect the polysaccharide after the oral administration in mice, we next sought to label the polysaccharides by radioactive ³H with the help of QZ Isotech (Shanghai, China) and examined by HPLC connected with β ray detector. Surprisingly, chromatograms showed a typical peak with quantitative signals (Supplementary Fig. 2j). Subsequently, C57BL/6J mice were orally administrated by ³H-GFPBW1 and WGE, and the concentration-time curves of GFPBW1 and WGE in mice serum are displayed in Supplementary Fig. 2k, further confirmed the results we obtained from rats and fluorescein labeled polysaccharides in mice. Further, we were still surprised to observe a peak with quantitative signals in mice plasma after oral administration of both ^99mTc labeled GFPBW1 and WGE (Fig. 2c, e, left panels). More importantly, we even obtained canonical peak in mice liver lysate after oral administration of both GFPBW1 and WGE (Fig. 2c, e, right panels), and the retention time of liver lysate was closed to the original labeled ones, further implied that polysaccharides were absorbed partially in the macromolecular form without significantly degradation (similar molecular weights) within the test time and could accumulate in the liver. Moreover, we obtained similar results from the plasma and liver lysate of mice after intravenously injected ^99mTc labeled GFPBW1 (Fig. 2d) and WGE (Fig. 2f). The pharmacokinetic parameters of ^99mTc- GFPBW1 and WGE in mice were summarized in Supplementary Tables 4 and 5, respectively. Furthermore, fluorescence labeled GFPBW1 (Fig. 2g, h) and WGE (Fig. 2i, j) with Cy5.5 were employed to trace the polysaccharides in vivo using living imaging after orally administration (Fig. 2g, i) or intravenous injection (Fig. 2h, j). Obvious fluorescence was observed in both mice liver and intestine, further indicating the absorption of both polysaccharides.

Next, to investigate when the absorption of the polysaccharides commences in mice intestine, BALB/c mice were orally administered with fluorescein-labeled GFPBW1 and WGE, and we indeed visualized the internalized GFPBW1 and WGE in mice intestine after the oral administration for 20 and 15 min, respectively. Moreover, the maximum internalization of GFPBW1 and WGE occurred at 80 and 30 min after oral dosing, respectively (Fig. 2k). To rule out the possible interference caused by mouse strain, we performed similar experiment in C57BL/6 J mice. Microscopic analysis showed that WGE was internalized mostly in the intestinal enterocytes of the duodenum, while less was internalized in the jejunum and only a trace was found in the ileum in WGE-1 h group. Conversely, we observed different absorption pattern in WGE-2 h group. The internalized WGE was increased sequentially in duodenum, jejunum, and ileum segments, and even could be observed in mice ileum crypt (Fig. 2l, m). Collectively, we demonstrated that polysaccharides could be absorbed by intestinal cells in vivo and detectable in rodent blood and liver.

Clathrin mediated endocytosis is critical for polysaccharides uptake in cultured cells

Some macromolecules, such as EGFR and asialoglycoproteins, were reported to be transported into the cells via endocytosis^36,37. Inspired by this, we speculated that polysaccharides, which are hydrophilic macromolecules, might be absorbed via vesicular endocytosis. Firstly, we confirmed the localization of internalized polysaccharides using DiI to label cell membrane, the internalized polysaccharides were observed mostly in cytoplasm and partially on cell membrane (Fig. 3a).

Fig. 3. Disruption of Clathrin-mediated endocytosis attenuated the uptake of polysaccharides.

a HIEC-6 cells were incubated with GFPBW1 or WGE for the indicated time, then cells were fixed and stained with DiI (red). Subsequently, HIEC-6 cells were pretreated with different inhibitors such as CPZ, Nystatin, Cyto D, or methyl-β-cyclodextrin, and the uptake experiments of GFPBW1 and WGE were then performed; the images are displayed in b, d, f, h, separately. Scale bars, 10 μm. The mean fluorescence intensity was quantified in c, e, g, and i, respectively. Data are presented as mean ± SEM from three biologically independent experiments. Unpaired two-tailed t test, *p < 0.05, **p < 0.01, ***p < 0.001. j–l Furthermore, HIEC-6 cells were transfected with siRNAs against CAV-1 for 48 h, then following the treatment with fluorescein labeled GFPBW1 and WGE, and visualized by microscopy (j), the quantitative results were shown in (k). l Meanwhile, western blots were used to detect the expression of caveolin-1. Scale bar, 10 μm. Data are presented as mean ± SD from three biologically independent experiments. Unpaired two-tailed t test. Source data are provided as a Source Data file.

Next, we utilized inhibitors to block various endocytosis routes and then measured polysaccharides uptake in intestinal cells. HIEC-6 cells were pretreated with the chloropromazine (CPZ, inhibitor of CME), Nystatin (inhibitor of lipid raft mediated endocytosis), Methyl-β-cyclodexrin (inhibitor of caveolin mediated endocytosis) or cytochalasin D (Cyto D, inhibitor of macropinocytosis) for indicated time, then incubated with polysaccharides or endocytosis route markers. Interestingly, the uptake of both GFPBW1 and WGE was dramatically inhibited after CPZ disturbance, while transferrin was used as the marker of CME, suggesting that polysaccharide internalization might be mediated by CME (Fig. 3b, c). Then, similar studies were also performed on IEC-6 cells. Consistently, uptake of GFPBW1 and WGE was reduced in CPZ treated group than that of control cells (Ctrl) in IEC-6 (Supplementary Fig. 3a, b). However, the uptake of both GFPBW1 and WGE in Nystatin treated HIEC-6 cells showed no obvious difference with that of Ctrl and DMSO group, while the uptake of cholera toxin B (marker of lipid raft mediated endocytosis) was nearly abolished (Fig. 3d, e), which indicated the effective inhibition of lipid raft endocytosis. We observed similar phenomenon in IEC-6 cells (Supplementary Fig. 3c, d). Furthermore, the fluorescence images indicated that the internalization of GFPBW1 or WGE was not affected by Cyto D (Fig. 3f, g); this observation was confirmed in IEC-6 cells under the same condition while the uptake of FD40 (marker of macropinocytosis or fluid-phase endocytosis) was severely impaired by Cyto D (Supplementary Fig. 3e, f). The above data suggested that internalization of GFPBW1 or WGE might be mediated by CME but not through macropinocytosis or lipid raft mediated endocytosis.

It is reported that cell membrane-expressed proteoglycan is internalized via caveolin-containing endosomes³⁸. Thus, we tried to determine whether caveolae-mediated endocytosis was involved in exogenous polysaccharides internalization. A caveolin-free Huh7 cell³⁹ was therefore used to test the hypothesis. However, out of our expectation, the tested polysaccharides could still internalize into Huh7 cells, further implying that polysaccharides uptake did not rely on caveolae mediated endocytosis (Supplementary Fig. 10a). Furthermore, the internalization of GFPBW1 and WGE was unaffected by methyl-β-cyclodextrin, a specific inhibitor that blocks caveolae-mediated endocytosis (Fig. 3h, i). More importantly, we then knockdown CAV1 (caveolin-1, hallmark protein and key regulator of caveolae-mediated endocytosis) in HIEC-6 cells, western blots results confirmed the good efficiency of the siRNAs (Fig. 3l). And the following uptake experiment showed that the internalization of GFPBW1 and WGE in si-CAV1 groups were comparable with that of Sham or Ctrl group, indicating that caveolin-1 did not affect polysaccharide uptake (Fig. 3j, k). Taken together, these data provide strong evidence that caveolae-mediated endocytosis does not contribute to polysaccharide uptake.

As mentioned above, we labeled the polysaccharides using various strategies (radioactive and fluorescent labeling) and detected their presence in rodent plasma following oral administration. This poses the question: how do the polysaccharides enter the systemic circulation? To address this, we employed human microvascular endothelial cells (HMEC-1) as a cell model. We first performed the kinetic uptake assay of GFPBW1 and WGE, fluorescence was detectable as early as 5 min after the treatment and became more pronounced at 30 min and 1 h, revealing that both the two polysaccharides can enter the microvascular endothelial cells (Supplementary Fig. 4). To further investigate the underlying molecular mechanism, we subsequently employed established pharmacological inhibitors (CPZ, inhibitor of CME; dynasore, inhibitor of dynamin; cytochalasin D, inhibitor of actin driven macropinocytosis; amiloride and EIPA, inhibitors for macropinocytosis; Nystatin, inhibitor of lipid/raft mediated endocytosis; methyl-β-cyclodexrin, inhibitor of caveolin mediated endocytosis) as we did in the intestinal cells. The results showed the uptake of GFPBW1 and WGE was markedly decreased by the treatment of both CPZ (Supplementary Fig. 5) and dynasore (Supplementary Fig. 6), however, the internalization of polysaccharides did not affect by cytochalasin D (Supplementary Fig. 7), amiloride (Supplementary Fig. 8a, b), EIPA (Supplementary Fig. 8c, d), Nystatin (Supplementary Fig. 9a, b) or methyl-β-cyclodexrin (Supplementary Fig. 9c, d), suggesting that the uptake of polysaccharides in HMEC-1 cells might rely on clathrin dependent endocytosis. In other words, polysaccharides might enter the blood stream via CME.

