The component herbs of LJAW used in this study are as follows: Pseudostellariae Radix, Poria, Macrocephalae Rhizoma, Radix Rhizoma Glycyrrhizae, Pinelliae rhizome, Citrus reticulata Blanco, Galli Gigerii Endothelium Corneum, Massa Medicata Fermentata, Setariae Fructus Germinatus, Hordei Fructus Germinatus, Crataegi Fructus, Amomi Fructus, and Aucklandiae Radix at a weight ratio of 15:5:5:3:5:5:5:5:5:5:3:3. The LJAW used in the study were purchased from Beijing Tcmages Pharmaceutical Co., Ltd. (Beijing, China). Prior to each experiment, the formula LJAW were milled to powders and dissolved in filtered deionized water. The concentrations of LJAW in the study refer to the crude drug concentrations.

Eight-week-old female Wistar rats weighing 180 ± 20 g was used. The animal facility was kept at 23°C and 10% humidity, with a cycle of 12 hours of light and 12 hours of dark. The rats were acclimated for 1 week in the facility prior to the experiment.

Pica (eating nonnutritive substances such as kaolin) in rats, analogous to vomiting in animals that developed emetic reflex, is regarded as an alternative model for nausea and vomiting in rodents (Sharma et al., 1997). To determine the efficacy of LJAW in treating CINV, rats were injected with cisplatin intraperitoneally at a dose of 6 mg.kg^-1 (cisplatin-induced pica) and then randomly assigned to DDP, LJAW and ondansetron. The rats were gavage with distilled water (10 ml.kg^-1), LJAW (12.78 g.kg^-1) or ondansetron (2.6 mg.kg^-1) for 7 days. Another group of rats were set as vehicle-control without any treatment. The dose of LJAW is equal to that used in humans in the clinic. The dose of LJAW was choose based on the results of our preliminary experiment. Briefly, we treated rats with high-, medium- and low-dose of LJAW. Clinically relevant dose was set as medium-dose, 2-times clinically relevant dose was set as high-dose, and 0.5-time of clinically relevant dose was set as low-dose. Preliminary data showed that medium dose of LJAW exerted the best therapeutic effect. Kaolin and chow intake as well as body weight were monitored every day. Feces were collected prior to cisplatin administration, 3 and 7 days after cisplatin injection. The rats were euthanized, and the intestine and medulla oblongata were resected 3 and 7 days after cisplatin injection, respectively. The samples of jejunum, medulla oblongata and feces were flash frozen in liquid nitrogen and stored at -80°C. Part of the jejunum tissues was fixed in a 10% formalin-PBS solution for hematoxylin-eosin staining (HE) and immunohistochemistry staining (IHC).

Both intestine and medulla oblongata tissues were lysed in ice-cold lysis buffer (Thermo Fisher Scientific, Waltham, MA) and homogenized with a tissue homogenizer (FLUKO Shanghai Equipment, Shanghai) followed by centrifugation at 10,000 g for 10 minutes at 4°C. Protein levels were quantified using a bicinchoninic acid (BCA) protein assay kit (cwbiotech, Beijing). An equal amount of protein (20 µg) was applied to a 10% to 15% SDS gel and then transferred onto polyvinyl membranes according to standard procedure. We blocked the membranes with 5% nonfat dry milk blocking buffer prepared in Tris-buffered saline with 0.1% Tween 20 for 1 hour at room temperature. The membranes were then probed with primary antibodies against NF-κB p65 (Cell Signaling Technology, Danvers, MA), mdy88 (Cell Signaling Technology, Danvers, MA), TLR4(Cell Signaling Technology, Danvers, MA), TPH1(Cell Signaling Technology, Danvers, MA) or beta-actin (Cell Signaling Technology, Danvers, MA) overnight at 4°C. The membranes were washed and incubated with secondary antibodies (anti-rabbit IgG) prepared in 5% nonfat dry milk blocking buffer with 0.1% Tween 20 for 1 hour at room temperature. The membranes were then washed again and incubated with the ECL+ detection kit for 5 minutes. Then, lights were turned off, and the protected membrane was exposed to the X-ray film. The exposure time varied according to the bands. Gel Imagestem ver. 4.00 (Tanon, Beijing) was used for protein band quantification.

