Synergistic Sorafenib with Jianpi Huayu Decoction Resulted in Tumor Regression and Prevented Adverse Events in Hepatocellular Carcinoma by Remodeling the Gut Microbiota


 Background: There is an urgent need for effective treatments for hepatocellular carcinoma (HCC). Sorafenib is first-line treatment for HCC, which has a modest efficacy due to severe adverse effects(AE), acquired resistance and others. Combination therapy may be able to overcome this limitation. Jianpi Huayu Decoction (JHD) is a traditional Chinese medicine formulation, which has been shown to be effective as an alternative and complementary therapy of HCC. We investigated the synergistic effect of JHD with sorafenib by a xenograft model.Methods: Growth of mouse-derived HCC cells and adverse events(AE) of treatment were evaluated in a subcutaneous model with JHD and clinical-dose sorafenib combination treatment. Diarrhea, the most frequently reported AE, was evaluated by diarrhea score. The gut microbiota(GM) composition of the mice was analyzed by Illumina NovaSeq. Results: JHD administration in mice synergistically enhanced the anti-tumor response, thereby suppressing HCC, and prevented occurrence of the most common AEs(diarrhea and body weight loss) of sorafenib. Sorafenib induced increased proinflammatory GM(Helicobacter) which promoted the progression of HCC and against anti-tumor treatment. JHD reduced the abundance of anti-inflammatory microbiota Muribaculum, Fusicatenibacter, and Dorea. Following the modulation of GM, the proinflammatory signaling interleukin 6/signal transducer and activator of transcription-3 pathway was downregulated by JHD.Conclusions: Our finding suggested that differences in the microbial gut flora may modulate resistance to sorafenib through IL-6/STAT3 signaling. JHD with microbiota modulation properties could potentiate sorafenib and provided a promising approach for HCC treatment.

The study protocol was approved by the Animal Care and Use Committee of Southern Medical University (Guangzhou, China).

Gut microbiota analysis
Mice fecal samples were collected(23 days after implantation)and stored at − 80°C immediately. Total genome DNA from samples was extracted using CTAB-SDS method. 16S rRNA genes were ampli ed used the speci c primer with the barcode. PCR products was puri ed with GeneJET Gel Extraction Kit(Thermo Scienti c). Sequencing libraries were established using Illumina TruSeq DNA PCR-Free Library Preparation Kit (Illumina, USA) following manufacturer's instructions and index codes were added. The library quality was assessed on the Qubit@ 2.0 Fluorometer (Thermo Scienti c) and Agilent Bioanalyzer 2100 system. At last, the library was sequenced on an Illumina NovaSeq platform and 250 bp paired-end reads were generated.
Paired-end reads from the original DNA fragments were merged by using FLASH [11], and was assigned to each sample according to the unique barcodes. Sequences were analyzed using QIIME [12] software package (Quantitative Insights Into Microbial Ecology), and in-house Perl scripts were used to analyze alpha and beta diversity. First, reads were ltered by QIIME quality lters. Then we used pick_de_novo_otus.py to pick operational taxonomic units (OTUs) by making OTU table. Sequences with ≥ 97% similarity were assigned to the same OTUs. We picked representative sequences for each OTU and used the RDP classi er [13] to annotate taxonomic information for each representative sequence. We rari ed the OTU table and calculated three metrics Chao1 and Shannon index. We used weighted unifrac, calculated by QIIME, for principal coordinate analysis (PCoA). Signi cance test was conducted with some statistical analysis methods, including T-test, similarity percentages breakdown (SIMPER) and linear discriminant analysis effect size(LEfSe).