Polysaccharides uptake was impaired in CLTC knockdown intestine cells and CLTC deficient animals

To identify critical proteins during polysaccharide internalization process, HIEC-6 and IEC-6 cells were treated with fluorescein-labeled GFPBW1 and WGE, followed by immunostaining with anti-clathrin heavy chain (CLTC). Intriguingly, we found that the internalized polysaccharides could co-localize with CLTC protein in both HIEC-6 (Fig. 4a, b) and IEC-6 (Fig. 4c, d), indicating that CLTC might play a direct role in polysaccharides uptake. More importantly, we further observed the co-localization of the internalized polysaccharides and CLTC protein in Huh7 cells (Supplementary Fig. 10a). To further ascertain the requirement of CLTC for polysaccharides uptake, we firstly performed CLTC knockdown in HIEC-6 cells and then assessed polysaccharides uptake. The results showed that both the internalization of GFPBW1 and WGE was significantly decreased upon CLTC knockdown. The efficiency of the knockdown was confirmed by immunoblotting (Fig. 4e–g). We next overexpressed CLTC in the cells by transfection with its full-length plasmid and followed by the polysaccharide uptake assay. The results showed that CLTC overexpression significantly enhanced the uptake of both GFPBW1 and WGE (Fig. 4h–j). This result encourages us to ask whether this situation also occurs in animals. Then, we generated intestinal epithelial cell (IEC) -specific CLTC-knockout mice using Cre-loxP conditional gene targeting CLTC (CLTC CKO mice). The fluorescein labeled WGE was orally administrated to both the CLTC^flox/flox (Ctrl) and CLTC CKO mice. Surprisingly, we observed trace amount of internalized WGE in the ileum of the CLTC CKO mice, while obvious fluorescence was observed in the ileum of Ctrl  mice, suggesting that silencing CLTC sharply blocked the uptake of polysaccharides in vivo (Fig. 4k, l). Hence, our findings revealed a fundamental role of CLTC in polysaccharide uptake with these in vitro and in vivo validations. We then asked whether a direct interaction exists between clathrin and polysaccharides? Surface Plasmon Resonance (SPR) strategy was employed to conduct the exploration. Surprisingly, the SPR results showed that both GFPBW1 and WGE bound directly to clathrin with an estimated equilibrium dissociation constant (KD) of 1.149 × 10⁻⁶ and 3.506 × 10⁻⁶, respectively (Fig. 4m, n).

Fig. 4. Clathrin heavy chain was required for the internalization of polysaccharides.

HIEC-6 (a) or IEC-6 cells (c) were treated with fluorescein labeled polysaccharides and then subjected to immunostaining for  clathrin heavy chain (CLTC). Scare bar, 10 μm, 2 μm for the enlarged images. The colocalization analysis results are shown in b, d, respectively. HIEC-6 cells were transfected with si-CLTC or scrambled siRNA (Sham); the uptake experiments results are displayed in (e), siRNA efficiency was detected by Western blotting (f), and the fluorescence intensity was measured in (g). Data are presented as mean ± SEM from three biologically independent experiments. Unpaired two-tailed t test, ***p < 0.001, ****p < 0.0001. Additionally, cells were transfected with CLTC and the vector plasmid for 48 h, then treated with fluorescein labeled GFPBW1 and WGE, and imaged using confocal microscopy (h). The quantified results were shown in (i). j The expression of CLTC detected by western blots was shown. Scale bars, 10 μm. Data are presented as mean ± SD from three biologically independent experiments. Unpaired two-tailed t test, *p < 0.05, **p < 0.01. k Representative images of the jejunum and ileum from the CLTC^flox/flox (Ctrl) and CLTC CKO mice after oral administration of fluorescein labeled WGE, scare bar, 100 μm, the mean intensity was analyzed and showed in (l). Data are presented as mean ± SEM. Three independent experiments, n = 5 mice for each group. Unpaired two-tailed t test, *p < 0.05, **p < 0.01. The interactions between GFPBW1 (m) and WGE (n) with recombinant human CLTC (981-1100 aa) were determined by SPR. Source data are provided as a Source Data file.

Dynamin1 and Rab5 were required for polysaccharides uptake

Although the role of CLTC protein in polysaccharide absorption has been characterized in our above investigation, other proteins involved in polysaccharide uptake are largely unknown. Dynamin, best studied for its role in CME, is a prototypical member of a family of multi-domain GTPases involved in fission and remodeling of multiple organelles⁴⁰. To examine whether dynamin plays a role in polysaccharide uptake, HIEC-6 cells were treated with dynasore (dynamin inhibitor) in presence of GFPBW1 and WGE. The results showed that the internalized GFPBW1 and WGE were substantially reduced after blockage of dynamin function (Fig. 5a, b), suggesting the extracellular to intracellular transport of the polysaccharide is likely reliant on dynamin. Unexpectedly, uptake of GFPBW1 in IEC-6 cells was not affected by dynamin inhibition, while the internalization of WGE and transferrin were sharply decreased after dynasore treatment (Supplementary Fig. 11), indicating even uptake of the same polysaccharide is cell species-dependent. Since mammalian genomes contain three closely related tissue-specific dynamin genes, siRNAs against DNM1 and DNM2 were employed to determine which isoform was required for the polysaccharide uptake by the cells. Interestingly, uptake of GFPBW1 was impaired by si-DNM1, and the uptake of WGE and transferrin was almost completely abolished by DNM1 depletion, while the knockdown efficiency of DNM1 was confirmed by immunoblot (Fig. 5c–e). Interestingly, although the expression of DNM2 was reduced sharply by si-DNM2, the internalization of GFPBW1 and WGE was not significantly disturbed by DNM2 silencing (Supplementary Fig. 12). Collectively, we speculated that DNM1 rather than DNM2 was responsible for polysaccharides uptake at least in the tested intestine cells. To further confirm this phenomenon in vivo, DNM1 conventional knockout mice were generated firstly, unfortunately, we failed to obtain the homozygous DNM1 KO mice due to that homozygous DNM1 offspring died after birth. Therefore, wild type (Ctrl) and heterozygous mice (DNM1 KO, heterozygote) were employed to do the flowing exploration. Interestingly, both GFPBW1 and WGE were internalized in Ctrl mice intestine after polysaccharides orally administrated, while the intestinal internalized GFPBW1 and WGE were slightly or sharply decreased in DNM1 KO heterozygous mice, respectively (Fig. 5f, g). The above data suggested that dynamin 1 is required for the GFPBW1 and WGE internalized in the intestine cells. More importantly, when we overexpressed dynamin1, the uptake of GFPBW1 and WGE was dramatically enhanced, further confirming the critical role of dymanmin1 in this process (Fig. 5h-j). Even more interestingly, the SPR result revealed that DNM1 could bind to GFPBW1 with the KD of 0.137 × 10⁻⁶ (Fig. 5k), while WGE showed no direct interaction with DNM1 (Supplementary Fig. 13).

Fig. 5. Dynamin 1 played a critical role in polysaccharide uptake.

a HIEC-6 cells were treated with dynasore and then performed the uptake experiment of GFPBW1 or WGE. Scare bar, 10 μm. The intensity was quantified in (b). Further, images of the internalized GFPBW1 and WGE in HIEC-6 cells after the disturbing of si-DNM1 are presented in (c). Western blotting results showed the siRNA efficiency (d), the quantified results are shown in (e). f Images of the internalized polysaccharides in mice intestine from wild type mice (Ctrl, upper panel) and DNM1 KO mice (heterozygotes, bottom panel), n = 5 mice for each group, the intensity was quantified in (g). Data are presented as mean ± SEM from three biologically independent experiments. Unpaired two-tailed t test, *p < 0.05, **p < 0.01, ***p < 0.001. h The internalization of GFPBW1 and WGE in the cells transfected with DNM1 and the vector plasmid was examined by microscopy, and the quantitation of the fluorescence was exhibited in (i). j The expression of DNM1 detected by Western blots was shown. Scale bar, 10 μm. Data are presented as mean ± SD from three biologically independent experiments. Unpaired two-tailed t test, *p < 0.05, **p < 0.01. k SPR assay determined the interaction of GFPBW1 and DNM1. Source data are provided as a Source Data file.

