Abstract

   Oral cancer is one of the most common cancers worldwide and ranks
   fourth for the mortality rate of cancers in males in Taiwan. The oral
   microbiota is the microbial community in the oral cavity, which is
   essential for maintaining oral health, but the relationship between
   oral tumorigenesis and the oral microbiota remains to be clarified.
   This study evaluated the effect of microbiome dysbiosis on oral
   carcinogenesis in mice, and the impact of the microbiome and its
   metabolic pathways on regulating oral carcinogenesis. We found that
   antibiotics treatment decreases carcinogen-induced oral epithelial
   malignant transformation. Microbiome analysis based on 16S rRNA gene
   sequencing revealed that the species richness of fecal specimens was
   significantly reduced in antibiotic-treated mice, while that in the
   salivary specimens was not decreased accordingly. Differences in
   bacterial composition, including Lactobacillus animalis abundance, in
   the salivary samples of cancer-bearing mice was dramatically decreased.
   L. animalis was the bacterial species that increased the most in the
   saliva of antibiotic-treated mice, suggesting that L. animalis may be
   negatively associated with oral carcinogenesis. In functional analysis,
   the microbiome in the saliva of the tumor-bearing group showed greater
   potential for polyamine biosynthesis. Immunochemical staining proved
   that spermine oxidase, an effective polyamine oxidase, was upregulated
   in mouse oral cancer lesions. In conclusion, oral microbiome dysbiosis
   may alter polyamine metabolic pathways and reduce carcinogen-induced
   malignant transformation of the oral epithelium.

   Keywords: microbiota, oral cancer, antibiotics-induced microbiome
   dysbiosis, polyamine, spermine oxidase

Introduction

   Oral squamous cell carcinoma (OSCC) is one of the most common human
   malignancies and ranks fourth in terms of the mortality rate of cancer
   in males in Taiwan [41]^1. Tobacco, alcohol, and betel nuts are widely
   considered major risk factors for oral cancer, and emerging studies
   have also found that poor oral hygiene is associated with a high
   incidence of oral cancer [42]^2. Despite improvement in surgical
   techniques and advances in chemotherapy and radiotherapy, the 5-year
   survival rate of OSCC patients with regional lymph node or other organ
   metastases remains less than 50%, with no significant improvement over
   the past 30 years [43]^3^,[44]^4. Therefore, the prevention of cancer
   formation, early diagnosis, and development of novel targeted therapies
   are crucial for oral cancer.

   Although cancer is generally considered to be a disease of host
   genetics and environmental factors, microorganisms are still closely
   related to human malignancies [45]^5. For example, Helicobacter pylori
   is the main microorganism associated with gastric cancer [46]^6,
   Fusobacterium nucleatum has been indicated as a closely related
   microbiome to colorectal carcinogenesis [47]^7, and Streptococcus
   gallolyticus has been known to promote colorectal cancer development
   [48]^8. Although the advances in next-generation sequencing
   technologies have been used to analyze the microbiome in oral cancer
   specimens and salivary samples and confirmed that microbial
   disturbances are associated with OSCC in recent years, no consistent
   bacterial species or flora has been identified [49]^9^-[50]^12.
   Antibiotic-induced microbiome depletion has been used frequently to
   study the role of the gut microbiome in pathological conditions
   [51]^13^,[52]^14. In oral cancer, previous studies have used germ-free
   mouse models to assess the role of the microbiome in carcinogen-induced
   oral cancer, but with conflicting results [53]^15^,[54]^16. Therefore,
   the effects of antibiotic-induced oral microbiome dysbiosis or
   depletion on oral carcinogenesis remain to be investigated.

   Metabolic pathways or metabolites have been known to be altered in
   cancer cells and considered critical for the initiation and progression
   of cancer. The crosstalk of microbial communities and human cells
   contributes to the maintenance of cellular metabolism. Changes in the
   balance between the host and microbiota create favorable conditions for
   cancer development. For example, microbial metabolites can modulate
   inflammatory responses in the tumor microenvironment or create
   immunosuppressive conditions that promote cancer development and induce
   epithelial-mesenchymal transition to promote tumor metastasis [55]^17.
   Polyamines are small polycationic molecules, which are produced from
   ingested foods and microbiota metabolism, and are involved in multiple
   cellular processes, such as gene expression regulation and cell
   proliferation [56]^18. In cancer, polyamine metabolism is frequently
   dysregulated by cancer-associated microbes, such as in colorectal
   cancer [57]^19 and pancreatic cancer [58]^20. However, whether the
   polyamine metabolism in OSCC is disturbed by microbial alterations
   needs to be further clarified. In this study, we aimed to investigate
   the effects of microbiome dysbiosis on oral carcinogenesis in mice and
   to decipher the impact of the microbiome and its metabolic pathways on
   regulating oral carcinogenesis.

