Sporamin Intervention Altered the Gut Microbiome and Tumor Tissue Transcriptome in Mice Intraperitoneally Xenografted With the Human Colorectal Carcinoma LoVo Cells

Accumulating evidence indicates that sporamin, the main storage protein in the sweet potato (Ipomoea batatas), can suppress the development of colorectal cancer (CRC), but the changes in the gut microbiome after sporamin intervention and its relationships with the pathogenesis of CRC have not been investigated.

Colorectal cancer (CRC) is one of the most prevalent gastrointestinal cancers worldwide. It is a major contributor to worldwide cancer mortality and morbidity and, in 2018, over 1.8 million new cases and 881,000 deaths were reported [1]. Its etiology is very complex that involves multiply risk factors, such as genetic susceptibility, environmental toxins, lifestyle and dietary factors, and their complex interactions [2]. However, to date, the interactions between these factors have been very di cult to study and explain. On the other hand, the effective therapeutic strategies for CRC are also hampered by many limitations such as drug resistance and low long-term survival. Therefore, early prevention still is the best way to reduce the cancer burden.
In this regard, the crucial roles played by the dietary factors in the prevention and treatment of CRC have been increasingly recognized [3]. They can in uence the onset and development of CRC through multiple mechanisms such as regulating the gut microbiota dysbiosis, in uencing the integrity and function of the intestinal mucosal barriers, modifying the intestinal immune responses that may lead to in ammatory bowel diseases and CRC [4,5]. In this process, the gut microbiome may be one of the key mediators that can modify the process both directly through contact with the intestinal mucosa and indirectly through fermenting the intestinal contents and changing the secondary metabolites produced [5][6][7]. In addition, dietary factors and gut microbiota also play an important role in modulating the host's response to chemotherapeutics and other oral treatments by in uencing their bioavailability, toxicity, and e cacy [8,9]. Consequently, using probiotics and/or prebiotics to intentionally modulate the gut microbiota and favorably in uence the host's immune response and prevent the progression of CRC has been proposed as a potential therapeutic strategy [10]. What's more, more and more foods and food-derived bioactive components that are e cacious in reversing gut microbiota dysbiosis and preventing the development of CRC have been found [11][12][13]. This may open up a new avenue for the chemoprevention of CRC.
Sporamin is the major storage protein of the sweet potato (Ipomoea batatas) roots [14]. It can suppress the growth of a variety of cancer cells in vitro and in vivo [15][16][17][18][19]. However, to date, the impact of sporamin consumption on the host gut microbiome has not been explored, particularly in the context of CRC. Therefore, in the present study, the changes in the gut microbiome following oral administration of sporamin for four weeks were investigated in both healthy and tumor-bearing mice by using the bacterial 16S rRNA sequencing technology. The formation of tumors from the intraperitoneally (i.p.) transplanted human colorectal cancer LoVo cells was evaluated and the transcriptomic pro les of the tumor tissues after sporamin intervention were investigated by using the high throughput RNA sequencing (RNA-seq) technology. The potential molecular targets and mechanisms of action of sporamin were evaluated according to the changes in the gut microbiome and the tumor tissue transcriptome as well as their possible interactions.

Materials
Sporamin was extracted from fresh sweet potatoes as previously reported [20]. The human colorectal cancer LoVo cell line (Tumor Cell Bank of the Chinese Academy of Medical Sciences, Beijing, China) were cultured in Dulbecco's Modi ed Eagle's Medium (DMEM) (Thermo Fisher Scienti c, Shanghai, China) supplemented with 10% fetal bovine serum (Thermo Fisher Scienti c, Shanghai, China) and 1% penicillin-streptomycin at 37°C in a humidi ed incubator with 5% CO 2 . Male BALB/c athymic nude mice (Vital River Laboratory Animal Technology Co. Ltd., Beijing, China) aged 6 weeks were housed in an access-restricted room with a 12-h/12-h light/dark cycle, controlled temperature (22°C, daily temperature uctuations ≤ 3°C), relative humidity of 40%-70%, an air ow velocity ≤ 0.18 m/s, and a room air pressure gradient of 20-50 Pa under speci c pathogen-free conditions.