Rab proteins are a large family of GTPases that regulate vesicle transport in cells. Rab5 is a marker of early endosomes and plays an important role in CME⁴¹. To understand the possible role of Rab5 in polysaccharide uptake mediated by clathrin, immunostaining for  Rab5 was performed after both HIEC-6 (Fig. 6a, b) and IEC-6 cells (Supplementary Fig. 14) were incubated with GFPBW1 and WGE. Interestingly, Rab5 was partially or completely colocalized with the internalized GFPBW1 and WGE in both cells, suggesting that Rab5 probably involved in polysaccharides uptake. Moreover, fluorescence imaging showed that the uptake of GFPBW1 and WGE was dramatically inhibited by RAB5A silencing, while Western blotting analyses confirmed the knockdown efficiency (Fig. 6c–e), suggesting that Rab5 was also critical for polysaccharides uptake. To further evaluate the role of Rab5 in polysaccharide trafficking in vivo, RAB5A conventional knockout mice were generated. Then the fluorescein-labeled GFPBW1 or WGE were orally administrated to both wild type (Ctrl) and RAB5A^−/− (RAB5A KO) mice. Indeed, internalized polysaccharides in both duodenum and jejunum segments of the Ctrl mice were observed obviously, while trace amounts of internalized polysaccharides could be detected in RAB5A KO mice, indicating that littermates of the RAB5A KO mice displayed a marked impairment in their ability to internalize polysaccharides (Fig. 6f–i). Taken together, these data demonstrated that Rab5 was also required for polysaccharides uptake in vitro and in vivo. More importantly, when we overexpressed Rab5, the uptake of GFPBW1 and WGE was dramatically increased, further confirming the important role of Rab5 in this process (Fig. 6j–l). Even more interestingly, the SPR results showed that both GFPBW1 and WGE bound directly to Rab5 with an estimated equilibrium dissociation constant (KD) of 9.193 × 10⁻⁸ and 1.080 × 10⁻⁶, respectively (Fig. 6m, n).

Fig. 6. Rab5 was required for polysaccharide uptake.

HIEC-6 (a) were treated with fluorescein labeled polysaccharides and then subjected to immunostaining for  Rab5. Scare bar, 10 μm, 2 μm for the enlarged images. The colocalization analysis results are showed in (b). HIEC-6 cells were transfected with siRNAs against RAB5A, then treated with polysaccharides. Images of the internalized GFPBW1 and WGE were shown in (c), and the quantified results were exhibited in (e), and Western blotting results showed the siRNA efficiency (d). Data are presented as mean ± SEM from three biologically independent experiments. Unpaired two-tailed t test, *p < 0.05, **p < 0.01, ****p < 0.0001. Control and RAB5A KO mice (n = 5 mice for each group) were oral administrated with fluorescein labeled GFPBW1 (f) and WGE (g), images of the duodenum and jejunum of the mice were shown, the fluorescence intensity was analyzed by Fiji and shown in (h, i), respectively. Scare bar, 100 μm, 20 μm for the enlarged images. Data are presented as mean ± SEM from three biologically independent experiments. Unpaired two-tailed t test, ***p < 0.001, ****p < 0.0001. j–l Cells were transfected with RAB5A and the vector plasmid for 48 h, treated with fluorescein labeled GFPBW1 and WGE, and imaged using confocal microscopy, the quantified results are shown in (l). k The expression of Rab5 detected by Western blots was shown. Data are presented as mean ± SD from three biologically independent experiments. Unpaired two-tailed t test, *p < 0.05, **p < 0.01. m, n SPR assay was performed to determine the interaction of polysaccharides and Rab5. Source data are provided as a Source Data file.

Internalized polysaccharides were detectable in endosome, lysosome, Golgi, and endoplasmic reticulum in intestinal cells

Further, we were wondering how the intracellular fates of the internalized polysaccharides are after the uptake? To address this question, HIEC-6 cells were incubated with fluorescein-labeled GFPBW1 and WGE for the indicated time and then immuno-stained for  EEA1 (endosome marker), LAMP1 (lysosome marker), AIF (mitochondria marker), RCAS1 (Golgi marker), or PDI (endoplasmic reticulum marker), respectively. Both GFPBW1 and WGE were mainly colocalized with EEA1 as expected (Fig. 7a, b). Furthermore, significant amount of colocalization between the internalized GFPBW1, WGE with LAMP1 was also observed (Fig. 7c, d). However, it seemed that both polysaccharides were not much in mitochondria, since there was little detectable association between polysaccharides and AIF (Fig. 7e). Moreover, we found that the internalized polysaccharides were significantly colocalized with the marker of Golgi apparatus-RCAS1 (Fig. 7f, g). Finally, the internalized polysaccharides were mainly colocalized with PDI, which is the marker of endoplasmic reticulum (ER) (Fig. 6h, i). Thus, based on the above results, we speculated that polysaccharide might be dynamically transported among organelles. Upon internalization, polysaccharides could move from the early endosomes to lysosomes or/and further moved to other organelles such as Golgi, ER but not mitochondria.

Fig. 7. The internalized polysaccharides located in endosome and lysosome, Golgi, endoplasmic reticulum (ER), but not in mitochondria, in HIEC-6 cells.

HIEC-6 cells were treated with labeled GFPBW1 and WGE for 1 h, then the cells were immune-stained for  EEA1 (a), AIF (e), RCAS1 (f), and PDI (h); the colocalization analysis results are exhibited in b, g, i, respectively. Meanwhile, HIEC-6 cells were incubated with GPFPBW1 and WGE for 1.5 h and immune-stained for LAMP1 (c), and the colocalization analysis is in d. Scale bar, 10 μm, 2 μm for the enlarged images. n = 3 samples per group. Source data are provided as a Source Data file.

EGFR, dectin-1, and BMPRIA were implicated in polysaccharides uptake

It is reported that transport of some polysaccharides was mediated by specific receptors⁴². Previously, we found that a sulfation derivative of WGE could bind to BMPRIA, while pectin RN1 (branched l,6-linked galactan with arabinose) could bind to EGFR⁴³, indicating that these receptors might mediate specifical functions of exopolysaccharides. Additionally, dectin-1 was the well-known receptor for β-1,3 glucan,  this phenomenon is also demonstrated by this group, evidenced by the binding between GFPBW1 and dectin-1 in our previous work³². To further understand the role of cell membrane receptors in polysaccharides uptake, HIEC-6 cells were treated with fluorescein GFPBW1 and WGE following EGFR knockdown. The results showed that the uptake of both polysaccharides was dramatically diminished, while the internalization of transferrin showed no difference between the control and EGFR knockdown cells, while the efficiency of siRNAs was confirmed by immuno-blotting (Fig. 8a–c). Furthermore, we observed similar phenomenon in T24 cells either by Gefitinib (EGFR inhibitor, Supplementary Fig. 15a, b) or by si-EGFR interference (Supplementary Fig. 15c-e), further validated the important role of EGFR in polysaccharides uptake. Even more critically, overexpression of EGFR led to a marked increase in the uptake of both GFPBW1 and WGE (Fig. 8d–f). Surprisingly, when we assessed the interaction using SPR, no direct binding was detected between the polysaccharides (both GFPBW1 and WGE) and EGFR (Supplementary Fig. 16). Therefore, further investigation is required to elucidate the molecular mechanism by which EGFR mediates polysaccharides uptake. Secondly, siRNAs were utilized to knock down the expression of CLEC7A  in HIEC-6 cells, the fluorescence images revealed that the internalization of GFPBW1 was significantly abolished, while the uptake of transferrin did not affect (Fig. 8g–i). Meanwhile, we validated the importance of dectin-1 in GFPBW1 internalization using anti-dectin-1 treatment in T24 cells (Supplementary Fig. 17). To further confirm the role of dectin-1 in GFPBW1 uptake, we overexpressed this protein in HIEC-6 cells. The subsequent cellular uptake assay demonstrated that overexpression of dectin-1 significantly promoted the uptake of GPFBW1, while the uptake of WGE in overexpressed dectin-1 group was comparable to that of Ctrl and vector groups (Fig. 8j–l). More importantly, the SPR results revealed the direct interaction of GFPBW1 and dectin-1 with the KD value of 1.063 × 10⁻⁸, while WGE could not bind to dectin-1 (Supplementary Fig. 18), further emphasized the importance of dectin-1 in GFPBW1 uptake. Thirdly, the internalization of WGE was indeed dramatically blocked by BMPRIA silence using siRNA in HIEC-6 cells, while transferrin showed no differences (Fig. 8m–o). Moreover, the penetration of WGE was sharply blocked after BMPRIA depletion via anti-BMPRIA antibody treatment or siRNA interference in T24 cells (Supplementary Fig. 19). Next, we overexpressed BMPRIA in HIEC-6 cells,  the following uptake experiment showed that the internalization of WGE was significantly augmented by BMPRIA overexpression, while the uptake of GFPBW1 was not affected (Fig. 8p–r). However, the subsequent SPR experiment revealed no direct binding between WGE and BMPRIA. Intriguingly, GFPBW1 showed a strong binding with BMPRIA (Supplementary Fig. 20). Nevertheless, these interesting findings still needed more detailed exploration. Hence, we speculated that the receptors EGFR, BMPRIA, and dectin-1 were involved in the uptake for their corresponding polysaccharides.

