Original Article
Lenvatinib induces diarrhea in rats by promoting intestinal barrier injury via targeting AQP4
2025 Volume 50 Issue 12 Pages 677-690
Details
2025 Volume 50 Issue 12 Pages 677-690
Our study evaluated whether lenvatinib induces the disease process of diarrhea by facilitating intestinal epithelial barrier damage. Sprague-Dawley rats were orally administrated with 0.2 or 2 mg/kg lenvatinib for six consecutive days to induce diarrhea models. The diarrhea rate was monitored every day, and rats were sacrificed on day 6. We found that rats began to develop diarrhea on day 3 after lenvatinib treatment. Almost all rats treated with lenvatinib (2 mg/kg) developed grade 3 diarrhea. Intestinal villi structure damage and obvious inflammatory cell infiltration were observed in the colon tissues of lenvatinib-administrated rats. Lenvatinib significantly upregulated serum contents of intestinal injury biomarkers (D-lactate and DAO) but downregulated colon levels of tight junction proteins (ZO-1, Occludin, and Claudin-1) in rats. In vitro results showed that lenvatinib higher than 5 μM significantly attenuated the viability of human intestinal epithelial cell line Caco-2. Lenvatinib suppressed ZO-1, Occludin, and Claudin-1 levels, decreased the transepithelial electrical resistance value, and elevated paracellular permeability in Caco-2 cells. Mechanically, lenvatinib targeted AQP4 and inhibited its expression. Overexpressing AQP4 reversed lenvatinib-induced intestinal epithelial barrier injury in Caco-2 cells by inhibiting the MLCK/p-MLC2 signaling pathway. Collectively, lenvatinib triggers diarrhea by disrupting the intestinal barrier through downregulating AQP4 and activating the MLCK/p-MLC2 signaling pathway.
Over the past decade, with the advancement of oncology research, attention has been drawn to many specific biological pathways involved in cancer development. Novel drugs targeting specific proteins involved in the process of cell proliferation and spread have been developed, which have to some degree replaced the use of traditional chemotherapeutic approaches in the treatment of advanced solid tumors (Tsimberidou, 2015). Tyrosine kinase inhibitors (TKIs), since their introduction in the early 2000s, have attracted much attention as the most effective pathway-targeted anticancer drugs (Vergoulidou, 2015). TKIs can effectively block tyrosine kinase activity and suppress cell signaling, thus suppressing tumor cell growth and proliferation (Mongre et al., 2021). TKIs, in combination or as monotherapy, are increasingly being used as a first-line treatment option and have shown great utility in the treatment of a wide range of solid tumors and hematologic malignancies (Huang et al., 2020a). Due to their widespread applications, reports of adverse reactions caused by TKIs are becoming more frequent (Pottier et al., 2020). It has been demonstrated that TKIs can affect multiple organs in the body, including the heart, skin, thyroid, kidneys, gastrointestinal tract, liver, and lungs (Shyam Sunder et al., 2023). Gastrointestinal toxicity, usually manifested as diarrhea, is a common side effect of TKI-targeted therapy, for which there is no specific prevention or treatment strategy (Liu et al., 2024a). Diarrhea and related intestinal ulcers can result in a range of serious problems, including dehydration, malnutrition, fatigue, renal insufficiency, and increased risk of systemic infections (Stein et al., 2010).
Lenvatinib, as a multi-targeted TKI, has been approved by the U.S. FDA to treat thyroid cancer, renal cancer, hepatocellular carcinoma, and endometrial cancer (Hao and Wang, 2020). Lenvatinib can prevent the phosphorylation and subsequent activation of many tyrosine kinases that participate in tumor cell proliferation and neo-angiogenesis, thereby effectively counteracting tumor progression (Motzer et al., 2022). In addition, it prevents the formation and maturation of new blood vessels and reduces the vascular permeability of the tumor microenvironment by inhibiting the activity of several kinases, including RET, KIT, PDGFRα, FGFR1-4, and VEGFR1-3 (Qin et al., 2019). Although it is successful in the treatment of cancer, it may cause life-threatening side effects such as irritation of the gastrointestinal mucosa, drug sensitization, damage to the nervous system, injury to the liver and kidneys, and bone marrow suppression (Motzer et al., 2015). Lenvatinib is digested and absorbed in the gastrointestinal tract, often causing irritation of the gastrointestinal mucosa, leading to nausea, vomiting, abdominal pain, diarrhea, and other symptoms, but the exact pathogenic mechanism needs to be explored in depth (Inukai et al., 2023). The intestinal tract is the largest digestive and immune organ of the body and is crucial for maintaining homeostasis of the body's internal environment (Schoultz and Keita Å, 2020). The intestinal mucosal epithelial barrier not only resists microbial entry but also allows for better digestion and absorption of nutrients, being one key component of the body's defense and protection (Turner, 2009). Impairment of the intestinal mucosal epithelial barrier function can induce the development of various autoimmune and inflammatory diseases, thus affecting overall health (Groschwitz and Hogan, 2009). Studies have revealed that improving the intestinal barrier integrity and function helps to prevent and ameliorate diarrhea (Guan et al., 2021). Importantly, TKIs were reported to induce diarrhea by promoting calcium-activated chloride secretion and epithelial barrier damage (Kim et al., 2020). Accordingly, lenvatinib might also trigger the disease process of diarrhea through facilitating intestinal epithelial barrier damage.