Experimental design
The experiment protocol was approved by the institutional ethics committee at Capital Medical University (license number: AEEI-2016-018). After a one-week acclimation period, twelve BALB/c athymic nude mice were randomly divided into four groups (n=3 per group). CG1: healthy mice treated with vehicle (control group); CG2: healthy mice treated with sporamin; TCG: tumor-bearing mice treated with vehicle; TTG: tumor-bearing mice treated with sporamin. Mice in the CG1 group were intragastrically infused with 0.5 mL of distilled water (vehicle) per day for four weeks (28 days). Mice in the CG2 group were intragastrically infused with 0.5 mL of sporamin at 0.5 g/kg BW per day in distilled water for four weeks.
Mice in the TCG group were i.p. transplanted with 2.5×10 7 LoVo cells in 0.2 mL of PBS and then intragastrically infused with 0.5 mL of distilled water (vehicle) daily for four weeks. Mice in the TTG group were i.p. xenografted with 2.5×10 7 LoVo cells in 0.2 mL of PBS and then intragastrically infused with 0.5 mL of sporamin at 0.5 g/kg BW per day in distilled water for four weeks. All animals had free access to feed and drinking water throughout the experiment. Their body weights (BW) were recorded every 4 days during the experiment. At the end of the experiment, all mice were anesthetized with 10% chloral hydrate (0.1 mL/10 g BW) and sacri ced by cervical dislocation.

Sample collection
Fresh fecal samples were collected from the mice under sterile conditions the day before sacri cing and were stored at -80°C until analysis. An autopsy was performed and tumors formed in the abdominal cavity were carefully examined, counted, measured for weight and size, collected and washed with RNAlater™ Stabilization Solution (Thermo Fisher Scienti c, Shanghai, China), and stored at -80°C until analysis.
Bacterial genomic DNA extraction, sequencing, and bioinformatic analysis Bacterial genomic DNA was extracted from 0.18 to 0.22 g of the fecal samples using the E.Z.N.A ® Mag-Bind Soil DNA Kit (Omega Bio-Tek, GA, USA) according to the manufacturer's instructions. The V3-V4 hypervariable region of the bacterial 16S ribosomal RNA gene was ampli ed by polymerase chain reaction (PCR) using primers 341F 5 -CCCTACACGACGCTCTTCCGATCTG-barcode-CCTACGGGNGGCWGCAG-3 and 805R 5 -GACTGGAGTTCCTTGGCACCCGAGAATTCCAGACTACHVGGGTATCTAATCC-3 under the following condition: 94˚C for 3 min, followed by ve cycles at 94°C for 30 s, 45˚C for 20 s and 65°C for 30 s; this was followed by another 20 cycles at 94˚C for 20 s, 55°C for 20 s and 72°C for 30 s, with a nal extension at 72°C for 10 min. An 8-bp barcode sequence, unique to each sample, was attached to the primers for multiplexing. PCR was performed in triplicate in 50 μL reaction mixtures containing 5 μL of 10 × PCR buffer, 0.5 μL of 10 mM dNTPs, 10 ng of genomic DNA, 0.5 μL of each primer (50 μM), and 0.5 μL of Plantium Taq polymerase (5 U/µL). The PCR products were separated by 1.2% agarose gel electrophoresis, and bands of the desired size (> 600 bp) were puri ed using the SanPrep DNA Gel Extraction Kit (Sangon Biotech, Shanghai, China) according to the manufacturer's instructions.
Then, the bacterial genomic DNA samples were sent to Sangon Biotech Co. Ltd. (Shanghai, China) for 16S rRNA sequencing, where the DNA concentration of each PCR product was determined using a Qubit2.0 DNA uorometer (Thermo Fisher Scienti c, Shanghai, China). Equimolar puri ed products were pooled, and paired-end sequenced (2 × 300) on an Illumina MiSeq PE300 platform (Illumina, San Diego, CA, USA) according to the standard protocol at Sangon Biotech.