Fig. 8. Blockage of EGFR, BMPRIA, dectin-1 could suppress polysaccharide internalization in HIEC-6 cells.

The uptake results of GFPBW1 and WGE in HIEC-6 cells after transfecting with si-EGFR or si-CLEC7A, or si-BMPRIA are displayed in a, g, m; the fluorescence intensity was quantified and is exhibited in b, h, n, respectively. Data are presented as mean ± SEM from three biologically independent experiments. Unpaired two-tailed t test, *p < 0.05, **p < 0.01, ***p < 0.001. Western blotting was used to detect the expression of EGFR (c), Dectin-1 (i), and BMPRIA (o) to confirm siRNAs efficiency. Next, HIEC-6 cells were transfected with full length of EGFR, Dectin-1, BMPRIA, and their corresponding vectors, the images of uptake experiments performed 48 h later are shown in d, j, p. The fluorescence intensity was quantified and is exhibited in e, l, q, respectively. Western blotting was used to detect the expression of EGFR (f), Dectin-1 (k), and BMPRIA (r) to confirm the overexpression efficiency. Scare bar, 10 μm. Data are presented as mean ± SD from three biologically independent experiments. Unpaired two-tailed t test, *p < 0.05, **p < 0.01, ***p < 0.001. Source data are provided as a Source Data file.

Wnt/β-catenin and NF-κB signaling pathways play a role in polysaccharides uptake

To further understand whether some signaling pathways are also implicated in the polysaccharide uptake, HIEC-6 (Supplementary Fig. 21a, b), Caco-2 (Supplementary Fig. 21c, d), or T24 cells (Supplementary Fig. 21e, f) were pretreated with XAV939 (Wnt/β-catenin signaling inhibitor), then incubated with fluorescein labeled GFPBW1 and WGE. The results showed that internalization of the two polysaccharides was significantly attenuated after XAV939 treatment in all the tested cell line (Supplementary Fig. 21a–f), indicating wnt/β-catenin signaling might be implicated in polysaccharides uptake. Interestingly, uptake of GFPBW1 and WGE was also sharply blocked by TPCA-1 (NF-κB inhibitor) (Supplementary Fig. 21g, h). Thus, these findings revealed that wnt/β-catenin and NF-κB signaling pathway might be critical in polysaccharide uptake.

Polysaccharide could be internalized in T24 cells via CME, dynamin1, and Rab5 were required during this process

To determine whether the polysaccharide uptake phenomenon is widespread and to assess its cell-type specificity, we used a cancer cell line of epithelial morphology named T24. This approach also helped to exclude variation between individual cells. Firstly, the uptake dynamics of GFPBW1 and WGE were examined in T24 cells. The results demonstrated that uptake of GFPBW1 showed similar pattern in T24 cells with that in HIEC-6 cells among different time points, while WGE exhibited a different manner with that in intestine cells (Fig. 9a, b). Meanwhile, both GFPBW1 and WGE displayed a concentration-dependent manner as in HIEC-6, IEC-6, and Caco-2 cells (Fig. 9c, d).

Fig. 9. Polysaccharides could be internalized in T24 cells, and clathrin mediated endocytosis played an essential role during this process.

a, b T24 cells were treated with fluorescein labeled GFPBW1 and WGE for the indicated time points, the images were detected by microscopy, the quantification results were shown in (b). Data are presented as mean ± SEM. n = 3 samples per group. c, d The effects of polysaccharides concentration on their uptake in T24 cells were examined, and the images and quantification results were displayed in c, d, respectively. Data are presented as mean ± SEM. n = 3 samples per group. e The binding curves for the interaction of GFPBW1 to living and immobilized T24 cells, detected by QCM, are shown. f–h T24 cells were pretreated with the inhibitors of various endocytosis routes (CPZ, clathrin-mediated endocytosis inhibitor, Nystatin, lipid raft mediated endocytosis inhibitor, Cyto D, macropinocytosis inhibitor) for the indicated time intervals, then the uptake experiments of GFPBW1 and WGE were performed, the images are displayed in f, h, g, while the fluorescence intensity were quantified in i, j, k separately. Data are presented as mean ± SEM. n = 3 samples per group. Unpaired two-tailed t-test. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Scale bar, 10 μm. Source data are provided as a Source Data file.

To gain insight to the dynamic uptake of polysaccharide, the interaction of GFPBW1 to living and immobilized T24 cells was detected by quartz crystal microbalance (QCM) biosensor, which has proved to be a powerful tool for studying macromolecular interactions⁴⁴. Interestingly, the binding activity of GFPBW1 to living T24 cells was dramatically stronger than that of the immobilized cells, suggesting that GFPBW1 was dynamically internalized into living T24 cells (Fig. 9e).

Next, similar with what we had done in intestine cells, inhibitors were employed to preliminarily determine the uptake route of polysaccharides in T24 cells. CPZ treatment caused a dramatic decrease in both uptake of GFPBW1 and WGE (Fig. 9f, i), while no obvious changes were observed in neither Nystatin (Fig. 9h, j) nor Cyto D (Fig. 9g, k) treated cells compared with Ctrl, revealing that CME might be also critical in polysaccharides uptake in T24 cells.

To further confirm this result, the immunostaining of CLTC was conducted. Most of the internalized polysaccharides were indeed colocalized with CLTC as expected (Fig. 10a, b). Moreover, we found the binding of GFPBW1 to living T24 cells was blocked by anti-CLTC treatment, further confirmed the critical role of clathrin in GFPBW1 uptake (Fig. 10c). Subsequently, we found that knockdown CLTC with siRNA also cause internalization inhibition of both GFPBW1 and WGE (Fig. 10d, f), while the siRNA efficiency was checked by immunoblot in the T24 cells (Fig. 10e). Furthermore, we also utilized QCM to confirm siRNA efficiency (Fig. 10g). Interestingly, the binding curve of the blank group was much higher than that of si-CLTC group after GFPBW1 treatment, indicting the inhibition of GFPBW1 internalization was caused by CLTC knockdown (Fig. 10h). To further confirm, we wondered whether overexpression of clathrin would promote polysaccharide internalization. To test, full-length CLTC cDNA plasmid was transfected in T24 cells, Western blotting analysis confirmed the over-expression of CLTC (Supplementary Fig. 22c). However, the amount of internalized GFPBW1 or WGE was not significantly increased (Supplementary Fig. 22a, b). Interestingly, polysaccharides could not colocalize with caveolin-1 in T24 cells (Supplementary Fig. 10b), further confirmed that polysaccharides uptake did not require caveolin.