Western-Blot analysis and qRT-PCR
Total protein and RNA was extracted using extraction kit following the manufacturer's instructions. Protein was quanti ed, separated, transferred and blocked as previous description. The membranes were incubated with primary antibodies (β-actin, a nity 1 : 5000; STAT3, CST 1 : 1000; pSTAT3, CST 1 : 2000; iNOS, CST 1 : 2000). Protein bands were quanti ed using ImageJ software with β-actin as the internal control. The expression of mRNA was measured via qRT-PCR using a SYBR PrimeScript RT-PCR Kit (Takara Bio, Shiga, Japan) in accordance with the manufacturer's instructions. We used β-actin as an internal control. The primers used are listed in Table 1. We calculated relative mRNA levels based on the Ct values and normalized using β-actin expression. volume was 2 uL and the ow rate was set to 0.4 mL/min. The mass spectrometric data was collected using a UHPLC-Q Exactive Mass Spectrometer (Thermo, USA) equipped with an electrospray ionization source operating in either positive or negative ion mode. The optimal conditions were set as followed: Aus gas heater temperature, 400℃ ; Sheath gas ow rate 40 psi; Aus gas ow rate 30 psi; ion-spray voltage oating,-2800V in negative mode and 3500V in positive mode, respectively; Normalized collision energy, 20-40-60V rolling for MS/MS. Data acquisition was performed with the Data Dependent Acquisition mode. The detection was carried out over a mass range of 70-1050 m/z.

Statistical analysis
Tumor weight, body weight, the expression of mRNA and protein, the relative abundance of gut microbiota and the numbers of MDSCs were analyzed using graphpad prsim 5 (GraphPad Software, Inc. USA). Part of the 16S rRNA analysis and spearman correlation was carried out in R software. All data was expressed as mean ± S.E.M. ANOVA and t-test were used when data accorded with normal distribution and homogeneity of variance. Tukey test was used multiple comparisons. A p-value < 0.05 indicated that the difference was statistically signi cant.

Identi cation of components of JHD
The components of JHD were identi ed by HPLC-MS. Ten potential compounds, that is, Citropten 7, Formononetin, 1-Kestose, Biochanin A, Xanthohumol, Cytisine, Ferulic acid, Gallic acid Hesperetin, and Quercetin 3-O-glucoside were identi ed (Supplement 1A, Table 2), which were lined to what we identi ed before [7]. The characterizations and sources of these compounds are listed in Table 2. 3.2. JHD inhibited the growth of tumor and enhanced the therapeutic effect of sorafenib in vivo.
To investigate the combined effect of JHD and sorafenib, we determined the anti-HCC effect of sorafenib at a clinical dose (30 mg/kg) in syngeneic mouse models. Obviously, sorafenib inhibited tumor growth, and the combination treatment of sorafenib and JHD showed more dramatically suppression than JHD alone or sorafenib alone (Fig. 1A, B). We also weighed the tumor tissue excised from tumor-bearing mice, and the weight of tumor from the JHD group was lower than that of the vehicle group (Fig. 1C). To further con rmed the synergistic effect of JHD on sorafenib, we assessed the proliferation and angiogenesis of tumor tissue by staining with proliferating cell nuclear antigen (PCNA), CD31 and vascular endothelial growth (VEGF) stain. JHD had slight effect on the proliferation of tumor cells, whereas the sorafenib combined with JHD strongly suppressed cell proliferation in vivo (Fig. 1D). Sorafenib signi cantly inhibited the expression of VEGFA (Fig. 1G), which is one of the target of sorafenib [14], and decreased tumor angiogenesis(as indicated by reduced microvessel density in tumors). The combined of JHD and sorafenib further inhibited tumor angiogenesis (Fig. 1F). These results indicated that JHD induced a synergistic antitumor effect when combined with sorafenib for HCC treatment.

JHD protected occurrences of sorafenib-induced diarrhea and subsequent occurrences of body weight loss.
Although tumor growth was e ciently suppressed, sorafenib led to diarrhea and body weight loss, suggesting that side effects were induced. In clinical, ~ 80% of patients treated with sorafenib suffer AEs, such as diarrhea, body weight loss, hand-foot skin reaction, and hypophosphatemia [3,9].
The most frequent AEs (any grade) were diarrhea. Body weight loss is common in patients experienced diarrhea. In this study, signi cant reduction in body weight of mice treated with sorafenib was observed. Treatment of JHD exhibited remarkably improvement on the loss of body weight and diarrhea control induced by sorafenib. Moreover, we observed a decrease of diarrhea accompanied by less body weight loss in the mice treated with the combination of sorafenib and JHD compared with that of treatment of sorafenib alone( Fig. 2A, B).