Fluid transport is one of the most important functions of the gastrointestinal tract, which is essential for maintaining the body's water balance and physiological functions such as digestion and absorption (Ma and Verkman, 1999). Aquaporins (AQPs) are a family of intrinsic membrane proteins widely distributed in human and mammalian cell membranes and closely related to transmembrane water transport (Ye et al., 2023). To date, 13 AQP isoforms (AQP0-12) have been identified in humans and mammals, of which at least 9, including AQP4, are abundantly expressed in gastrointestinal tissues (Laforenza, 2012). In recent years, studies applying knockout mice have confirmed that AQP4 not only plays a vital role in intestinal water absorption, secretion and metabolism, as well as in regulating intestinal mucosal tight junctions and intestinal barriers, but also participates in cell migration, proliferation, and angiogenesis, which have been associated with tumor formation (Lv et al., 2022). Accumulating evidence has demonstrated the role of AQP4 in the development of diarrhea. For example, AQP4 mRNA and protein levels in the ileum of Enterotoxigenic Escherichia coli-induced mouse diarrhea models gradually decrease over 7 days (Zhang et al., 2018). The probiotics Akkermansia muciniphila can ameliorate antibiotic-associated diarrhea in mouse models and simultaneously enhance water and electrolyte absorption through increasing AQP4 expression (Liu et al., 2024b). Treatment with aconite aqueous extract in mouse models of diarrhea mitigates colon pathological changes, reduces fecal water content and diarrhea symptom, and increases AQP4 content in the colon (Zhang et al., 2023). Berberine, a compound that is extensively applied as a non-prescription agent for diarrhea treatment, can elevate AQP4 expression in both human intestinal epithelium cell line (HIEC) and sennoside A-induced diarrhea mouse models (Zhang et al., 2012). The above literature suggests that AQP4 downregulation is associated with the occurrence and progression of diarrhea, while upregulating AQP4 is beneficial for preventing or hindering diarrhea development. Nevertheless, whether AQP4 plays a role in lenvatinib-induced diarrhea remains unclear. Thus, our research used animal and cellular experiments to assess our hypothesis that lenvatinib might target and inhibit AQP4 to promote intestinal barrier injury, thereby participating in the disease progression of diarrhea.
Healthy adult male and female Sprague-Dawley rats (230-280 g) obtained from Charles River Laboratories (Beijing, China) were used in this study following the approval of the Animal Ethics Committee of Wuhan Myhalic Biotechnology Co., Ltd. (No. HLK-20220415-002). Rats were acclimatized for at least 1 week prior to use and housed in clean metabolic cages at 22 ± 2°C with a humidity of 55% ± 5% in a 12-hr light/dark cycle, with tap water and laboratory chow provided ad libitum.
Experimental designTo establish the lenvatinib-induced diarrhea model, the rats were randomly allocated into the vehicle, lenvatinib (0.2 mg/kg), and lenvatinib (2 mg/kg) groups. Water containing 10% carboxymethyl cellulose was used to prepare lenvatinib suspension. The lenvatinib groups were orally administrated with 0.2 or 2 mg/kg lenvatinib (#LZR5903; Rayzbio, Shanghai, China) for six consecutive days, while the vehicle group was administrated with water containing 10% carboxymethyl cellulose. The diarrhea rate was monitored every day, and rats were sacrificed by isoflurane inhalation on day 6 and the colon tissues from the upper 4 cm of the anus were immediately collected.
Evaluation of diarrheaThe diarrhea state of rats was monitored two times daily. Diarrhea is classified into four grades according to its severity: Grade 0=normal, no diarrhea; Grade 1=mild diarrhea, loose stools that remain the same shape; Grade 2=moderate diarrhea, loose and unformed stools; Grade 3=severe diarrhea, watery stools.
Histopathological stainingThe colon samples were placed for 24 hr in 10% neutral buffered formalin, dehydrated, paraffin-embedded, and sliced into 5‐μm sections using an ultra-thin semiautomatic microtome. After routine dewaxing and dehydration, the sections were stained using the commercial hematoxylin and eosin (H&E) Staining Kit (#G1120; Beijing, China) as directed by the manufacturer. Colon pathology was observed under an Olympus optical microscope.