Fig. 10. Requirement of CLTC, dynamin1, and Rab5 for the uptake of polysaccharides in T24 cells.

a Colocalization examination was conducted in T24 cells after the incubation of GFPBW1 and WGE for 1 h (green, polysaccharides, red, CLTC), the colocalization analysis results are shown in b. c The interaction of GFPBW1 to living cells in the absence and presence of CLTC antibodies (4 μg/mL). d–f T24 cells were transfected with si-CLTC and scrambled siRNA (Sham). The results of the uptake experiments are presented in d (red, polysaccharides). The efficiency of siRNA was detected by Western blotting (e), and the fluorescence intensity was quantified in (f). g, h The binding assays of fixed wild-type (Ctrl group) and CLTC-depleted T24 cells (si-CLTC group) to anti-CLTC were performed after the transfection with si-CLTC for 48 h. After regeneration of the chips, an injection of GFPBW1 (100 μg/mL) to both channels performed and the binding curves are displayed in (h). The immunostaining results of Rab5 after the treatment of polysaccharides and the colocalization analysis results were presented in i and j, respectively. k Uptake experiments of GFPBW1 and WGE were conducted after the transfection of si-RAB5A, the fluorescence intensity was quantified in l and the efficiency of siRNA was measured by immunoblot (m). n–q Further, cells treated with GFPBW1 and WGE were then fixed and stained with dynamin, the immunofluorescence images and the colocalization analysis were shown in (n) and o, respectively. p Images of the internalization of polysaccharides in T24 cells after the disturbing of si-DNM1, the quantified fluorescence result was exhibited in q. Western blotting results showed the siRNA efficiency (r). Representative images of the polysaccharide uptake were shown in T24 cells after the treatment of dynasore (s), and the fluorescence intensity was quantified in (t). Scale bar, 10 μm, 2 μm for the enlarged images. Data are presented as mean ± SEM from three biologically independent experiments. Unpaired two-tailed t-test. *p < 0.05, **p < 0.01, ***p < 0.001. Source data are provided as a Source Data file.

Similar as our investigation in intestine cells, the immunostaining for  Rab5 was performed and the results showed that Rab5 could colocalize with internalized GFPBW1 and WGE (Fig. 10i, j) in T24 cells. Furthermore, the uptake of GFPBW1 and WGE was blocked after RAB5A depletion, while the knockdown efficiency was examined by immunoblot (Fig. 10k–m). Similarly, T24 cells were cultured with fluorescein labeled polysaccharides, followed by immuno-staining with anti-dynamin antibody. Indeed, the internalized GFPBW1 and WGE showed colocalization with dynamin both on the cell membrane and in the cytoplasm (Fig. 10n, o), indicating that dynamin may also be involved in polysaccharides uptake. Subsequently, the expression of DNM1 was disturbed by siRNA, the microscopy analysis indicated that the uptake of GFPBW1 and WGE was sharply blocked in the DNM1 silence T24 cells (Fig. 10p–r). Meanwhile, the internalization of the polysaccharides was dramatically attenuated by dynamin inhibition caused by dynasore (Fig. 10s, t) in the same cells. Similar as that of intestine cells, we also found that the internalization of polysaccharides did not affect by DNM2 disturbing (Supplementary Fig. 23).

Eps15 (an acronym for EGFR pathway substrate 15) is an important protein involved in CME^45,46. To determine the role of Eps15 in polysaccharides uptake, immunostaining assays were performed in IEC-6 and T24 cells after treatment with labeled polysaccharides. The results revealed that GFPBW1 and WGE were co-localized with Eps15 in most areas of the cytoplasm in both cells (Supplementary Fig. 24), suggesting that Eps15 might also be involved in polysaccharide uptake.

The internalized polysaccharides might be examined in endosome, lysosome, and Golgi in T24 cancer cells

The above results outlined extracellular to intracellular transport of polysaccharides in human intestinal epithelial cells, but how is their intracellular feature in cancer cells? Thus, similar strategy was employed as we used in HIEC-6 cells. The immuno-staining experiments revealed that both GFPBW1 and WGE showed strong colocalization with EEA1 (Supplementary Fig. 25a, b) and LAMP1 (Supplementary Fig. 25d, e) as expected. Moreover, we found the internalized polysaccharide were significantly colocalized with RCAS1, consistent with the results in HIEC-6 cells (Supplementary Fig. 25g, i). However, by using AIF and PDI as markers for mitochondria and ER, respectively, we found that there was no obvious association between polysaccharides and AIF or PDI (Supplementary Fig. 25c, f). Thus, we speculated that polysaccharide could transport among organelles in cancer cells, they could move from the early endosomes to lysosomes and might further deliver to Golgi, but not mitochondria or ER in cancer cells.

Discussion

This study demonstrated that some natural polysaccharides could be absorbed in intestine after oral administration via CME, including dynamin1, Rab5 as the co-factor proteins based on the results of genetic, in vivo, in vitro, and pharmacological experiments. The major findings of this study include the following: (i) Different linkages and various charged polysaccharide tested in this study were able to penetrate the Caco-2 monolayer with the P_app values of all the tested polysaccharides were greater than 1 × 10⁻⁶ cm/s. Therefore, theoretically, they could be absorbed in vivo according to the correlation between oral drug absorption and P_app values in Caco-2 model³⁴. (ii) By using two different model polysaccharides: GFPBW1 (β-1,3 linked glucan, 300 kDa) and WGE (α-1,4 linked glucan, 700 kDa), we characterized in vivo absorption features in rats and mice using unlabeled, fluorescein-labeled, and radioactive (^99mTc, ³H) labeled strategy. Surprisingly, at least some branched β-1,3 and α-1,4 linked glucan might be detected in mice and rat plasma without significant degradation evidenced by little shift in their retention time on HPLC before and after the polysaccharide oral administration, suggesting that at least some polysaccharides might be absorbed as original macromolecular form with little degradation and could penetrate the intestinal lining and be transported to the blood circulation system. (iii) Although some researchers have tried to move the field forward by figuring out the pharmacokinetic characteristics of several natural polysaccharides^18,19, the mechanism underlying polysaccharides internalization remained enigmatic. In the present study, we show that clathrin contributes to intracellular polysaccharide trafficking by endocytosis, a process also requiring dynamin 1 and Rab5 as well as Eps15 (Figs. 5–7 and Supplementary Fig. 24). Depletion of clathrin, Rab5, or dynamin 1 leads to sharp reduction in polysaccharide uptake in both in vitro and in vivo studies. To our knowledge, this is the first study to comprehensively explore polysaccharide uptake mechanism using pharmaceutic agents, siRNAs, and knockout mouse models. (iv) Previous study regarding natural polysaccharides uptake was mostly investigated in Caco-2 cells, which a colorectal cancer cells. In this study, we demonstrated the universality of polysaccharides internalization in vitro using normal human, rat intestinal cell line, and cancer cell lines. We showed that polysaccharides could internalize into all the tested cell lines, and the uptake dynamics are partially  cell line specific. To explore the underlying mechanism, we examined the expression of the receptors involved in uptake processing (EGFR, Dectin-1, BMPRIA) and CME related proteins (CLTC, DNM1, Rab5). As shown in Supplementary Fig. 26a, EGFR was highly expressed in Caco-2 and T24 cells, and lowly expressed in normal epithelial cells such as HIEC-6, IEC-6, and HMEC-1cells, and no notable differences were observed among these three cell lines. Interestingly, the expression profiles of BMPRIA and Rab5 were similar to those of EGFR in the tested cell lines (Supplementary Fig. 26c, e). Moreover, DNM1 was high expressed in T24 and HMEC-1 cells, but low expressed in HIEC-6, IEC-6 and Caco-2 cells (Supplementary Fig. 26a). The expression level of Dectin-1 and CLTC did not show significant variation across the cell line panel (Supplementary Fig. 26c). Furthermore, we also examined the expression of the proteins regarding to cell polarity and tight junction in these cell lines. ZO-1 was expressed higher in IEC-6, Caco-2, T24 cells than that in HIEC-6 and HMEC-1 cells. While occludin was highly expressed in Caco-2, T24 and HMEC-1 cells and lowly expressed in HIEC-6 and IEC-6 cells. N-cadherin was highly expressed in T24 and HMEC-1 cells and lowly expressed in Caco-2, HIEC-6 and IEC-6 cells (Supplementary Fig. 26d). Based on the experimental data above, we propose that the differences in the dynamic uptake profiles of polysaccharides among the cell lines are at least partially attributable to their distinct expression patterns of receptors and the proteins demonstrated in this study. (v) We characterized the intracellular transport and cellular destination of the internalized polysaccharides in intestinal cells, and we also further showed that the critical roles of some specific receptors, such as EGFR, BMPRIA, and dectin-1 were required for the tested polysaccharides uptake.