Sorafenib induced increased proin ammatory microbiota
Emerging evidences suggest that GM plays a vital role in progression of HCC [15] and cancer immunotherapy [16]. Meanwhile, microbiota dysbiosis, as indicated by drastic bacterial population changes at the phylum and genus levels is associated with a higher risk of diarrhea and can be a consequence of diarrhea. Then we try to understand these ndings by the changes of GM. Overall, 82,887 useable reads and 1,487 operational taxonomic units (OTUs) were obtained from 20 samples(Supplement 2). The relative abundance of GM was analyzed at the phylum level, Bacteroidota and Firmicutes accounted for 90% of the total community of GM (Fig. 3A). We noted that Firmicutes/Bacteroidetes (F/B) ratio, associated with disease or imbalance in metabolism, [17,18], was increased after treated with sorafenib, whereas had a reduction in the combination of sorafenib and JHD group (Fig. 3B).Then, we sought to determine if differences existed in the alpha diversity and beta diversity. The alpha diversity was signi cantly lower in the sorafenib group than that in the other three groups based on the Shannon index (Fig. 3C), and the beta analysis showed a opposite result with weighted unifrac (Fig. 3D). Four clusters were separated on the principal coordinates analysis (PCoA) plot, in which GM of sorafenib group were far from those of sorafenib + JHD group and JHD group (Fig. 3E).
At genes level, the changes of speci c microbiota were observed, and we noticed that proin ammatory GM, such as Helicobacter, saccharimonas, faecalibacterium and enterorhabdus were increased in the mice treated with sorafenib.
Following that, we paid our attention on the difference in GM between the sorafenib group and the sorafenib + JHD group. The contribution to the average dissimilarity was investigated by SIMPER procedure. Helicobacter species (21.88%) and Lactobacillus species (14.13%) contributed most at the genus level (Fig. 4A). Although dramatic increase of Helicobacter was shown in the sorafenib group, signi cance difference was not observed between these two groups (Fig. 4B). The t-test analyses showed bacteria of the genera Muribaculum, Fusicatenibacter and Dorea were enriched in the sorafenib + JHD group (Fig. 4C). At last, Lefse analyses was used to nd biomarker, and Helicobacter species was enriched in the sorafenib group, and decreased in mice treated with the combination of sorafenib and JHD (Fig. 4D). Together, these data clearly indicated that sorafenib treatment induced expanded proin ammation microbiota which was suppressed by JHD.
3.6. JHD decreased the in ltration of in ammatory cells and inhibited the IL-6/STAT3 pathway in tumors following the modulation of GM Pathological changes in the composition of the GM that lead to intestinal in ammation are a common factor for HCC [19]. The GM can gain access to the liver as a result of a chronic in ammation disease associated to dysfunction of the intestinal barrier. Following the result that proin ammatory microbiota was induced by sorafenib, we examined in ltration of in ammatory cells in the main organs, liver, lung, spleen, and tumor tissue of mice by staining (H&E). The liver displayed the most signi cant changes, and dense punctate in ammatory cells were seen in the liver of sorafenib group, whereas fewer changes were observed in the sorafenib + JHD group and JHD group (Fig. 5A). The "leaky" intestinal membrane allows for translocation of bacteria-derived LPS (gram-negative bacteria) and lipoteichoic acid (LTA, derived from gram-positive bacteria) [20] which initiates in ammatory signaling pathway and ultimately leads to production of the in ammatory cytokines. IL-6/STAT3 signaling pathways link in ammation to cancer and was vital to the progression of HCC. We measured expression of the key signaling pathway IL-6/janus kinase 2 (JAK2)/STAT3 at mRNA and protein levels. Expression of the cytokine and pathway activator IL-6 was downregulated in the sorafenib + JHD and JHD group. Moreover, expression of the downstream molecules JAK2 and pSTAT3 was suppressed ( Fig. 5B-E). Expression of the proin ammatory mediator iNOS was increased in tumor cells in sorafenib group. (Fig. 5F, G)