ELISAAfter rats were anesthetized with isoflurane inhalation, abdominal aortic blood samples were collected into an ethylene diamine tetraacetic acid (EDTA) anticoagulant tube. The samples were centrifuged at 3000 × g, 4°C for 10 min to acquire the serum, which was kept at -20°C until analysis. The contents of D-lactate and diamine oxidase (DAO) in the serum were detected by using D-lactate ELISA Kit (#KBR-hlk4373; Keboruibio, Shanghai, China) and DAO ELISA Kit (#KBR-hlk3932; Keboruibio) following the protocols of the manufacturers. Finally, the absorbance values were measured using a microplate spectrophotometer.
Cell culture, treatment, and transfectionThe human intestinal epithelial cell line, Caco-2, was provided by SUNNCELL (Wuhan, China), and maintained at 37°C in complete Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal bovine serum, 25 mmol/L HEPES, 50 U/mL streptomycin, 50 U/mL penicillin, and 4.5 mg/mL glucose in a humidified atmosphere with 5% CO2. After subculturing by 0.25% trypsin solution containing EDTA, Caco-2 cells differentiated into enterocyte-like cells 18-20 days later. Fully differentiated cells were used for follow-up assays. Subsequently, cells were treated for 24 hr with lenvatinib (0, 0.1, 0.5, 1, 2, 5, 10, and 15 μM) and the cytotoxicity of lenvatinib was analyzed using the Cell Counting Kit-8 (CCK-8). AQP4 overexpression vector (OE-AQP4), MLCK overexpression vector (OE-MLCK), and the negative control (OE-NC) were designed by GenePharma (Shanghai, China) and transfected into Caco-2 cells for 24 hr using Lipofectamine 2000 (#11668019; Invitrogen, Carlsbad, CA, USA).
CCK-8 assayCaco-2 cells after the above transfection and treatment were seeded (3000 cells/well) into 96-well plates for 24 hr. Afterwards, 10 μL of CCK-8 solution (#abs50003; Absin, Shanghai, China) was added to each well, and the cells were cultivated for an additional 2 hr in the dark at 37°C. Ultimately, a microplate reader (Model 680, Bio-Rad, Hercules, CA, USA) was adopted to estimate the absorbance at 450 nm.
Lactate dehydrogenase (LDH) release assayCaco-2 cell death was detected by measuring the LDH activity using an LDH assay kit (#C0016; Beyotime, Shanghai, China). Briefly, Caco-2 cells were seeded (5000 cells/well) into 96-well plates and received the above-mentioned transfection and treatmen for 24 hr at 37°C. Afterwards, the cell culture medium was harvested, and the absorbance was detected at 490 nm with a microplate reader to determine LDH level.
Measurement of transepithelial electrical resistance (TER)Caco-2 cells were plated at a density of 1 × 105 cells/mL on transwell inserts (Millipore, Billerica, MA, USA) with polyethylene terephthalate membrane. Cells were cultured for about 3 weeks after reaching confluence and completely differentiated. The changes in TER were measured with a transepithelial voltohmeter (World Precision Instruments, FL, USA). Before measurement, Hanks’ Balanced Salt Solution (HBSS) was used to wash both apical and basolateral sides of the transwell three times. TER was examined until similar values were recorded for three consecutive measurements and calculated as Ohm·cm2 (Ω·cm2).
Examination of paracellular permeabilityFITC-Dextran 40 kDa (FD-40), an established paracellular marker, was used to measure Caco-2 monolayer permeability. After being washed with HBSS, the well-grown Caco-2 monolayers were cultured with serum-free DMEM without phenol red containing 10 mg/mL FD-40 for 2 hr. Thereafter, 100 µL were taken from the outer chamber to assess the paracellular flux. Fluorescent signals were detected using Synergy H2 microplate reader (Biotek Instruments, USA) with excitation at 485 nm and emission at 535 nm. The permeability coefficient (PE) was calculated according to the formula described in the previous study (Yang et al., 2016).
RT-qPCRTRIzol reagent (#RY0871; Jisskang, Qingdao, China) was used to extract total RNA from Caco-2 cells. Then, 1 µg of total RNA was reversely transcribed into cDNA using PrimeScript RT Master Mix (#RR036A; TaKaRa, Dalian, China). RT-qPCR was conducted with the ABI StepOne Plus system (Thermo Fisher Scientific, USA) by employing the SYBR Green PCR kit (#RR420A; TaKaRa). GAPDH was adopted as the internal control. The relative expression was calculated using the 2−ΔΔCT method. Primer sequences are listed in Table 1.
;%0A%09%09%09newWindow.document.open();%0A%09%09%09newWindow.document.write('<img src=%22./Graphics/50_677-t1.jpg%22>');%0A%09%09%09newWindow.document.close();%0A%09%09)
Caco-2 cells after lenvatinib treatment were washed thrice with phosphate-buffered saline (PBS), followed by 30 min fixation with 4% poly-methyl fermentation solution and 10 min permeabilization with 0.5% Triton X-100. After being blocked for 30 min with 10% goat serum, the anti-ZO-1 (#AF5145; 1:100; Affinity Biosciences) primary antibody was added to the cells and incubated at 4°C overnight. On the second day, Fluor488-conjugated goat anti-rabbit IgG secondary antibody (S0018; 1:100; Affinity Biosciences) was added and incubated away from light at 37°C for 45 min. For nuclear staining, cells were rinsed with PBS and counterstained in a dark environment for 10 min at room temperature with 4′-6-diamidino-2-phenylindole (DAPI; #HXSJ-021134; Jisskang). Cells were observed under a confocal laser scanning microscope.