Most clinical biological products, such as proteins and peptides, are primarily delivered via the parenteral route due to their negligible oral bioavailability. This can be attributed to several factors: their large molecular size that hampers interaction with biological barriers, their susceptibility to degradation by proteolytic enzymes, and their high hydrophilicity. Recently, extensive studies make efforts to develop alternative delivery strategies, including permeation enhancers, enzyme inhibitors, and mucoadhesive polymers⁴⁷. Lee group designed a hydrophobic ion-pairing (HIP) of insulin and found that the P_app of insulin was about 6.36 folds enhanced in Caco-2 models⁴⁸. Furthermore, the transport mechanism of the exogenous proteins also attracted researchers’ attention. Wang et al. showed that inhibition of CME predominantly blocked the transportation of 55–100 kDa proteins from the hemolymph into the fat body verifying by RNAi⁴⁹. Bagnat and his group demonstrated that lysosome rich enterocytes internalize dietary protein via receptor mediated and fluid phase endocytosis for intracellular digestion and transcellular transport in the immature gut of zebrafish and suckling mice⁵⁰. Thus, the major molecular mechanism underlying protein transport was endocytosis, consistent with what we demonstrated for the uptake of some natural polysaccharides in this study. However, some natural polysaccharides seem to be more stable in gastrointestinal tract than proteins. Our data showed that at least partial of the absorbed polysaccharide was exhibited as macromolecular form with little degradation in plasma (Figs. 1 and 2).

Furthermore, we were also able to rule out the involvement of some routine endocytosis pathways, such as lipid raft-mediated endocytosis, caveolin-mediated endocytosis, or macropinocytosis, in the internalization of polysaccharides in human intestinal cells (Fig. 3). More interestingly, the intracellular transport of the internalized polysaccharides was traced using colocalization analysis. We uncovered that the internalized polysaccharide could transport to endosome, lysosome, Golgi as well as ER but not mitochondria in HIEC-6 cells (Fig. 7). What surprising us is that the destination of intracellular transport of polysaccharides rely on cell type. The polysaccharides could be transported to endosome, lysosome and Golgi but not ER or mitochondria in T24 cells (Supplementary Fig. 25). Nevertheless, the detailed mechanism underlying this difference still need further exploration.

Except clathrin and its partner proteins, we showed that some cell membrane receptors also play a role in polysaccharides uptake. Moreover, we demonstrated that as receptors, BMPRIA and dectin-1 mediated the endocytosis of WGE and GFPBW1, respectively (Fig. 8). Since polysaccharides with different structure feature may bind to their specific receptors on cell membrane, thus we performed the binding test of the polysaccharide and the receptors. Interestingly, the SPR experiments did not detect a direct interaction between the polysaccharides and EGFR (Supplementary Fig. 16). Furthermore, no interaction was observed between WGE and BMPRIA (Supplementary Fig. 20). Therefore, we speculate that additional mechanisms may exist in the EGFR-mediated polysaccharide uptake and BMPRIA mediated WGE uptake process, which require further experimental validation. Nevertheless, GFPBW1 showed a strong binding with dectin-1 as examined by SPR (Supplementary Fig. 18), and the internalization of GFPBW1 was enhanced by dectin-1 overexpression, further emphasizing the critical role of dectin-1 during GFPBW1 uptake process. Although the mechanism of uptake of polysaccharide has partial understood, this study has some limitations. First, we have demonstrated that we could detect the polysaccharide in rodent plasma after orally administration, we sought to explore the molecular mechanism by which polysaccharides enter the systemic circulation. Indeed, we found that the internalization of polysaccharides in human microvascular endothelial cells (HMEC-1 cells) was dependent on clathrin mediated endocytosis, and dynamin was involved in this process (Supplementary Figs. 5 and 6). Thus, we propose that CME is one mechanism contributing to polysaccharide entry into the blood. However, the exact molecular mechanism requires further investigation. Secondly, we found some membrane receptors could mediate polysaccharide uptake, but further research is required to clarify the specificity of these receptors. In addition, our preliminary experiments showed that when we antagonized FGFR1 using its specific antibody in HIEC-6 cells, the uptake of GFPBW1 and WGE was significantly inhibited, suggesting that the receptor FGFR1 may also be involved in the uptake process (Supplementary Fig. 27). However, given the structural diversity of polysaccharides and the complexity of their bioactivities, we speculate that additional receptors are likely involved in mediating polysaccharides uptake.

Collectively, this study clearly demonstrate that some natural polysaccharides can be absorbed by intestine and enter blood circulation system after oral administration via serials of in vitro and in vivo evidence. We also showed that polysaccharides internalization was via CME with the synergistic action of Rab5, dynamin1 and Eps15. Furthermore, the receptors BMPRIA, dectin-1 and EGFR, and wnt/β-catenin, NF-κB signaling were also critical for polysaccharide endocytosis. Taken together, our findings provide a crucial theoretical foundation for understanding the oral absorption of natural polysaccharides and will facilitate the development of oral drugs based on polysaccharides for clinical use.

Methods

Animal treatment

Male Sprague Dawley rats (SD rats, male), BALB/c and C57BL/6J mice (both female and male) were obtained from Shanghai SLAC Laboratory Animal Center and maintained in Animal Center of Shanghai Institute of Materia Medica, Shanghai, China. All procedures involving experimental animals were carried out following protocols approved by the Institutional Animal Care and Use Committee (IACUC) of the Shanghai Institute of Materia Medica (IACUC code: 2020-10-DK-94, 2020-10-DK-95, 2021-10-DK-102, 2021-10-DK-103) and conformed to the Guide for the Care and Use of Laboratory Animals. All animals were maintained under constant humidity and temperature at standard facilities under specific pathogen-free conditions with tap water and standard laboratory rodent chow.

In vivo pharmacokinetics study in rats

Male SD rats (weight, 250–300 g) for orally administered were conducted jugular vein cannulation for blood sample collection; another group of rats for intravenously administered were subjected to intubation in both the jugular (blood sample collection) and femoral vein (for intravenous administration). Following a two-day recovery after the intubation, lentinan, GFPBW1 and WGE were orally administrated (450 mg/kg, 550 mg/kg, and 750 mg/kg, respectively) or intravenously administered (450 mg/kg, 135 mg/kg, and 450 mg/kg, respectively) to rats. Then, blood samples (500 μL) were collected alternately from the jugular vein at the indicated times (n = 5). Plasma was separated from the collected blood immediately by centrifugation (12,000 rpm, 5 min) and stored at −20 °C for analysis.

Polysaccharides were extracted from the plasma before analysis. Shortly, 100 μL GFPW (2 mg/mL) and 200 μL trichloroacetic acid (30%) were added into 100 μL of the plasma for deproteinization (4 °C, 2 h). Then, 2 mL MilliQ-H₂O was added, followed by centrifugation (10,000 rpm, 10 min). The supernatant was transferred into the mixture of chloroform and butanol (v/v = 4/1) and vigorously shaken for 7 min. This extract was subjected to two more rounds of extraction followed by centrifugation (12,000 rpm, 10 min). The upper supernatant was collected and desalted twice using the Zeba spin desalt column (Thermo Fisher Scientific, USA) and then lyophilized.

The analysis of the samples was performed on an Agilent 1200 high performance liquid chromatography (HPLC) equipped with an Agilent G1362A refractive index (RI) detector and a serially linked combination of an Ultrahydrogel^TM 2000 and Ultrahydrogel^TM 500 column (Waters, USA). The mobile phase was MilliQ-H₂O, and column temperature was set at 25 °C. By comparing the peak height of the polysaccharides extracted from the plasma in the chromatograms with that of the original samples in the chromatograms (standard curve), the concentrations of the serum polysaccharides were obtained. Then, the concentration-time curves of the polysaccharides were achieved. The curves were analyzed using the WinNonlin software (Scientific Consulting Inc., USA) to determine the pharmacodynamic of polysaccharides in rats.

Synthesis and characterization of fluorescein-labeled polysaccharides

Saccharides were labeled using a modified version of Glabe’s³⁵. Briefly, 40 mg polysaccharides dissolved in 20 mL MilliQ-H₂O were initially activated by saturated CNBr solution (Fluka, USA) for 7 min (the pH of the reaction system throughout the process was monitored and maintained above 10.0 by adding 0.25 M NaOH intermittently). The mixture was then dialyzed against distilled water to remove redundant CNBr. 48 h later, the retentate was concentrated and dissolved in Na₂B₄O₇ (0.2 M, pH 8.0, Sinopharm, China), followed by the addition of fluorescein (dansyl ethylenediamine (DED, TRC Inc., Canada), 5-FITC cadaverine (Anaspec, USA) or Texas red (Biotium, USA) and vortexed in dark at room temperature for 18 h. Subsequently, the mixture was centrifugated, and the supernatant was then dialyzed against distilled water for 48 h in dark, then the retentate was concentrated and lyophilized, designated as fluorescein labeled polysaccharides. For fluorescein labeling maltose and raffinose, saccharides were subjected to Bio-Gel-P2 column for further purification after activation with CNBr.