After sequencing, demultiplexing and quality-ltering of the raw fastq les were performed using QIIME (version 1.8.0) [21]. UCHIME [22] was applied to identify and remove chimeric sequences.
UPARSE [23] was used to cluster operational taxonomic units (OTUs) with a 97% similarity cutoff level according to Greengenes (version 13.8). RDP Classi er [24] was used to assign sequences generated from samples to the corresponding taxonomy with a con dence threshold of 70% [25]. The alphadiversity indices, including Chao1, ACE, Shannon, and Simpson, were calculated to summarize the richness, evenness, and distributions of the gut bacterial community. The principal coordinates analysis (PCoA) based on the weighted UniFrac distance matrix and the hierarchically clustered heat maps were constructed in R (version 3.2) to assess the beta-diversity of the gut microbiome across groups. Phylogenetic Investigation of Communities by Reconstruction of Unobserved States (PICRUSt) [26] was used to predict the functional potential of the gut microbiome based on the OTU table, where information relating to gene annotations was obtained from the Kyoto Encyclopedia of Genes and Genomes (KEGG) database. The abundance of each functional category was calculated according to OTU abundance and STAMP (version 2.1.3) [27] was used to analyze differential metabolic pathways across groups.
Transcriptome pro ling of the tumor tissues Tumor tissue samples from the TCG and TTG group were sent to Beijing Genomics Institute (BGI, Wuhan, China) for RNA extraction and sequencing (RNA-seq) according to the following procedures: (i) 1 μg of the total tissue RNA was extracted using the TRIzol ® reagent and treated with DNaseI at 37 °C for 20 min to digest DNA. Oligo dT magnetic beads were used to select mRNA with polyA tail; (ii) the puri ed mRNA were fragmented and reverse transcribed to double-strand cDNA (dscDNA) by using the N6 random primer (Sequence: 5' d(N6) 3' [N=A, C, G, T]); (iii) the ds cDNAs were treated by a traditional process, including end repairing with phosphate at the 5′ end and stickiness 'A' at the 3′ end, ligation, and adaptor with stickiness 'T' at the 3′ end; (iv) two speci c primers were used to amplify the ligation product; (v) the double-stranded PCR products were heat-denatured to single-strand and circularized by splint oligo and DNA ligase; (vi) sequencing was performed on the BGISEQ-500 platform with the prepared library. (vii) bioinformatics: low-quality reads, reads with adaptors, and reads with unknown bases were ltered to get clean reads and were stored in FASTQ format [28]. The clean reads were aligned to reference genome sequences via Bowtie2 [29]. Gene and isoform expression levels were quanti ed with RSEM [30]. Differentially expressed genes (DEGs) were detected with the DEseq2 package in R [31]. The criteria for selecting DEGs were fold-change ≥ 2.00 and adjusted P-value < 0.05. Hierarchical clustering analysis for DEGs was performed by the pheatmap package in R. Gene ontology (GO) and pathway enrichment analysis based on the KEGG database were performed online. Transcription factor (TF) prediction was performed by using the DIAMOND [32] combined with the information recorded in the human TF database (Human TFDB). A protein-protein interaction (PPI) network was plotted using the STRING database [33] and the Cytoscape network analysis platform [34].  Supplementary Table S1. All reactions were performed in triplicate under the following procedures: 95 ˚C for 10 min, followed by 37 cycles of 95 ˚C for 15 s, 60 ˚C for 30 s and 72 ˚C for 30 s, and a nal extension at 72°C for 10 min. The Ct values of the genes were used to calculate the degree of gene expression using the 2 -∆∆CT method normalized to that of the β-actin gene [35].