Western blottingAfter the extraction of total cellular and tissue proteins with radioimmunoprecipitation assay buffer (#LM-D8001; LMAl Bio, Shanghai, China), protein concentrations were examined using a bicinchoninic acid protein assay kit (#ZY80815; Zeye, Shanghai, China). Next, protein aliquots (10 µg) were separated by 10% SDS-PAGE and transferred onto PVDF membranes with a wet transfer system. Then, the membranes were blocked for 1 hr with 5% non-fat milk in TBST, followed by incubation overnight at 4°C with primary antibodies and for 1 hr at room temperature with HRP-conjugated secondary antibodies. Primary antibodies against ZO-1 (#AF5145; 1:1000), Occludin (#AF4605; 1:1000), Claudin-1 (#DF6919; 1:1000), AQP4 (#AF5164; 1:1000), MLCK (#AF5314; 1:1000), p-MLC2 (#AF8618; 1:1000), MLC2 (#DF7911; 1:1000), and GAPDH (#AF0911; 1:5000) purchased from Affinity Biosciences (Jiangsu, China) were used. In the end, protein bands were visualized with BeyoECL Moon Imaging Kit (#P0018FS; Beyotime), and the staining images were captured using ImageQuant LAS 4010 (GE Healthcare, Little Chalfont, UK). GAPDH was used as the internal control to normalize the relative expression of each protein. The band intensity was quantified using Quantity One 4.6.2 Software (Bio-Rad, California, USA).
Molecular dockingTo determine whether lenvatinib directly targets AQP4, the interaction between lenvatinib and the key core target was evaluated using the molecular docking method. The three-dimensional (3D) crystal structure of AQP4 was obtained from the Protein Data Bank (PDB; http://www.rcsb.org/pdb/). The two-dimensional (2D) SDF format of lenvatinib was downloaded from PubChem (http://pubchem.ncbi.nlm.nih.gov). Docking experiments were conducted using CDOCKER with receptor-ligand interactions. The bonding was assessed by measuring the -CDOCKER_INTERACTION_ENERGY value. The binding results were presented as 3D and 2D diagrams using Discovery Studio (DS) 2.5 (Accelrys Software Inc., San Diego, CA, USA.
Immunofluorescence stainingCaco-2 cells were treated with 100 μM biotin-labelled lenvatinib for 24 hr, followed by 30 min fixation with 4% paraformaldehyde, three washes with cold PBS, 10 min permeabilization with 0.2% Triton X-100 at 4°C, and 1 hr blocking with 10% goat serum at room temperature. The blocked cells were then incubated with an anti-AQP4 primary antibody (#ab259318; 1:200; Abcam, Shanghai, China) at 4°C overnight and with an Alexa Fluor 488-conjugated anti-rabbit IgG secondary antibody (#ab150077; 1:200; Abcam) at room temperature for 1 hr, followed by 10 min incubation in the dark at 37°C with DAPI. The images were obtained under a laser confocal microscope. The cells co-localized with biotin-labelled lenvatinib (red) and AQP4 (green) appeared orange.
Pull-down assayThe secondary amine (-NH-) in the cyclopropylcarbamoyl group of lenvatinib was covalently linked with NHS-PEG4-biotin (Themo Fisher Scientific) to generate biotinylated lenvatinib for the pull-down assay. Cell lysates of Caco-2 cells were harvested and incubated with biotin-labelled lenvatinib at 4°C overnight, followed by the addition of streptavidin magnetic beads (#ZY130521; Zeye) for 2 hr incubation at room temperature. Subsequently, the beads were washed thrice with PBS and then resuspended in SDS-containing loading buffer to elute the bound proteins. The eluted proteins were detected by western blotting using anti-AQP4 antibody.
Cellular thermal shift assay (CETSA)Caco-2 cell lysates were divided into two equal portions, one of which was used as the lenvatinib group and incubated with lenvatinib (15 µM) at room temperature for 2 hr, and the other was used as the control group and incubated with dimethyl sulfoxide (DMSO). The two cell lysates after incubation were divided equally into six tubes and were gradually heated from 46 to 56°C, with temperature increments of 2°C for 3 min each time. Afterwards, the tubes were incubated at room temperature for 3 min, after which the loading buffer was added to denature the protein at 95°C for 10 min. The expression of AQP4 was finally examined by western blotting.
Statistical analysisAll in vitro assays were conducted in triplicate. Data are expressed as the mean ± standard deviation, and statistical analyses were carried out with SPSS 23.0 (IBM, Armonk, NY) software. Two or multiple comparisons were done using unpaired Student's t-test or one-way ANOVA followed by the least significant difference (LSD) post hoc test. p < 0.05 indicated statistical significance.