Discussion
Emerging evidences show TCM acts as a concernful role in anti-tumor adjuvant therapy. JHD as a TCM formula, has been utilized in clinical for years. The application of JHD for the treatment of HCC has been studied in inducing apoptosis[6] and attenuates immunosuppressive status [21]. Sorafenib is still widely used for advanced HCC therapy. In this study, we combined JHD with sorafenib to treat HCC in a mouse model. This new strategy represents hopeful progress. First, JHD enhanced the therapeutic effect of sorafenib and alleviated reduce the adverse effect of sorafenib. Second, JHD protected sorafenib-induced increasing proin ammatory microbiota and led to downregulation of in ammation signaling. To our knowledge, this study is the rst to link the composition of the intestinal microbiota of HCC with response to sorafenib treatment.
HCC is highly lethal, and effective therapeutic treatments are still needed [5]. Immunotherapy has brought an option for cancer treatment [22], which provides survival bene t while remains low cost-effective [23]. E cient and cost-effective treatment regimens are needed. Several studies demonstrated the dramatic effect of TCM in potently synergistic and cognitive adverse effects [24,25], present a hopeful treatment for HCC. JHD inhibited the progression of HCC in our previous study, then we explored the synergistic effect of sorafenib in HCC.
We demonstrated that JHD in combination with anti-angiogenesis immune therapy sorafenib synergistically enhanced the anti-tumor response, thereby suppressing HCC and preventing the most common AEs(diarrhea and body weight loss) of sorafenib in BNL tumor-bearing mice. The mechanism of sorafenib-caused diarrhea is unclear and the management of diarrhea include dietary changes and preemptive use of loperamide. If severe or persistent diarrhea is unresponsive to management, interruptions/reductions of the sorafenib dose should be considered [26]. Severe AEs resulted in treatment interruption or discontinuation, which almost inevitably leads to treatment failure and tumor progression. We showed JHD inhibited the growth of tumor as well as protected diarrhea and the body weight loss of tumor-bearing mice from sorafenib-induced, even if increasing tumor burden. Yang and colleagues present that compound kushen injection improves the therapeutic outcomes of low-dose sorafenib and avoids body weight loss [24]. JHD alleviated diarrhea of the mice treated with clinical-dose sorafenib. Together, we provided the rst evidence for the combination treatment of JHD and sorafenib for HCC patients. We con rmed this nding by the investigation of expression of proliferation marker and target marker of sorafenib. Of note, we observe a signi cant effect of sorafenib combined with JHD on tumor cell proliferation, in spite of either sorafenib or JHD alone showing a modest suppression of proliferation in this study. Ki67 or PCNA, two common indices of proliferation, are predictive of tumor chemotherapy e cacy in terms of recurrence free survival, but Ki67 didn't show predictive of sorafenib e cacy in the phase 3 STORM trial [27]. In a previous study, JHD decreased the expression of PCNA in a H22 tumor-bearing mouse model [21]. VEGF and CD31 have been used to assess the status of vasculature and as predicters of the prognosis of cancer patients [28][29][30], although limited speci city. We found sorafenib further decreased the expression of VEGFA and CD31 when combined with JHD, showing a synergistical treatment of JHD.
GM plays an important role in health, disease, and responses to medication. Microbes exert indirect effects on the progression of tumor cells at distant sites or treatment responses by altering the types of circulating metabolites and immune responses, which, in turn, affect the general physiology of the host [31][32][33]. Nevertheless, little attention is paid to interaction between sorafenib and GM. Yamamoto K and colleagues investigated the GM of HCC patients treated with sorafenib, patients who did not suffer diarrhea had a higher abundance of Butyricimonas species and a lower abundance of Citrobacter, Peptostreptococcus, and Staphylococcaceae than that in patients with diarrhea [34]. As far as we know, we compared the composing of GM of mice treated with sorafenib or vehicle for the rst time. We showed sorafenib induced dysbiosis and decreased anti-in ammatory GM(Muribaculum, Fusicatenibacter and Dorea) which against the e cacy of sorafenib. In previous study, low level of bacterias of the genera Muribaculum [35],Fusicatenibacter[36-38]and Dorea [39,40] are associated with diarrhea and in ammation, though a part of evidence coming from non-neoplastic model. Given the critical role of chronic in ammation in progression of tumor, we speculated that sorafenib-induced changes of GM impaired the e cacy of sorafenib, which could be reversed by JHD.