Saccharides and their fluorescein-labeled derivatives (30 mg) were deuterium-exchanged and dissolved in 0.5 mL D₂O (99.8% D, Sigma-Aldrich, USA). The ¹H and ¹³C NMR spectra were measured at 25 °C with the Bruker AM-400 NMR spectrometer (Karlsruhe, Germany), and acetone was used as the internal standard (δ 31.50 ppm for ¹³C-NMR and δ 2.29 ppm for ¹H-NMR).

Preparation of ^99mTc-labeled polysaccharides

The labeling process were performed according to the previous report¹⁷. Shortly, about 40 mg polysaccharides were dissolved in DMSO, then 4 mg of DTPAA and 2 mg of DMAP were added and reacted at 40–45 °C. 24 h later, the solution was further precipitated with ethanol and dialyzed with distilled water for 48 h, followed by the purification of Sephadex G50 column and gave the DTPA-polysaccharide intermediate. Subsequently, 10 μL of freshly prepared SnCl₂·2H₂O was mixed with 1 mL of Na^99mTcO₄ solution for stirring for 5 min. 1 mL of DTPA-polysaccharide intermediate and 50 μL saturated sodium bicarbonate were then added and incubated for 1 h at 60 °C, purification on a PD-10 column (General Electric, Fairfield CT, USA) was then conducted to obtain ^99mTc-labelled polysaccharides.

Generation of RAB5A and DNM1 conventional knockout mice by transcription activator-like effector nucleases (TALENs)

The RAB5A heterozygous KO mice were produced in SiDanSai Biotechnology Co., Ltd (Shanghai, China), then the heterozygous mice were crossed to generate the wild type (Ctrl) and homozygous RAB5A deficient mice (RAB5A KO mice). 8-10 weeks of the male or female mice were used in the following experiments. Similarly, the DNM1 heterozygous KO mice were produced in SiDanSai Biotechnology Co., Ltd (Shanghai, China), then the heterozygous mice were crossed to generate the wild type and homozygous DNM1 deficient mice. Unfortunately, the homozygous DNM1 KO mice died shortly after birth in our study; thus, the heterozygous mice and wild type mice were used to do the polysaccharides absorption experiment.

Generation of intestinal epithelial cell specific CLTC knockout mice

CLTC loxP heterozygous (CLTC^flox/wt) mice were produced by SiDanSai Biotechnology Co., Ltd (Shanghai, China). Then the CLTC flox heterozygous were crossed to obtain a CLTC^flox/flox genetic background, subsequently, the CLTC^flox/flox mice were crossed with C57BL/6 mice expressing Cre recombinase driven by the villin promoter, which was extensively in intestine. Next, resulting progeny (CLTC^flox/wt mice heterozygous for villin-Cre-ERT2) were mated with CLTC^flox/flox mice to generate CLTC^flox/flox mice (Ctrl mice) and intestine-specific CLTC knockout mice (CLTC CKO mice). Male and female mice aged 8-10 weeks were used for the experiments.

In vivo absorption studies in mice

For the absorption experiments in BALB/c or C57BL/6J mice, mice (weighted 20–25 g) were fasted overnight except for water. Mice were orally administrated with fluorescein-labeled polysaccharides for the indicated time points, the intestinal loops were removed and rinsed with saline. After washing, frozen sections of the intestine samples were obtained using a Leica CM1950 cryostat (Leica, Germany): 6-μm thick tissues sections were snap-frozen in Tissue-Tek® O.C.T. compound (Sakura, USA) and then affixed onto gelatin-coated glass slides. Then, the sections were fixed and stained, imaged by confocal microscopy (FV1000, Olympus, Japan).

For the in vivo absorption experiment in KO and CKO mice, 8-10 weeks old male and female mice (Ctrl mice (CLTC^flox/flox) and CLTC CKO mice; RAB5A KO mice and wild type mice (Ctrl); DNM1^+/− mice and wild type mice (Ctrl)) were orally administrated with fluorescein-labeled polysaccharides for the indicated time intervals, the mice were euthanized, and the small intestine was taken for analyses. Small intestine was embedded in Tissue-Tek OCT cryostat molds (Leica) and frozen at −80 °C, then 6-μm thick sections were generated in a cryostat (Leica CM1950, Germany). Then the tissue sections were stained with DAPI and imaged.

For the absorption experiments in C57BL/6 J mice with ³H-labeled polysaccharides, C57BL/6 J mice were orally administrated with ³H-labeled GFPBW1 or WGE (1200 μCi/kg), the blood samples were collected after dosing for the indicated time intervals (1 h, 2 h, 4 h, 8 h for GFPBW1, 0.5 h, 1 h, 2 h, 4 h for WGE). The radioactivity of the dosing formulation and plasma samples was determined with a liquid scintillation counter (LSC) 600 SL (Hidex, Lemminkäisenkatu, Turku, Finland).

For the in vivo absorption experiment with analysis with HPLC, C57BL/6J mice for orally administrated with fluorescein-labeled WGE 100 mg/kg), and after dosing, the blood samples were obtained after centrifugation at 3000 rpm for 5 min, plasma samples were subsequently quantified and analyzed by HPLC.

For the absorption experiments in BALB/c mice with ^99mTc-labeled polysaccharides: mice were dosing by the polysaccharides at 111 MBq (3 mCi/0.5 mL) via oral administration. The blood samples and the liver were collected at 15 min, 30 min, 1 h, 2 h, 4 h, 8 h, 12 h, 16 h, and 24 h, stored properly until analysis.

Analytical conditions and sample pretreatment procedure

Sample pretreatment (for fluorescein-labeled polysaccharides): 400 µL of ethanol was added into a 100 µL portion of plasma sample, then the mixture was vortexed for 2 min and then centrifuged at 12,000 rpm for 20 min to precipitate the polysaccharides. The precipitate was then dissolved with distilled water and vortexed. After the centrifugation for 20 min at 12,000 rpm, the supernatant was collected and lyophilized, then subjected to HPLC analysis.

Sample pretreatment (for ^99mTc-labeled polysaccharides): the plasma or the liver homogenate was mixed with the saturated (NH₄)₂SO₄ with the same volume, then stirred for 10 min, followed by centrifugation at 12,000 rpm for 15 min, the supernatant was collected and subjected to Radio-HPGPC analysis.

For fluorescein-labeled polysaccharides analysis: the assay system consisted of an Agilent 1200 series HPLC with a fluorescence detector set at λex 492 nm and λem 516 nm. Samples were separated by HPGPC using a Shodex KS-804 HQ column. The eluent was 0.1 M phosphate buffer (pH 7.4), delivered at a flow rate of 0.5 mL/min; the chromatographic procedures were performed at 35 °C.

For ^99mTc-labeled polysaccharides analysis: the 99mTc-labeled polysaccharides and the biological samples were analyzed by the Thermo Scientific^TM Dionex^TM UltiMate^TM 3000 HPLC (Thermo, USA) system equipped with a Berthold LB 514 radio detector. Samples were separated by HPGPC using a Waters Ultrahydrogel Liner column (300 mm × 10 μm). The eluent was 0.1 M NaNO₃, delivered at a flow rate of 1.0 mL/min, the chromatographic procedures were performed at 50 °C. Data was analyzed by the Thermo Scientific^TM Chromeleon ^TM chromatography data analysis system (Thermo, USA).

Plasmids and siRNA

Clathrin heavy chain cDNA clone and the vector control (pcmv-6) plasmids were purchased from OriGene (Rockville, MD, USA). PGMLV-CMV-H_DNM1-3 × Flag-PGK-Puro, PGMLV-CMV-H_CLEC7A-3 × Flag-PGK-Puro and GMLV-PGK-Puro plasmids were bought from Genomeditech (Shanghai) Co., Ltd. (Shanghai, China). pCMV-BMPR1A (human) -3 × FLAG-Neo, pCMV-T7-MCS-3 × FLAG-Neo, pLV3-CMV-EGFR (human) -3 × FLAG-CopGFP-Puro, pLV3-CMV-RAB5A (human) -3 × FLAG-CopGFP-Puro and pLV3-CMV-MCS-3 × FLAG-CopGFP-Puro plasmids were purchased from miaoling plasmid (Wuhan, China). siRNA sequences were provided in table S6 in the Supplementary Information.