Lenvatinib-induced diarrhea in rats was classed into four grades (0-3). Rats were administrated with 0.2 or 2 mg/kg lenvatinib for six consecutive days, and diarrhea in rats was recorded every day. As shown in Fig. 1A, rats in both dose groups began to develop diarrhea on day 3 after lenvatinib treatment. Almost all rats subjected to high-dose lenvatinib administration developed grade 3 diarrhea. After rats in each group were sacrificed, their colon tissues were dissected and stained by H&E to assess the histopathological changes. No erosion and ulcer formation were observed in the colon tissues of control rats, and the crypt structure was normal and relatively intact. Nevertheless, intestinal villi structure damage and obvious inflammatory cell infiltration were shown in the colon tissues of lenvatinib-administrated rats, with more severe pathological injury in the high-dose group than in the low-dose group (Fig. 1B). ELISA revealed that lenvatinib treatment led to a marked increment in the contents of D-lactate and DAO, biomarkers of intestinal injury, in the serum of rats (Fig. 1C-D). Besides, through western blotting, we discovered that the levels of tight junction proteins ZO-1, Occludin, and Claudin-1 in rat colon tissues were notably reduced following administration with lenvatinib (Fig. 1E).
Lenvatinib induces diarrhea and disrupts the intestinal mucosal barrier in rats. (A) Diarrhea incidence in rats after treatment with 0.2 or 2 mg/kg lenvatinib for six consecutive days. (B) Representative H&E staining images showing the histopathological changes in the colon tissues of rats. (C-D) Detection of D-lactate and DAO contents in the serum of rats by ELISA. (E) Measurement of ZO-1, Occludin, and Claudin-1 protein levels in rat colon tissues through western blotting. n=10. *p < 0.05, **p < 0.01, ***p < 0.001.
As for the in vitro assay, Caco-2 cells were treated with different concentrations of lenvatinib for 24 hr. As shown by CCK-8 assay, lenvatinib at concentrations higher than 5 μM significantly attenuated Caco-2 cell viability (Fig. 2A). Consistently, lenvatinib (5, 10, and 15 μM) treatment resulted in a dose-dependent increase in LDH release, indicating that lenvatinib induced Caco-2 cell death (Fig. 2B). Through RT-qPCR, we discovered that ZO-1, Occludin, and Claudin-1 mRNA levels in Caco-2 cells were markedly reduced after lenvatinib treatment (Fig. 2C). Western blotting revealed that lenvatinib treatment resulted in a notable decrement in ZO-1, Occludin, and Claudin-1 protein levels in Caco-2 cells (Fig. 2D-E). Immunofluorescence staining manifested that lenvatinib disturbed the periplasmic localization of ZO-1 and caused its cytoplasmic internalization (Fig. 2F). Next, TER was measured to evaluate the permeability of intestinal epithelial cells. The data indicated that the TER considerably dropped after stimulation by lenvatinib and reached the lowest on day 4 (Fig. 2G). As a paracellular marker, FITC-Dextran 40 kDa can be used to examine mucosal permeability. As indicated in Figure 2H, the paracellular permeability in lenvatinib-treated Caco-2 cells was elevated over time and peaked on day 4.
Lenvatinib triggers intestinal epithelial cell injury. (A) Detection of intestinal epithelial cell viability after 24 hr treatment with lenvatinib (0, 0.1, 0.5, 1, 2, 5, 10, and 15 μM) by CCK-8 assay. (B) Assessment of Caco-2 cell death after lenvatinib (5, 10, and 15 μM) treatment by LDH release assay. (C) Analysis of ZO-1, Occludin, and Claudin-1 mRNA levels in Caco-2 cells after lenvatinib stimulation by RT-qPCR. (D-E) Measurement of ZO-1, Occludin, and Claudin-1 protein levels in Caco-2 cells through western blotting. (F) Representative immunofluorescence staining images showing ZO-1 expression in Caco-2 cells treated by lenvatinib. (G-H) Assessment of the paracellular permeability of Caco-2 cells by measuring the transepithelial electrical resistance and using the fluorescent probe FITC-Dextran 40 kDa. *p < 0.05, **p < 0.01, ***p < 0.001.
To further explore the downstream molecular mechanism underlying the effects of lenvatinib on intestinal barrier injury, we analyzed the interaction between lenvatinib and AQP4 by molecular docking (Fig. 3A). 2D molecular docking data revealed that lenvatinib produced conventional hydrogen bonds with TRP231, ASN229, HIS230, TRP234, PHE218 and GLU228 of AQP4 and a Pi-Pi stacked interaction with TRP227 (Fig. 3B). Immunofluorescence experiments showed the co-localization of biotin-labelled lenvatinib and AQP4 in the cytoplasm (Fig. 3C). The pull-down assay in vitro revealed that lenvatinib-biotin could precipitate with AQP4 protein in Caco-2 cells, suggesting that lenvatinib has a stable interaction with AQP4 (Fig. 3D). Furthermore, the results of CETSA assay confirmed that the AQP4 protein disappeared as the temperature increased in the control group. However, a strong AQP4 band was detected even with increasing temperature in the lenvatinib group, which indicated that the combination of lenvatinib and AQP4 could promote the thermal stability of AQP4 in Caco-2 cells (Fig. 3E). These data clearly demonstrate that lenvatinib directly binds with AQP4. Through western blotting, we found that AQP4 protein levels were prominently downregulated in rat colon tissues and Caco-2 cells after lenvatinib treatment (Fig. 3F-I).