5-HT and cytokine levels in jejunum and medulla oblongata tissues were measured by Elisa assay kit. About 30mg of tissue for each sample was taken from the above-mentioned sites after their dissection. Samples were rinsed with ice-cold PBS and dried with filter paper to remove feces or blood and weighted. Tissues was homogenized in PBS(1:10) by tissue homogenizer and followed by sonicating by using 2 cycles of 30s each to break cell membranes. Then, homogenates were centrifugated for 15 min at 5000rpm and the supernatants were collected and used for Elisa analysis.

5-HT(Cambridge, MA) and cytokines levels including IL-6, IL-1β, IL-10, TNF-α and TGF-β(Elabscience, Wuhan) were measured by ELISA according to the manufacturer’s instructions. Then, the concentration of the samples was calculated according to the OD value.

Formalin-fixed, paraffin-processed jejunum tissues were used for biomarker identification using IHC staining. The slides were baked at 60°C for over 2 hours and then deparaffinized and rehydrated. The antigens were unmasked by heat-induced antigen retrieval. The slides were then immersed in 3% H₂O₂-methanol solution followed by blocking with 5% goat serum in 0.3% Triton X-100 PBS. The slides were stained with ZO-1(proteintech, Wuhan) and occludin(proteintech, Wuhan) antibodies in a humidified chamber overnight at 4°C, washed three times with PBS and further incubated with secondary antibody at room temperature for 45 minutes. The slides were then incubated with avidin-biotin complex (Vector Laboratories, Burlingame, CA) followed by DAB substrate for antibody visualization and counterstained with Mayer’s hematoxylin, dehydrated, and mounted with ClearMount mounting medium (American MasterTech, Lodi, CA) (Pan et al., 2015).

16S rRNA sequencing was carried out by Beijing Guoke Biotechnology. Total genome DNA from samples was extracted using CTAB/SDS method.16S rRNA genes were amplified used the specific primer with the barcode. Then, mix same volume of 1X loading buffer (contained SYB green) with PCR products and operate electrophoresis on 2% agarose gel for detection. Samples with bright main strip between 400-450bp were chosen for further experiments. Then, mixture PCR products was purified with GeneJET Gel Extraction Kit (Thermo Scientific). Sequencing libraries were generated using Illumina TruSeq DNA PCR-Free Library Preparation Kit (Illumina, USA) following manufacturer’s recommendations and index codes were added. The library quality was assessed on the Qubit@ 2.0 Fluorometer (Thermo Scientific) and Agilent Bioanalyzer 2100 system. At last, the library was sequenced on an Illumina HiSeq platform and 250 bp paired-end reads were generated.

Raw sequence data were analyzed using the QIIME 2 pipeline (https://qiime2.org/) (version 2020.2). Raw sequence data were denoised using DADA2 in the QIIME 2 package. The paired-end FASTQ files were processed by Phred quality score-based quality filtering, merging of the paired ends, chimera removal, singleton removal, and construction of a feature table consisting of amplicon sequence variants (ASV). Using DADA2 denoise-single method, we removed low quality regions of the sequences. All ASV were aligned to mafft through q2 alignment, and phylogenetic analysis was performed using fasttree2. A scikit-learn naïve Bayes machine-learning taxonomy classifier against the SILVA (https://www.arb-silva.de/) 16S rRNA Version 138 reference sequences was trained with the q2-feature-classifier plugin using the V3-V4 regions of 16S rRNA sequences. Taxonomy was assigned to each ASV using the ‘classifysklearn’ command in q2-feature-classifier. The α-diversity was calculated between groups using the Kruskal-Wallis pairwise test and permutational multivariate analysis of variance, respectively. In order to compute α-diversity, we rarify the ASV table and calculate Chao1 to estimates the species abundance. We used unweighted unifrac for Principal Coordinate Analysis (PCoA). To mine deeper data of microbial diversity of the differences between the samples, significance test was conducted with some statistical analysis methods, including MetaStat, LEfSe, Anosim and MRPP.