Materials and Methods

Animal experiments and Specimen Collection

   All animal care and experiments protocols were approved by the
   Institutional Animal Care and Use Committee (IACUC), NCKU, and complies
   with ARRIVE (Animal Research: Reporting of In vivo Experiments)
   guidelines. A total of twenty 6-week-old C57BL/6 mice (10 males and 10
   females) were randomly divided into a control group and an antibiotics
   treatment group. A brief summary of the administration of
   4-nitroquinoline 1-oxide (4-NQO)/arecoline and antibiotics, as well as
   the collection of saliva and stool samples, is shown in Figure [59]1A.
   The antibiotic regimen consisted of a cocktail of three different
   antibiotics, ampicillin (1 mg/mL), streptomycin (5 mg/mL), and colistin
   (1 mg/mL), which were dissolved in sterile drinking water [60]^13. Mice
   in the antibiotic treatment group drank antibiotic-containing water for
   1 week and 3 weeks before and after carcinogen treatment, and the water
   was changed twice per week. All mice underwent induced oral cancer
   formation for 8 weeks by drinking 4-NQO (50 μg/mL) and arecoline (500
   μg/mL) in water, and the water was changed twice per week [61]^21. The
   murine oral mucosa and tongue were gently wiped with cotton swabs to
   absorb the saliva and then soaked in 500 μl PBS buffer for further
   analysis after the 12^th week. Stool samples of mice in the control and
   the antibiotic treatment groups were collected separately at week 22.
   After the 12^th week, the formation of oral tumors was observed and
   photographed every three days. Observational data from all mice were
   included in the analysis, except for one mouse in the antibiotic group
   that died of injury in week 17. At the 30^th week, pathological changes
   in the tongue were observed in most of the mice, and some tumors
   significantly affected the feeding of the mice and resulted in weight
   loss of 25%, which was set as the end of the experiment. All mice were
   euthanized with CO[2], and their tongues were collected and fixed in
   10% formalin for further histological examination. All experiments in
   the study were single-blind.

Figure 1.

   [62]Figure 1
   [63]Open in a new tab

   Antibiotic-induced microbiome dysbiosis decreases
   4-NQO/arecoline-induced murine oral tumorigenesis. (A) Schematic
   diagram of the experimental protocol of antibiotic treatment and
   4-NQO/arecoline-induced oral cancer in mice. (B) Representative images
   of mouse tongues without OSCC and with OSCC lesions, and histological
   staining of tongue tissue sections. Scale bar: 500 µm. (C) Incidence of
   oral lesions in mice. Fisher's exact test was used to analyze the P
   value of the incidence of oral lesions in the two groups.

Bacterial genomic DNA extraction

   Saliva specimens were collected and centrifuged 1500 rpm to remove cell
   debris and then harvested the bacteria by spinning 5000 × g for 10
   minutes. Resuspended the bacteria with 500 ul PBS, bacterial genomic
   DNA was isolated with QIAamp DNA Mini Kit (Qiagen, Germany) in
   accordance with the manufacturer's protocol. Stool samples were
   randomly collected from the control and antibiotic group mouse cages
   and dissolved in 500 ul PBS, respectively. The gDNA of mouse feces was
   extracted by using DNeasy PowerSoil Kit (Qiagen) according to the
   manufacturer's instructions. The quantity and quality of isolated gDNA
   were determined using a NanoDrop 2000 (Thermo Scientific, USA).

Microbiome analysis

   The microbial taxonomy of mouse saliva and fecal samples was analyzed
   by sequencing the V3-V4 hypervariable region of the 16S rRNA gene. 1μg
   gDNA was first amplified by polymerase chain reaction (PCR) with
   bacteria-specific 16S rDNA primers (27F 5'-AGAGTTTGATCCTGGCTCAG-3';
   1492R 5'-GGTTACCTTGTTACGACTT-3'). The V3-V4 hypervariable region of 16S
   rDNA was amplified using the bacterial-specific forward and reverse
   primer set using the 16S Metagenomic Sequencing Library Preparation Kit
   (Illumina, San Diego, CA, USA). Indexed adapters were added to the
   amplicons using the Nextera XT Index Kit (Illumina, San Diego, CA,
   USA), according to the manufacturer's instructions. The amplified DNA
   sizing accuracy was checked using the 4200 TapeStation System (Agilent
   Technologies, Santa Clara, CA, USA). After library construction,
   samples were mixed with MiSeq Reagent Kit v3 (600-cycle) and loaded
   onto a MiSeq cartridge, then a 2x300 bp paired-end sequencing run was
   performed using the MiSeq platform (Illumina, San Diego, CA, USA).
   Sample preparation, library construction and sequencing were performed
   by Welgene Biotech Co., Ltd. (Taipei, Taiwan).