Lenvatinib targets AQP4 and inhibits its expression. (A) Three-dimensional docking of lenvatinib and AQP4. (B) Two-dimensional docking of lenvatinib and AQP4. (C) Immunofluorescence staining images showing the co-localization of lenvatinib-biotin with AQP4 in Caco-2 cells. (D) Pull-down assays confirmed that biotin-labelled lenvatinib could precipitate with AQP4 in Caco-2 cell lysis. (E) Cellular thermal shift assays and western blotting were used to assess the binding of lenvatinib and AQP4 in Caco-2 cells (F-I) AQP4 protein levels in colon tissues of rats (n=10/group) and in Caco-2 cells after lenvatinib treatment were determined by western blotting. *p < 0.05, ***p < 0.001.
Finally, a series of rescue experiments were conducted to verify that lenvatinib promotes intestinal barrier injury through targeting AQP4. AQP4 mRNA and protein levels in Caco-2 cells were considerably elevated after transfection with AQP4 overexpression vector OE-AQP4 (Fig. 4A-B). CCK-8 assay illustrated that AQP4 overexpression rescued lenvatinib-triggered decline in Caco-2 cell viability (Fig. 4C). AQP4 overexpression also significantly reduced the release of LDH from Caco-2 cells caused by lenvatinib treatment (Fig. 4D). Lenvatinib-induced reduction in ZO-1, Occludin, and Claudin-1 mRNA and protein expression was overturned by overexpressing AQP4 (Fig. 4E-G). In addition, lenvatinib stimulation resulted in a distinct decrement in the TER value and an increment in the paracellular flux of FITC-Dextran 40 kDa in Caco-2 cells, which however, were reversed after overexpression of AQP4 (Fig. 4H-I), indicating that AQP4 upregulation alleviated the enhanced paracellular permeability in lenvatinib-stimulated Caco-2 cells.
Overexpressing AQP4 debilitates lenvatinib-induced intestinal barrier injury. Caco-2 cells were transfected with OE-NC or OE-AQP4 for 24 hr and then treated with lenvatinib (15 μM) for 24 hr. (A-B) Estimation of AQP4 mRNA and protein levels in Caco-2 cells through RT-qPCR and western blotting. (C) Evaluation of Caco-2 cell viability by CCK-8 assay. (D) Detection of Caco-2 cell death by LDH release assay. (E-G) Assessment of ZO-1, Occludin, and Claudin-1 mRNA and protein levels in Caco-2 cells via RT-qPCR and western blotting. (H-I) Determination of the paracellular permeability of Caco-2 cells by measuring the transepithelial electrical resistance (TER) and using the fluorescent probe FITC-Dextran 40 kDa. ***p < 0.001 versus Control; #p < 0.05, ##p < 0.01, ###p < 0.001 versus Lenvatinib + OE-NC.
Finally, to ascertain the mechanism by which AQP4 overexpression mitigates lenvatinib-induced intestinal barrier injury, we detected the involvement of the MLCK/p-MLC2 signaling pathway in this process. As revealed in Figure 5A-B, lenvatinib treatment markedly enhanced the protein levels of MLCK and phosphorylated-MLC2 in rat colon tissues. Consistently, MLCK and phosphorylated-MLC2 protein levels in Caco-2 cells were also markedly elevated after lenvatinib treatment. AQP4 overexpression reversed lenvatinib-induced increase in MLCK and phosphorylated-MLC2 protein levels, which however, was reversed after overexpression of MLCK (Fig. 5C-D). Furthermore, MLCK overexpression antagonized the effects of AQP4 overexpression on the protein levels of ZO-1, Occludin, and Claudin-1 as well as the paracellular permeability in lenvatinib-stimulated Caco-2 cells (Fig. 5E-H), suggesting that AQP4 mitigates lenvatinib-induced intestinal barrier injury through inactivating the MLCK/p-MLC2 pathway.