Statistical Analysis
Data are expressed as mean ± SEM (standard error of the mean). Differences among groups were examined by one-way analysis of variance (ANOVA) using SPSS Statistics version 18.0 (IBM, New York, USA). The Turkey-Kramer Post-hoc test was used when the homogeneity of variance was equal between groups. The Game-Howell test was used when the variance was unequal. P < 0.05 was considered statistically signi cant.

Results
Sporamin alleviated the tumor burden of mice without any observable adverse effects During the experiment, the healthy mice in the CG2 group showed no obvious side effects compared with the control mice in the CG1 group, indicating that oral consumption of sporamin is safe for healthy mice. For tumor-bearing mice in the TCG and TTG group, all of them developed tumors in their abdominal cavities after about eight days of the i.p. transplantation of the LoVo cells, which became touchable and bulged gradually, indicating that the transplantation was successful. During the experiment period, tumorbearing mice (the TCG group in particular) were listless, inactive, and slightly thinner than their counterparts in the sporamin intervention group (TTG) who were more physically active and mentally poise. Throughout the experiment period, the BW of the mice were comparable across the four groups, indicating that consumption of sporamin for four weeks does not signi cantly in uence the growth of the mice (P>0.05, Figure 1A). Autopsy of tumor-bearing mice revealed that a great number of tumors were grown in the abdominal cavity of the mice in the TCG group, which were greatly relieved by sporamin intervention in the TTG group ( Figure 1B, C).
Sporamin intervention favorably changed the gut microbiome and its metabolome  Figure S1) indicates that our data volume and sequencing depth were su cient. There were no signi cant differences in the alpha-diversity indices, including the ACE, Chao1, Shannon, and Simpson indices, across the four groups (P > 0.05) (Supplementary Table  S2).
To explore the dissimilarities of bacterial communities across groups (beta-diversity), the PCoA and hierarchically clustered heat maps were constructed. The three-dimensional PCoA plot ( Figure 2A) shows that 45.4%, 23.8%, and 12.0% of the dissimilarities between the samples could be explained by PCoA1, PCoA2, and PCoA3, respectively. The two-dimensional PCoA plot ( Figure 2B) clearly shows that the healthy mice (CG1 and CG2) mainly locate in the second quadrant, indicating that consumption of sporamin did not signi cantly alter the structure of the gut microbiome in the healthy mice. This is consistent with its impacts on the BW of mice. Tumor-bearing mice (TCG) tend to aggregate to the fourth quadrant, indicating that the growth of the xenografted tumors had signi cantly altered the structure of the gut microbiome, which were partly pulled back to the rst and the second quadrant after sporamin intervention (TTG), suggesting that sporamin intervention could normalize the abnormal structure of the gut microbiome in tumor-bearing mice. To look at the changes more closely, hierarchically clustered heat maps ( Figure 2C) illustrating the differential composition of the fecal bacterial communities at the phylum and genus level were constructed, which show that mice who had received the sporamin intervention (CG2 and TTG) were clustered into a subgroup while those who had not received the intervention (CG1 and TCG) were clustered into the other subgroup, indicating that the consumption of sporamin had brought a signi cant impact on the taxonomic composition of the microbiome of both healthy and tumor-bearing mice. Therefore, we further compared the differences in the relative abundances of the bacterial taxonomies after sporamin treatment. Figure 3A shows that there were eight bacterial phyla and an archaea phylum, Euryarchaeota, identi ed in all fecal samples, among which, Bacteroidetes and Firmicutes were the two predominant phyla in all animals. After tumor transplantation, the proportion of Bacteroidetes was further increased while those of other phyla were correspondingly reduced. In comparison, sporamin intervention decreased the proportion of Bacteroidetes and increased that of Firmicutes in tumor-bearing mice, and subsequently increased the Firmicutes/Bacteroidetes ratio to a more balanced level. The details of the relative abundances of all phyla are provided in Supplementary Table S3. The distribution histogram of the gut microbiome at the genus level ( Figure 3B) shows that 48 bacterial genera were identi ed in all samples. Sporamin intervention increased the proportions of Muribaculaceae and Oscillibacter in healthy mice. Notably, the proportion of Bacteroides increased greatly after tumor transplantation, which was signi cantly inhibited by sporamin intervention. In addition, the proportions of Barnesiella and Lactobacillus were signi cantly increased by sporamin intervention in tumor-bearing mice. The details of the changes in the bacterial communities at the genus level are provided in Supplementary  Table S4.
To explore the functional relevance of the changes in the bacterial compositions brought by sporamin intervention, the 16S rRNA data were submitted to the PICRUSt for functional prediction. Figure 4A shows that sporamin did not signi cantly alter the metabolic pathways that were inferred from the phylogenetic composition of the gut microbiome in the healthy mice (CG1 & CG2), but transplantation of the LoVo tumor cells led to a signi cant increase in the functional abundance of the proximal tubule bicarbonate reclamation pathway (from 0.02% to 0.03%, P=0.03, Figure 4B) compared with control mice, implying that the acid-base balance and the kidney function of tumor-bearing mice might have been impaired by the microbiome-related factors, which disappeared after sporamin intervention ( Figure 4C). In addition, in tumor-bearing mice, sporamin intervention signi cantly increased the relative proportions of the bacterial secretion system pathway (P=0.04), the glycan biosynthesis and metabolism pathway (P=0.04), and the stilbenoid, diarylheptanoid, and gingerol, biosynthesis pathway (P=0.04) while signi cantly reduced the proportions of the phenylalanine metabolism pathway (P=0.02) and the glyoxylate and dicarboxylate metabolism pathway (P=0.03), implying that sporamin intervention may exert a signi cant impact on the development of tumor through modulating the metabolome of the gut microbiome and then alter the pathophysiology of the host such as its intestinal barrier function, immune responses, and other cancerrelated metabolisms, etc.