Data analysis for 16S rRNA amplicon sequencing has been described in preceding paragraphs. Other data are analyzed using GraphPad 7.0 software. If the data conformed to a normal distribution, one-way ANOVA was used to compare the means of multiple samples. The LSD method was used to compare the groups with homogeneous variances, and the Dunnett’s T3 method was used to compare the data between groups with unequal variances. Non-parametric test such as Kruskal-Wallis test was used for non-normal distribution data, p<0.05 was considered statistically significant.

All rats showed pica behavior after cisplatin injection. We measured kaolin intake, food intake and body weight every 24 hours. Cisplatin induced substantial increased kaolin intake ranging from 0.53g to 1.09g during the first 24 hours after chemotherapy compared with vehicle control(saline) injected rats (cisplatin vs. vehicle-control, 0.78 ± 0.032g vs. 0.056 ± 0.001g, p<0.01). LJAW and ondansetron significantly reduced cisplatin-induced kaolin intake on the first day after cisplatin injection compared with cisplatin treated rats(cisplatin, LJAW and ondansetron, 0.78 ± 0.032g, 0.61 ± 0.22g and 0.42 ± 0.034g, respectively; LJAW vs. cisplatin, p<0.01; LJAW vs. ondansetron, p<0.05, ondansetron vs. cisplatin, p<0.01). Kaolin consumption was significantly decreased on the second day in LJAW and ondansetron treated mice compared with cisplatin-only treated rats, but no significant difference was observed in LJAW and ondansetron treated rats (cisplatin, LJAW and ondansetron, 0.65 ± 0.16g, 0.49 ± 0.12g and 0.44 ± 0.13g, respectively; LJAW vs. cisplatin, p<0.05; LJAW vs. ondansetron, p>0.05, ondansetron vs. cisplatin, p<0.05). On the third day, the kaolin intake was 0.53 ± 0.27g, 0.34 ± 0.11g and 0.38 ± 0.15g in cisplatin-, LJAW- and ondansetron-treated rats (LJAW vs. cisplatin, p<0.05; LJAW vs. ondansetron, p>0.05; ondansetron vs. cisplatin, p<0.05) ( Figure 1A ). Moreover, we observed that rats in the cisplatin group consumed much less chow within 72 hours after cisplatin injection compared with the Control rats (24hours, cisplatin vs. control, 9.23 ± 3.77g vs. 15.63 ± 3.32g, p<0.05; 48hours, cisplatin vs. control, 9.23 ± 3.77g vs. 15.63 ± 3.32g, p<0.05; 72hours, cisplatin vs. control, 9.23 ± 3.77g vs. 15.63 ± 3.32g, p<0.05). Interestingly, LJAW significantly increased the food intake on the first and third day after chemotherapy compared with cisplatin-only treated rats(chow consumption on the first day(cisplatin, LJAW and ondansetron), 9.23 ± 3.77g, 13.94 ± 4.37g, 11.94 ± 4.3g, LJAW vs. cisplatin, p<0.05; LJAW vs. ondansetron, p>0.05;ondansetron vs. cisplatin, p>0.05; chow consumption on the second day(cisplatin, LJAW and ondansetron), 10.60 ± 2.67g, 12.56 ± 3.21g, 11.88 ± 3.80g, LJAW vs. cisplatin, p>0.05; LJAW vs. ondansetron, p>0.05; ondansetron vs. cisplatin, p>0.05; chow consumption on the third day(cisplatin, LJAW and ondansetron), 11.30 ± 3.20g, 15.92 ± 2.98g, 10.72 ± 3.21g, LJAW vs. cisplatin, p<0.05; LJAW vs. ondansetron, p<0.05; ondansetron vs. cisplatin, p>0.05). However, increased food intake was not observed in the ondansetron-treated rats compared with cisplatin only treated rats ( Figure 1B ). Similarly, body weight of cisplatin or ondansetron treated rats decreased dramatically 24 hours after injection of cisplatin. Body weight of all rats started to recover 72 hrs. after the dosage of cisplatin. However, body weight loss was not present in LJAW-treated rats compared with rats in the Control group ( Figure 1C ).