AQP4 ameliorates lenvatinib-induced intestinal barrier injury by inhibiting the MLCK/p-MLC2 signaling pathway. (A-B) Examination of MLCK, MLC2, and p-MLC2 protein levels in the colon tissues of lenvatinib-induced diarrhea rats through western blotting. n=10. ***p < 0.001. (C-D) Caco-2 cells were transfected with OE-NC, OE-AQP4, or OE-AQP4 + OE-MLCK for 24 hr and then treated with lenvatinib (15 μM) for 24 hr. MLCK, MLC2, and p-MLC2 protein levels in Caco-2 cells were measured by western blotting. ***p < 0.001 versus Control; ###p < 0.001 versus Lenvatinib + OE-NC; &&&p < 0.001 versus Lenvatinib + OE-AQP4. (E-F) Measurement of ZO-1, Occludin, and Claudin-1 protein levels in Caco-2 cells via western blotting. (G-H) Assessment of the paracellular permeability of Caco-2 cells by measuring the transepithelial electrical resistance (TER) and using the fluorescent probe FITC-Dextran 40 kDa. **p < 0.01, ***p < 0.001 versus Lenvatinib; #p < 0.05, ##p < 0.01, ###p < 0.001 versus Lenvatinib + OE-AQP4.
Tyrosine kinases play critical roles in tumorigenesis and progression, and the development of molecularly targeted drugs oriented to them has become one of the focuses of modern antitumor drug research (Kumar et al., 2023). Lenvatinib is an oral multi-targeted tyrosine kinase inhibitor capable of achieving tumor ablation by inhibiting tumorigenesis and angiogenesis (Suyama and Iwase, 2018). Diarrhea is one of the common digestive side effects of lenvatinib (Kudo et al., 2018). Maintaining the intestinal epithelial cell integrity and controlling permeability between neighboring epithelial cells are the minimum requirements to ensure the proper functioning of the intestinal barrier (Camilleri et al., 2012). Accordingly, the induction of lenvatinib on diarrhea might be attributed to the disruption of the intestinal barrier. In our study, it was discovered that lenvatinib treatment not only damaged the intestinal mucosal barrier in rats but also caused intestinal epithelial barrier injury in Caco-2 cells. Additionally, the detrimental effects of lenvatinib in Caco-2 cells were confirmed to be associated with its inhibition of AQP4 expression.
The normal intestinal tract has a well-functioning isolation zone to prevent the invasion of pathogenic antigens (bacteria, toxic substances, food antigens, carcinogens, etc.) in the intestinal lumen, which is known as the barrier function of the intestinal mucosa (Khoshbin and Camilleri, 2020). The intestinal mucosal barrier, which is composed of intact intestinal epithelial cells and the connections between neighboring intestinal epithelial cells, is the most important barrier in the intestinal tract (Rohr et al., 2020). Tight junctions are the most significant connections between neighboring epithelial cells and play a pivotal role in the maintenance of the intestinal barrier (Suzuki, 2020). Once the tight junctions are damaged, the permeability between the intestinal epithelial cells increases, and bacteria, endotoxins, and macromolecules can enter the circulation through the tight junctions, which is associated with the development of many gastrointestinal diseases, including diarrhea (Salvo Romero et al., 2015). Transmembrane proteins (e.g., occludin and claudin) and cytoplasmic proteins (e.g., ZO-1) are involved in the formation of tight junctions (Buckley and Turner, 2018). Several studies have shown that intestinal barrier damage and changes in the expression of related markers are seen in mouse models of diarrhea. For example, Guan et al. disclosed that the serum contents of D-LA and DAO were upregulated while the colon levels of ZO-1, occludin, and claudin-1 were downregulated in mice with Folium Sennae-induced diarrhea (Guan et al., 2021). Fan et al. suggested that occludin and ZO-1 mRNA and protein levels were markedly lower in the intestinal samples of rats with post-weaning diarrhea than in those of control rats (Fan et al., 2017). Herein, our results showed that the serum contents of D-LA and DAO were elevated and the colon levels of ZO-1, occludin, and claudin-1 were reduced in lenvatinib-induced diarrhea rat models. For in vitro assays, lenvatinib also downregulated ZO-1, occludin, and claudin-1 mRNA and protein levels in Caco-2 cells. In addition, TER and FITC-Dextran 40 flux were used to confirm lenvatinib-triggered increase in Caco-2 monolayer permeability. Therefore, lenvatinib caused intestinal barrier injury in both diarrhea rats and intestinal epithelial cells.