In cancers, GM is mainly linked to colorectal cancer and HCC [41,42] [43]. Carcinogenesis included by bacterial pathogens in the gut mainly includes secretion of virulence factors (e.g., H.pylori), and induction of chronic in ammation; increased reactive oxygen species mediated genotoxicity(e.g., Fusobacterium nucleatum) [44]. We investigated whether the synergistic effect of JHD was due to alleviation of chronic in ammation following changes in the intestinal microenvironment. We noticed expanded lymphocyte in ltration in the liver, (but not in the lung or spleen) in sorafenib group, compared with that in the JHD group and sorafenib + JHD group. Dysbiosis and "gut leakiness" are the main contributors to liver in ammation and linked to each other. Dysbiosis may contribute to a more permeable intestinal barrier. A leaky gut enables bacterial metabolites to translocate and reach the liver readily [45]. These nding suggested JHD alleviated the in ammatory reaction of the body of mice by remodulation of GM.
As we all known, in ammatory statue is double-edged for tumor progression with activation of different pathway. We focus on the STAT3 pathway which is the central connections among in ammation, GM and HCC. Phosphorylated (p) STAT3 has been detected in ~ 60% of HCC patients, and STAT3-positive tumors have been deemed to be aggressive [46], and STAT3 is a major kinase-independent target of sorafenib in HCC [47]. A disturbed GM may activate IL-6/STAT3 signaling [48]. In this study, sorafenib upregulated expression of IL-6, JAK2, and pSTAT3/STAT3 and iNOS, while JHD or JHD combined with sorafenib inhibited these signaling. This nding is lined with the previous study DNMT3b/OCT4 expression confers resistance to sorafenib and a poor prognosis of HCC through regulation of IL-6/STAT3 signaling [49]. Moreover, in colitis, chronic stress promotes diseases by disturbing the GM and activating IL-6/STAT3 signaling [48]. The metabolite-sensing receptor Ffar2, is a short-chain fatty acidssensing G protein-couple receptor that exerts immunomodulatory effects and functions in gut homeostasis and regulation of in ammation. Ffar2 activated the STAT3 axis and increases IL-22 expression [50]. Dietary supplementation with foxtail millet can ameliorate colitis-associated colorectal cancer via activation of gut receptors and suppression of the STAT3 pathway [51]. Taken together, these data suggested that JHD sensitized the therapeutic effect of sorafenib by inhibiting IL-6/STAT3 signaling following changes in GM.
Notably, there are several shortcomings in this study. First, we didn't investigate the metabolite and toxic molecule of GM which dominantly inducted chronic in ammation in mice. Second, fecal microbiota transplant, an important approach to prevent and treat disease [52], was not employed to con rmed the effect of JHD due to the limitation of this method. Oral fecal transplantation has been applied in patients and animal model, whereas ine cient survival in complex gastrointestinal environments of the microbiota due to strongly acidic gastric uid, digestive enzymes, and bile salts [53,54]. In summary, these evidences are still limited, and further research is needed to address these issues in greater depth.

Conclusions
We investigated the anti-tumor effects of sorafenib and JHD in tumor-bearing mice. JHD improved the e cacy of sorafenib and signi cantly inhibited the growth of tumor in vivo, as well as protected tumorbearing mice from diarrhea and body weight loss induced by sorafenib. Interestingly, our study suggested JHD reshaped GM and reversed the proin ammation gut ora. The effect of JHD may be attributed to downregulated expression of the IL-6/JAK2/STAT3 pathway following changes in GM. Our investigation provided a promising approach for HCC treatment.   (D)Taxonomic cladogram obtained by LEfSe. Differences are represented by the color of the most abundant class. The diameter of each circle is proportional to the taxon's abundance. Comparison of relative abundance at the phylum (n = 5; ANOVA for four groups comparisons and Tukey test for multiple comparisons). Figure 5