Microbiome profiling was performed to explore composition of intestinal microbiome in cisplatin- and LJAW- treated rats before (Day 0), 3 days and 7 days after cisplatin injection. First, we compared the gut microbiota diversity among groups using the Chao index. A significant decline in alpha diversity was observed in the feces collected 3 or 7 days after cisplatin injection compared with the feces collected 3 or 7 days after LJAW treatment ( Figure 2C ). Then, principal coordinates analysis was performed to explore similarities between these two datasets. A notable clustering effect by cisplatin injection was exerted in the intestinal microbiome of rats ( Figure 2D ). However, no clustering effect was observed in samples collected after 7 days’ treatment of LJAW compared with samples collected prior treatment. Next, we sought to further examine the composition changes of microbiome induced by cisplatin as well as LJAW. Results showed that all rats were dominated by Bacteroidetes and Firmicutes at the phylum level. However, the abundance of Bacteroidetes at the phylum level was increased 3 days after cisplatin injection compared to the abundances in the samples collected before injection ( Figure 2A ). At the genus level, the bacterial composition in feces collected before cisplatin injection was dominated by Enterococcus, Muribaculaceae, and Lactobacillus but devoid of Escherichia-Shigella ( Figure 2B ). Cisplatin injection caused an elevated abundance of Enterococcus and Escherichia-Shigella on both the 3^rd and 7^th day after cisplatin injection, but decreased abundance of Lactobacillus ( Figures 2B ). By contrast, the intestine microbiome was dominated by ASVs belong to Muribaculaceae, Lactobacillus, Blautia and Bacteroides, and fewer ASVs belonging to the Enterococcus and Escherichia-Shigella after 3- or 7-days’ treatment of LJAW ( Figures 2B ). To further investigate these findings, we conducted high-dimensional class comparisons via linear discriminant analysis of effect size (LEfSe), which again validated differentially abundant bacteria in the fecal microbiome of cisplatin- and LJAW-treated rats, with Escherichia-Shigella, Parabacteroides and Enterococcus genera enriched in DDP-7 and Blautia and Lactobacillus enriched in Tx-7 ( Figure 2E ).

Escherichia-Shigella was well documented as harmful bacteria which involved in the bowel inflammation and carcinogenesis of intestine (Lang et al., 2020; Fan et al., 2021). Lactobacillus was identified as probiotics and was prescribed to treat inflammatory bowel diseases ( Wu et al., 2020 ). Consistently, Blautia could also exert an anti-inflammatory function in the intestine and has been named a new functional genus with potential probiotic properties (Davrandi et al., 2021; Liu et al., 2021). Moreover, an experimental survey mimicking autism syndrome reported a 50% reduction in 5-HT in both small and large intestine mucosal 5-HT levels with a certain correlation to the abundance of Blautia, indicating that Blautia might participate in the synthesis or release of 5-HT (Agus et al., 2018). Here, we evaluated both the 5-HT level and the expression of TPH1, a rate-limiting enzyme of 5-HT, and the levels of proinflammatory cytokines and the expression of possible molecules involved in microbe-mediated inflammation.

It has been reported that a variety of neurotransmitters and their receptors are involved in the development of CINV, and serotonin (also named 5-hydroxytryptamine) plays a key role in the pathogenic process of CINV (Marx et al., 2017). Cytotoxic medicine such as cisplatin could cause the release of 5-HT in enterochromaffin cells in the intestinal mucosal barrier, the elevated 5-HT that is released stimulates the 5-HT receptors in both the central and peripheral nervous systems and regulates the emetic pathways (Zhong et al., 2017).