Sporamin changed the transcriptome of the tumor tissues
To explore the possible molecular targets of the sporamin intervention in tumor-bearing mice, the gene expression pro le of the xenografted tumor tissues after sporamin intervention was investigated using RNA-seq technology. As a result, the sequencing generated 24065679±21370 raw sequencing reads and 24066787±21381 clean reads after ltering out low-quality reads. The numbers of the reads and their quality metrics for each sample are provided in Supplementary Table S5. The average mapping ratio with the reference genome was 85.25%; the average mapping ratio with the reference genes was 69.10% and a total of 21,655 genes were detected. RSEM-quanti ed gene expression levels and the numbers of identi ed expressed genes in each sample are shown in Supplementary Figure S2.
To elucidate the functional interplays between the proteins encoded by these DEGs, a protein-protein interaction (PPI) network was constructed using the homology with the known PPIs reported in the STRING database ( Figure 5D), which revealed a PPI network composed of seven proteins (i.e. HSP90AA1, HSPH1, HSPA4L, NEMF, CHORDC1, CHD1, and NIPBL) that were all upregulated after sporamin intervention.
The functional relevance of these DEGs was also analyzed by the gene ontology (GO) analysis in terms of the three classi cations: biological process (BP), cellular component (CC), and molecular function (MF). Figure 6A shows that, in the CC category, eleven DEGs were annotated to cell and cell party; ten were annotated to organelle. In the MF category, eleven DEGs were annotated to binding; three were annotated to catalytic activity. In the BP category, twelve DEGs were annotated to positive regulation of the cellular process; eleven were annotated to positive regulation of the metabolic process.
Then, the critical biological pathways associated with the DEGs were investigated based on the KEGG database ( Figure 6B) and found that three pathways (i.e. protein processing in endoplasmic reticulum, glycosylphosphatidylinositol (GPI)-anchor biosynthesis pathway, and mineral absorption pathway, respectively) were signi cantly enriched for the DEGs found between the TCG and TTG group. The KEGG-DEG relationship network analysis ( Figure 6C) revealed the DEGs that are responsible for the promotion of the protein processing in the endoplasmic reticulum (i.e. HSPH1, HSP90AA1, HSPA4L), the increase in the mineral absorption pathway (i.e. SLC26A9), and the reduction in the GPIanchor biosynthesis pathway (i.e. PIGL) after sporamin intervention.
Then, in the TF prediction analysis, four TFs were found upregulated after sporamin intervention in tumorbearing mice, i.e. NEMF, HSP90AA1, BCL11A, and ZBED6. Their activities are closely related with three zinc nger domains (i.e. zf-C2H2, zf-C2HC, and zf-BED, Figure 6D), implying that sporamin may increase their transcriptional activities by increasing the absorption and bioavailability of zinc ions in their activity center.