Cell culture and transfection

Human intestine epithelial cells (HIEC-6 cells, CRL-3266) and rat intestine epithelial cells (IEC-6, CRL-1592) were obtained from the American Type Culture Collection (ATCC, Manassas, USA). Caco-2 (human colorectal cancer cell), HEK-293, Huh7 cells (human hepatoma cell), human microvascular endothelial cells (HMEC-1), and T24 cells (human bladder cancer cell) were from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China). HIEC-6 cells were cultured in Opti-MEM Reduced Serum Medium (31985, Gibco, USA) supplemented with 20 mM HEPES (15630080, Gibco, USA), 10 mM GlutaMAX (35050, Gibco, USA), 10 ng/mL Epidermal Growth Factor (EGF, SinoBiological, Beijing, China), 4% fetal bovine serum (FBS, Gibco, USA), and antibiotics (100 U/mL penicillin and 100 μg/mL streptomycin, Gibco, USA). The other cell lines were maintained in Dulbecco’s modified Eagle’s medium (DMEM, Corning, USA) supplemented with 10% FBS (Gibco) and antibiotics. HMEC-1 cells were cultured in MCDB-131 medium (MeilunBio, Dalian, China) supplemented with 2 mM L-glutamine (MeilunBio, Dalian, China), 10 ng/mL EGF, 15% FBS, and antibiotics. All cells were cultured at 37 °C in an atmosphere of 5% CO₂. Cells were routinely transfected with plasmids and siRNAs using Lipofectamine 2000 (Invitrogen, USA) or Rfect (Baidai biotech, Changzhou, China), respectively, according to the manufacturer’s instructions.

In vitro transepithelial transport studies

The transepithelial transport studies were conducted according to the literature⁵¹. All solutions used were pre-warmed (37 °C). The Caco-2 cell monolayers were firstly washed with phosphate-buffered saline (PBS). Then, the test polysaccharide solutions (5 mg/mL, 0.5 mL) were added into the apical chamber, while 1.5 mL of PBS was added into the basolateral side. Then, 0.5 mL solutions were collected from the basolateral side every 2 h up to 8 h and replaced by 0.4 mL of fresh pre-warmed PBS. The collected samples were stored at -20 °C prior to analysis. The amount of polysaccharide that permeated the Caco-2 cell monolayers was calculated using the formula: equation (1) $Q_n=0.5\times C_n+Q_{n-1}$_, where C represents the concentration of the polysaccharides in the permeated samples, Q means the cumulative amounts of permeated polysaccharides in mg, while n represents the time at which the sample was collected (2, 4, 6, and 8 h). P_app (cm/s) was calculated using the equation: equation (2) $P_{\mathrm{app}}=\left(\mathrm{\Delta }Q/\mathrm{\Delta }t\right)/\left(A\times C_0\right)$, where ΔQ/Δt is the linear appearance rate of the samples on the receiver side in mg/s, A is the surface area in cm², and C₀ is the donor concentration in mg/mL.

In vitro cellular uptake studies of polysaccharides

To estimate the uptake of polysaccharides in HIEC-6, IEC-6, Caco-2, Huh7, HMEC-1 or T24 cells, the cells under various conditions were seeded in a 12-well plate for 24 h, then the cells were incubated with fluorescein-labeled polysaccharides for 1 h. Then, the cells were fixed with 4% paraformaldehyde for 15 min followed by staining with DAPI (#4083, Cell Signaling Technology, USA) and scanning using confocal microscopy (FV 1000, Olympus, Japan; Lecia TCS SPS CFSMP, Leica, Germany).

For the immunostaining assays, cells were blocked with 5% bovine serum albumin (BSA) (with 0.3% Triton X-100) for 1 h after the fixation with 4% paraformaldehyde (w/v, in PBS). Next, the cells were incubated with primary antibodies at 4 °C overnight. After washing with PBS, the cells were incubated with fluorescein-conjugated secondary antibody. After washing to remove the redundant secondary antibodies, the cells were stained with DAPI and imaged using confocal microscopy (FV 1000, Olympus, Japan; Lecia TCS SPS CFSMP, Leica, Germany).

Intracellular transportation studies

To expound the intracellular transportation of polysaccharides, HIEC-6 and T24 cells were cultured with fluorescein labeled GFPBW1 and WGE for 1 h or 1.5 h. Then, the organelle localization IF antibody sample kit (#8653, Cell Signaling Technology, USA) was used to stain the subcellular organelles. Antibodies including AIF rabbit mAb, EEA1 rabbit mAb, LAMP1 rabbit mAb, PDI rabbit mAb and RCAS1 rabbit mAb were employed to stain mitochondria, endosome, lysosome, endoplasmic reticulum, and Golgi, respectively. The colocalization of polysaccharides with the subcellular organelles was imaged by CLSM and analyzed by Fiji Image J.

Western blotting analysis

The treated cells or mouse intestine tissues were harvested and lysed with the cell lysis buffer (#9803, Cell Signaling Technology, USA) on ice for 30 min followed by denaturation for 30 min at 95 °C and centrifugation. The supernatants were subjected to SDS-PAGE for separation, and the gels were then transferred to nitrocellulose membranes (Pall, USA), and the membranes were blocked with 5% w/v nonfat milk in TBST buffer (20 mM Tris-HCl [pH 8.0], 150 mM NaCl, and 0.1% Tween-20, v/v). This was followed by incubation with the relevant primary antibody at 4 °C overnight, and further incubated with an appropriate horseradish peroxidase-conjugated secondary antibody. Finally, Super Signal West Dura Substrate (Thermo, USA) and Pierce ECL Western blot substrate (Pierce, USA) were used for signal detection, GAPDH, β-tubulin, and β-actin were employed as the internal control. All antibodies are listed in Table S6 in the Supplementary Materials.

QCM analysis of polysaccharide—cell interaction

The interaction measurements of polysaccharides—cell were performed using an Attana Cell 200 System QCM Biosensor (Attana AB, Stockholm, Sweden) according to the literatures. Shortly, T24 cells were seeded on the COP-1 chips for 24 h and then subjected to the various treatments. The cells were fixed with 4% paraformaldehyde (fixed cells) or were not fixed (living cells). Next, the cell crystals were placed in a chip holder and docked into QCM channels. Before the measurement, the chips needed to be equilibrated under a continuous flow of running buffer (PBS, pH 7.4). The flow rates were 1 and 25 μL/min for the living and fixed cells, respectively. The resonant frequency of the interaction was recorded using the Attester software (Attana AB, Stockholm, Sweden). After the experiment, the chips were stained by DAPI and the fluorescent images were taken under a fluorescent microscope (Olympus BX53, Japan).

Surface plasmon resonance (SPR) analysis

A Biacore T200 instrument was used to measure the binding kinetics of human CLTC981-1100aa (abmart, Shanghai, China), human DNM11-271aa (Bioss, Beijing, China), human DNM1 (TP306284, OriGene, USA), human Rab5a (NBC1-18504, Novusbio, USA), human EGFR1-645aa, human Dectin-1, human BMPRIA1-152aa (Sino Biological Inc, Beijing, China) to the polysaccharides. The above proteins were immobilized on the sensor chip (CM5, GE Healthcare) using the amine coupling method according to standard protocols. Measurements were performed at 25 °C, different concentrations of polysaccharides (GFPBW1, WGE) were injected into the chips. All the procedures were performed using the HBS-EP, pH 7.4 (GE Healthcare) running buffer. Data were processed using standard double-referencing and fit to a 1:1 binding model using Biacore T200 Evaluation software.

Statistical analysis

Statistical analysis was performed using the GraphPad Prism software package. Results are expressed as mean ± standard deviation (SD) or mean ± standard error of mean (SEM). Significance (p value) of the difference between two groups was determined using Unpaired two-tailed t-test, and p < 0.05 was considered to statistically significant (*, p < 0.05; **, p < 0.01, ***, p < 0.001, ****, p < 0.0001). Differences between multiple groups were analyzed using one-way or two-way ANOVA, followed by Tukey’s or Sidak’s multiple-comparisons test.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

W.F.L., D.X.C., and Y.W. conceived the experiments. W.F.L. carried out most of the experiments and wrote the manuscript. Z.Y.D. helped in the animal experiments. X. C., J. Y., C.W.M., and P.P.W. purified the polysaccharides. Y.D.Z., P.F.D., and Z.M.W. helped with the radioactive polysaccharides analysis. H.C., X.X.D., and K.P.W. helped analyze the data. K.D. directed the overall project and supervised the whole project.

Peer review

Peer review information

Nature Communications thanks Hiroshi Kitagawa, and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

All data essential for validating the conclusions outlined in this manuscript are available within the Source Data file. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-68542-w.