The family of membrane proteins AQPs efficiently and selectively transport water and/or small molecule solutes to hydrophobic cell membranes through osmotic gradients (Zhao et al., 2023). AQPs are widely distributed in various cells and tissues in mammals, including gastrointestinal secretory and absorptive epithelial cells that participate in transport and translocation processes (Liao et al., 2020). Among the AQPs family, AQP4 is mainly localized in the basolateral membrane of small intestinal crypt cells and colon surface epithelial cells, and is responsible for regulating intestinal water transport (Koyama et al., 1999). Imbalance of intestinal water secretion and/or water absorption is a significant factor in diarrhea development (Keely and Barrett, 2022). Wang et al. discovered that in mice with AQP4 gene knockout, colonic water absorption decreased and fecal water increased significantly, which indicated that colonic AQP4 is involved in colonic water absorption in the intestinal lumen under physiological conditions. Sakai et al. indicated that treatment with the chemotherapy drug 5-fluorouracil resulted in body weight loss and diarrhea as well as decreased AQP4 and AQP8 levels in the colons of mice (Sakai et al., 2014). Cao et al. clarified that AQP3 protein expression was upregulated while AQP1, AQP4, and AQP8 protein levels were downregulated in colon tissues of murine models of rotavirus diarrhea (Cao et al., 2014). Huang et al. revealed that AQP4 levels were reduced in rotavirus-infected Caco-2 cells, while genistein could regulate the cAMP/PKA/CREB signaling pathway to promote AQP4 gene transcription and inhibit rotavirus replication (Huang et al., 2015). Yamamoto et al. reported that a marked decrement in AQP4 and AQP8 mRNA and protein levels was observed in the proximal colon of mice with allergic diarrhea (Yamamoto et al., 2007). In our study, through molecular docking, we observed the interaction between lenvatinib and AQP4. Immunofluorescence co-localization analysis, pull-down assay, and cellular thermal shift assay also verified the binding between lenvatinib and AQP4. Subsequently, lenvatinib was identified as a negative regulator of AQP4, which reduced AQP4 protein levels in rat colon tissues and Caco-2 cells. Importantly, rescue experiments verified that overexpression of AQP4 antagonized lenvatinib-induced reduction in ZO-1, Occludin, and Claudin-1 levels and enhancement in paracellular permeability in Caco-2 cells, suggesting that lenvatinib promotes intestinal barrier injury by downregulating AQP4.
The MLCK/p-MLC2 signaling pathway has been demonstrated to play an essential role in regulating tight junction permeability in the intestinal epithelium and thereby influencing barrier function (Huang et al., 2020b). MLCK-driven MLC2 phosphorylation regulates cellular actin-myosin contractions, which is a critical step in maintaining barrier integrity by opening paracellular pathways (Zhao et al., 2021). Besides, it has been found that diverse extracellular stimuli can regulate MLCK-driven MLC2 phosphorylation, thus altering the expression and localization of tight junction proteins (Wang et al., 2025). MLCK expression and MLC phosphorylation are increased in the intestinal epithelium of patients with inflammatory bowel disease (Blair et al., 2006). Previously, multiple studies have revealed that inhibition of the MLCK/p-MLC2 pathway is beneficial for decreasing intestinal permeability and preventing intestinal barrier damage in several gastrointestinal disorders. For example, Dong et al. reported that the natural bioactive monosaccharide mannose attenuated intestinal barrier damage in two mouse colitis models through preventing MLCK-induced tight junction disruption (Dong et al., 2022). Du et al. clarified that suppressing the MLCK/p-MLC2 signaling pathway mitigated heat stroke-induced intestinal mucosal barrier injury in rats by increasing the expression of tight junction proteins and reducing intestinal permeability (Du et al., 2023). Xie et al. suggested that Atractylodes lanceolata oil protected against intestinal barrier damage and ameliorated the symptoms of diarrhea in rat models of diarrhea-predominant irritable bowel syndrome via downregulating the levels of MLCK and inhibiting the phosphorylation of MLC2 (Xie et al., 2021). In the present study, we also assessed whether AQP4 affects lenvatinib-induced intestinal barrier injury through regulation of the MLCK/p-MLC2 signaling pathway. Lenvatinib treatment was observed to enhance MLCK expression and MLC2 phosphorylation in colon tissues of rats and Caco-2 cells. However, overexpression of AQP4 overturned lenvatinib-induced activation of the MLCK/p-MLC2 pathway. Importantly, rescue experiments demonstrated that MLCK overexpression further abrogated the effects of AQP4 overexpression on tight junction protein expression and paracellular permeability in lenvatinib-exposed Caco-2 cells.
To be honest, our study also has several limitations. First, due to the lack of high-resolution crystal structures or homology models of AQP4, molecular docking cannot accurately predict the binding sites. Besides, our laboratory conditions restrict the implementation of accurate molecular docking, and we fail to conduct high-precision molecular docking simulations. We are committed to improving our laboratory equipment in future work. Second, we only examined pathological changes in the colon tissues of lenvatinib-treated rats, but did not evaluate small intestinal pathological changes. Third, even though our results confirmed that the intestinal toxicity of lenvatinib is associated with AQP4 downregulation. Whether similar toxicity can be observed with other multi-kinase inhibitors remains unclear, which requires further investigation in the future.
To sum up, our research demonstrated that lenvatinib induces diarrhea and damages the intestinal mucosal barrier in rats. Besides, lenvatinib promotes intestinal barrier injury in human intestinal epithelial cells by decreasing AQP4 expression. AQP4 overexpression ameliorates lenvatinib-induced intestinal barrier injury by inhibiting the MLCK/p-MLC2 signaling pathway. Therefore, AQP4 might be a promising potential target for treating lenvatinib-induced intestinal barrier injury and diarrhea.
Conflict of interestThe authors declare that there is no conflict of interest.