Discussion
In the current study, we demonstrated for the rst time that oral consumption of sporamin for four weeks could alter the composition and function of the gut microbiome and the transcriptome of the tumor tissues in tumor-bearing mice that were i.p. xenografted with human colorectal cancer LoVo cells. Sporamin mainly exerts its role on the gut microbiome by increasing the relative abundances of Barnesiella and Lactobacillus and reducing that of Bacteroides in tumor-bearing mice. As a result, the metabolome of the gut microbiome was changed, which possibly mediated part of the inhibitory effect of the sporamin intervention on the growth of the tumor. Transcriptomic analysis of the tumor tissues revealed that sporamin mainly upregulated the genes involved in the protein processing in the endoplasmic reticulum and the mineral absorption pathway, and reduced the GPI-anchor biosynthesis pathway, thereby exerting a suppressive role in the growth of the xenografted tumor.
Sporamin has been previously shown to play an anti-cancer role in a diversity of cancers in vitro and in vivo [15][16][17][18][19]. However, its mechanisms of action in the context of different cancers have not been completely understood. Because it has been proved to be a potent trypsin inhibitor that can inhibit the activity of the main digestive enzymes [20], it is reasonable to hypothesize that it can survive gastrointestinal digestion and reach the large intestine and interact with the microbiome there to exert an effect on the host. This is possibly an important route by which sporamin alters the metabolism, immunity, and pathophysiology of the host and suppresses the growth of cancers in the body. Our ndings that consumption of sporamin for four weeks signi cantly altered the gut microbiome and its metabolome in tumor-bearing mice con rmed this hypothesis, suggesting that favorably improving the gut microbiome and its metabolome may at least partly mediate its anti-cancer effect on the host.
Regarding the etiology and pathology of CRC, it has been recognized that genetic susceptibilities, dietary patterns, and environmental toxins in combination are responsible for the onset and development of the disease. There are a lot of potentially harmful factors in the environment where the intestinal epithelial cells live, among which the intestinal microbiota constitute one of the most pertinent and persistent components of the intestinal microenvironment [36,37] and are closely associated with the pathogenesis of malignancies in the intestinal tract [38]. Some intestinal bacteria and their metabolites can stimulate in ammation and cause DNA damage and contributing to the occurrence and development of CRC [39]. On the other hand, some other intestinal bacteria, such as short-chain fatty acid (SCFA)producing bacteria, have been shown to favorably in uence the health of the gut epithelia and prevent their malignant transformation. In line with this, improving the composition and function of the gut microbiota via prebiotics or probiotics or antibiotics etc. has also been proposed as a method to prevent cancer and is successful in several studies [5,7,11]. Moreover, the scope of the agents that can be used to modulate the gut microbiome is expanding. More and more natural products that are not mainly composed of dietary bers are potent gut microbiota modulators and may exert bene cial roles in a variety of diseases [13,40,41].
In the present study, the proportion of Bacteroidetes increased signi cantly after tumor transplantation, which was partially reduced by sporamin intervention, resulting in a more balanced Firmicutes/Bacteroidetes ratio. At the genus level, sporamin intervention signi cantly increased the relative abundance of Barnesiella and Lactobacillus and reduced that of Bacteroides in tumor-bearing mice. Barnesiella is decreased in CRC patients [42] and plays anti-in ammatory, anti-cancer roles in a colitis-associated CRC model through increasing the production and secretion of acetate, butyrate, and propionate in the gut [43]. As to Lactobacillus, it is a widely accepted probiotic that demonstrates bene cial roles in preventing CRC [44,45].
Prediction of the functional potentials of the gut microbiome by PICRUSt indicated that the proximal tubule bicarbonate reclamation function was elevated in tumor-bearing mice, suggesting that they may be under the threat of acidosis and need to increase their bicarbonate reclamation capacity to maintain the acid-base balance [46]. Sporamin intervention abolished this elevation and produced ve other favorable changes in the metabolic pathways. First, sporamin reduced the activity of the phenylalanine metabolism pathway. A study involving a large cohort of multi-omics data indicated that the activity of the phenylalanine metabolism pathway led to an increase in intramucosal carcinomas in the colon [47]. A product of phenylalanine degradation, benzene-oxide, is reactive towards proteins and DNA and is potentially carcinogenic [48]. Second, sporamin intervention reduced the activity of the glyoxylate and dicarboxylate metabolism pathway that is enriched in colorectal tumor tissues [49]. Chronic exposure of breast epithelial cells to the product of glyoxylate oxidation, oxalate, has been found to promote the transformation of normal cells to tumor cells by inducing the expression of the c-fos proto-oncogene and the proliferation of cancer cells [50]. A reduction in this pathway might increase the abundance of Lactobacillus [51] and Enterococcus [52] which was also observed in the present study and is bene cial for CRC. Third, sporamin intervention increased the activity of the glycan biosynthesis and metabolism pathway, indicating that sporamin might promote the overall survival and growth of gut bacteria because the synthesis of glycans in the