OTU Analysis & Annotation

   The paired-end raw FASTQ reads generated from Illumina MiSeq platform
   were filtered to remove the Illumina PhiX Control using Bowtie 2
   [64]^22. Trimmomatic was used to remove sequences with average QV<20 to
   produce clean reads [65]^23. These clean paired-end reads were merged
   by overlaping sequences using FLASH [66]^24; then low-quality tails and
   primers were trimmed and filtered based on length using mothur to
   produce filtered tags [67]^25. USEARCH was used to remove PCR chimeras
   to produce effective tags and to construct OTUs (Operational Taxonomic
   Units) at 97% se-quence identity [68]^26. Taxonomy assignment of OTU
   sequences was performed using mothur with the SILVA database v132
   [69]^27. Taxa abundances, rank abundance curves were calculated and
   plotted using Qiime [70]^28. Rarefaction curves were calculated using
   the VEGAN R package [71]^29 as were the Beta diversity indices
   (Bray-Curtis matrix, PCoA and NMDS). OTU analysis and annotation was
   performed by Welgene Biotech Co., Ltd. (Taipei, Taiwan). Linear
   discriminant analysis (LDA) was performed using LEfSe (standalone
   v1.1.2) with default options [72]^30. Before analysis, OTUs annotated
   as “uncultured” or “unknown” were discarded. When the species could not
   be determined for an OTU, the species name was replaced by the OTU
   number. Association between species abundance and OSCC was detected
   using MaAsLin2 (v1.10.0; options: -t NONE, -f group, -n NONE, -z FALSE,
   -max_significance 0.25) [73]^31. Differentially abundant species were
   identified using ANCOM2 (v2.1; options: out_cut=0.05, zero_cut=0.9,
   lib_cut=1000, p_adjust_method=BH) [74]^32.

Metabolism Prediction

   Open database sources, including the KEGG and MetaCyc pathway database
   [75]^33, were used to identify metabolic pathways. A heat map of the
   identified key metabolites was performed by Welgene Biotech Co., Ltd.
   (Taipei, Taiwan).

Hematoxylin and Eosin (H&E) staining

   Mouse tongue tissues were collected, fixed in 10% formalin, dehydrated
   and embedded in paraffin. 5 μm-thick tissue sections were sliced and
   mounted on slides for histological and immunohistochemical (IHC)
   analysis. Tissue sections were deparaffinized in xylenes 5 minutes
   twice, 99% ethanol 2 minutes, 95% ethanol 1 minute, 85% ethanol 1minute
   and 75% ethanol 1 minute in order and washed by ddH[2]O for 1 minute.
   After ddH[2]O washing, hematoxylin was used to stain the nuclear parts
   of tissue sections for 15 minutes and washed for 20 minutes. Then,
   eosin was used to stain the cytoplasm parts of tissue sections for 1
   minute and washed for 20 minutes. Images of slides were taken using
   Olympus BX51 microscope at a magnification of 10X.

Immunohistochemical staining and assessment

   Serial 5-μm histological sections were analyzed using
   immunohistochemical staining with polyclonal anti-SMOX antibody at the
   dilution of 1:300 (ab150971, Abcam Inc., MA, USA). Briefly, unstained
   tissue sections were deparaffinized and rehydrated. By using the
   heat-induced epitope retrieval method, the tissue sections in retrieval
   solution (sodium citrate buffer 10 mmol/L, pH 6.0) were heat at 100 °C
   for 1 hour. Endogenous peroxidase was blocked by 3% hydrogen peroxide
   solution. Tissue sections were incubated with primary antibody
   overnight at 4 °C. Secondary antibodies were incubated for 1 hour using
   a Starr Trek Universal HRP Detection Kit (STUHRP700L10 kit, Biocare
   Medical, Concord, CA). For the negative controls, the primary antibody
   was omitted. DAB was used for examining the signal with hematoxyllin
   counterstaining. Slide examination was carried out using an Olympus
   BX51 light microscope. Expression of SMOX in the cytoplasm of tumor
   cells or tongue epithelial cells was recorded. Positive immune-stained
   cells were counted manually by two independent investigators. The
   following parameters were used to determine the degree of expression of
   immunostaining: if less than 10% of cells stained, the specimen was
   considered negative, while 10%-30%, 31%-70%, and 71%-100% were
   classified as weak, moderate, and strong expression, respectively.