To elucidate the possible mechanisms involved in LJAW-elicited anti-CINV efficacy, we measured the concentration of 5-HT in both rat jejunum and medulla oblongata tissues. As shown in Figure 3A , the level of 5-HT significantly increased in both jejunum and medulla oblongata tissues 3 days post-cisplatin injection compared with vehicle-control (5-HT levels in rats jejunum (Control, DDP and TX), 541 ± 59ng/mg, 835 ± 48 ng/mg and 687 ± 67 ng/mg, Control vs. DDP, p<0.01; Control vs. Tx, p>0.05; DDP vs. Tx, p<0.01. 5-HT in rats medulla oblongata (Control, DDP and TX), 270 ± 9ng/mg, 260 ± 9ng/mg and 170 ± 6ng/mg, Control vs. DDP, p<0.01; Control vs. Tx, p<0.05; DDP vs. Tx, p<0.01). The 5-HT levels were also significantly higher in the LJAW-treated rats than in the vehicle-control treated rats. However, it was still much lower compared with that of DDP-3 in rat jejunum tissue ( Figures 3A, B ). Then, we measured the expression of TPH1, a key enzyme participating in the synthesis of 5-HT in jejunum tissues, using western blotting. TPH1 expression was slightly upregulated 3 days after cisplatin injection. Conversely, it was downregulated in LJAW-treated rats ( Figure 3C ).

As most 5-HT-producing enterochromaffin cells reside in the jejunum, necrosis of enterochromaffin induced by cisplatin could also promote the release of 5-HT (Linan-Rico et al., 2016). We evaluated the integrity of the jejunum mucosa with HE staining and measured the expression of the tight junction-associated proteins occludin and ZO-1 in LJAW-treated rat jejunum tissue by IHC.

The HE staining showed that there was no disruption of the jejunal mucosa before cisplatin injection. Additionally, the morphology of the jejunal glands and villi was complete. Furthermore, the epithelial cells were neatly arranged, and cell degeneration and cell necrosis were not observed, as shown in Figure 4A(a) . Disrupted mucosa and shortened, sparse, or even lost villi were observed in the jejunum tissues of rats treated 7 days after cisplatin injection (Figure 4A(b) . As shown in Figure 4A(c) , the villi were normal in shape, and attenuated mucosa disruption was observed in LJAW treated rat’s intestine tissues.

Then, we sought to determine whether the intestinal mucosa barrier protective effect of LJAW was mediated by tight junction-associated proteins. We measured the expression of both occludin and ZO-1 in rat jejunum tissue through IHC staining and western blotting. Elevated occludin and ZO-1 protein expression was observed in the LJAW-treated rats compared with the cisplatin-treated rats ( Figures 4B–E ).

Bowel inflammation has already been validated in cisplatin-treated rat intestine tissues (Perales-Puchalt et al., 2018). In our study, we found that Escherichia-Shigella, and Enterococcus genera was enriched in rat feces collected 7 days after cisplatin injection, while Blautia and Lactobacillus genera was enriched in fecal samples collected 7 days after LJAW dosage, as shown in Figure 2E . Thus, ongoing inflammatory processes might be partially blocked or alleviated in LJAW-treated rat intestines. To further verify our hypothesis, we measured the pro-inflammatory cytokines and mucosa repair-related cytokines or chemokines in the rat jejunum with ELISA kits. The results showed that IL-6 and TNF-α were significantly elevated 72 hours after cisplatin injection; however, they were dramatically downregulated in LJAW-treated rats ( Figure 5D ). Then, to confirm whether the downregulated inflammation level in LJAW-treated rats was mediated by intestinal microbes through signaling to NF-κB by Toll-like receptors (TLRs), we performed western blotting to measure the protein expressions of TLR4, NF-κB and myd88 in rat jejunum tissues. As speculated, cisplatin induced upregulated protein expression of TLR4, myd88 and NF-κB compared with the control rats ( Figures 5A–C ). LJAW downregulated the expression of TLR4, myd88 and NF-κB in jejunum tissues compared with DDP treated rats.