cell wall is vital for bacteria [53]. Future studies should aim to quantify the total number of bacteria in the gut after sporamin intervention. Fourth, sporamin intervention increased the activity of the bacterial secretion system that might also be a re ection of an improved survival environment for the gut microbiome after sporamin intervention because the secretory function is an essential re ection of the viability of the gut microbiota. Fifth, sporamin intervention increased the stilbenoid, diarylheptanoid, and gingerol, biosynthesis pathways that might be also attributed to an overall increase in bacterial viability so that they can ferment the intestinal contents and produce more plant-derived bioactive compounds. Collectively, changes in these pathways after sporamin intervention suggest that the gut microbiome and its metabolome may at least partly mediated the bene cial role of sporamin intervention in tumor-bearing mice. Sporamin intervention may alter the pathophysiology of the host such as its intestinal barrier function, immune responses, and other cancer-related metabolisms via modulating the gut microbiome.
Examination of the transcriptome of the tumor tissues after sporamin intervention revealed that it mainly targets genes whose protein products participate in the protein processing in the endoplasmic reticulum (i.e. HSP90AA1, HSPH1, and HSPA4L), the GPI-anchor biosynthesis pathway (i.e. PIGL), and the mineral absorption pathway (i.e. SLC26A9). TF predictions identi ed four TFs (i.e. ZBED6, BCL11A, HSP90AA1,and NEMF), the activity of which are related to three conserved DNA binding zinc nger domains (i.e. zf-C2H2, zf-C2HC, and zf-BED). Among them, ZBED6 (zinc nger, BED-type containing 6) modulates gene expression primarily by repressing transcription. The gene targets of ZBED6, including the Wnt, Hippo, TGF-β, EGF receptor, or PI3K pathway components, are all involved in CRC development [54]. BCL11A (B-cell lymphoma/leukemia 11A) also functions mainly as a transcriptional repressor [55] and down-regulation of BCL11A has been found mediating the resistance to radiotherapy in the treatment of CRC [56]. HSP90AA1 is closely related to the folding and assembly of many signaling molecules in the Ras/Raf/ERK/1/2 signaling pathway [57]. The binding and dissociation of Hsp90 from these molecules mediate the transformation of these molecules between inactive and active forms [57]. Thus, upregulation of HSP90AA1 after sporamin intervention may suppress the activities of the kinases that drive cancer cell growth and protein synthesis by keeping them in inactive forms. Analysis of the GeneCards ® database revealed that NEMF (nuclear export mediator factor) is a component of the ribosome quality control complex and a similar protein in the y functions as a tumor suppressor but its role in CRC remains to be explored.
Collectively, it appears that the main changes in the transcriptome of the tumor tissues after sporamin intervention are closely associated with the protein processing in the endoplasmic reticulum.
Upregulation of the expression of the molecular chaperones (e.g. HSPH1, HSPA4L) and their cochaperones (e.g. CHORDC1) after sporamin intervention may result from the stresses to cancer cells posed by sporamin intervention. Depending on the speci c cellular context, these chaperone proteins may behave both as an oncosuppressor and as a proto-oncogene as is shown in the case of CHORDC1 [58,59]. However, the current work cannot answer the question of whether these hypotheses are true or not, and more experiments are warranted to investigate the intracellular signal transduction mechanisms involved in cancer cells after sporamin intervention. In addition, regarding the role played by the gut microbiome and its metabolome in the alterations in the transcriptome of the tumor tissues, it is obvious that the current work cannot directly answer this question too and more works are needed in the future to establish a much more solid link between them although this study has provided several clues. Of course, it should also be noted that the gut microbiome may only partly mediates the effect of sporamin because it can also be digested into small peptides, amino acids, and amines, etc. by digestive enzymes and the bacteria in the large intestine. And, once absorbed, their secondary metabolites in the body may also exert functionality at local and/or systemic levels.

Conclusion
In conclusion, this study is the rst that investigated the changes in the gut microbiome and tumor tissue transcriptome after the consumption of sporamin in both healthy and tumor-bearing mice. The ndings demonstrated that the consumption of sporamin for four weeks could favorably alter the gut microbiome and its metabolome, improving the gut microenvironment and the viability of the gut microbiota and increasing the detoxi cation and bioactive substance production activities in the large intestine, by which the host's metabolome may be altered and in turn exerts a suppressing effect on the protein synthesis and growth of tumor tissues. However, more experiments are needed to substantiate these hypotheses in the future.

Availability of data and materials
The datasets used and analyzed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests.

Funding
This work was supported by the grants from the National Natural Science Foundation of China (No.

& 81703216).
Authors' contributions PGL conceived and designed the study; GG provided assistance for specimen collection and the guidance of this study; SJC, XPZ and JJZ performed the experiment; CY analyzed data and wrote the manuscript; KWZ and BWC revised the manuscript; PGL had primary responsibility for nal content. All authors read and approved the nal manuscript.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. oatimage7.jpeg