Results

Antibiotic-induced Microbiome Dysbiosis Decreases 4-NQO/arecoline-induced
Murine Oral Tumorigenesis

   To establish the animal model of oral cancer, 4-NQO and arecoline were
   added to drinking water to induce tumorigenesis in the oral cavity of
   mice. Figure [76]1A shows a schematic diagram of the experimental
   protocol of antibiotic treatment and 4-NQO/arecoline-induced oral
   cancer in mice (Figure [77]1A). C57BL/6 mice were randomly divided into
   the control and antibiotic-treated groups, with five male and five
   female mice in each group. After histological examination, we divided
   the mouse tongue tissue lesions into OSCC-free (including no lesion,
   leukoplakia, epithelial hyperplasia, and tumor in situ) and OSCC groups
   (Figure [78]1B). We found that there were six and nine OSCC-free mice
   in the control and antibiotic groups, respectively, and one of the
   antibiotic-treated mice did not develop any tumors (Supplementary
   [79]Figure S1 and Figure [80]1C). Interestingly, invasive OSCC occurred
   only in the control mice (n = 4) and not in any of the antibiotic group
   mice (Supplementary [81]Figure S1 and Figure [82]1C). Although Fisher's
   exact test analysis showed no statistically significant difference in
   oral cancer lesions between the control and antibiotic mice (p =
   0.0867), the incidence of oral cancer lesions appeared to be reduced in
   the antibiotic-treated mice (Figure [83]1C).

Changes in the Microbiome in Mice Feces and Saliva after Antibiotic Treatment

   To determine microbial changes in the antibiotic-treated mice, we
   analyzed the V3-V4 hypervariable region of the 16S rRNA gene in murine
   feces and saliva. Rarefaction curves of operational taxonomic units
   (OTUs) indicated lower species richness and species evenness in the
   antibiotic-treated murine feces compared with the control mice (Figure
   [84]2A). However, the species richness in the murine saliva specimens
   did not change significantly with antibiotic treatment (Figure [85]2B).
   Bacteroidetes (65.0%) was the most abundant phylum, followed by
   Firmicutes (30.4%), Verrucomicrobia (2.0%), and Tenericutes (1.0%) in
   the feces of control mice (Figure [86]2C). After antibiotic treatment,
   the two increased bacterial phyla in the mouse feces were
   Verrucomicrons (2.0% vs. 26.0%) and Proteobacteria (1.0% vs. 13.4%),
   while the decreased bacterial phylum were Bacteroidetes (65.0% vs.
   54.1%) and Firmicutes (30.4% vs. 6.4%) (Figure [87]2D). Firmicutes
   (79.9%) was the most abundant phylum in the saliva of the control mice,
   followed by Actinobacteria (9.2%), Proteobacteria (6.6%), and
   Bacteroidetes (3.1%) (Figure [88]2E). Intriguingly, similar to the
   mouse feces specimens, anti-biotic treatment also increased the
   abundance of Verrucomicrons (0.1% vs. 3.2%) in mouse saliva, but the
   increase in Firmicutes abundance (79.9% vs. 84.0%) was opposite to the
   change in mouse feces (Figure [89]2F).

Figure 2.

   [90]Figure 2
   [91]Open in a new tab

   Changes in the microbiome in mice feces and saliva after antibiotic
   treatment. (A) Rarefaction curves of operational taxonomic units (OTUs)
   diversity for the stool samples. Compared with the control group, the
   number of OTUs in the antibiotic group was significantly reduced. (B)
   Rarefaction curves of OTUs diversity for the salivary samples. (C) The
   percentage of bacterial phyla in the feces of mice. The top five
   bacterial phyla are shown. (D) The percentage change in the top four
   bacterial phyla in the feces of mice in the antibiotic group and the
   control group. (E) The percentage of bacterial phyla in the saliva of
   OSCC-free mice. The top five bacterial phyla are shown. (F) The
   percentage change in the top four bacterial phyla in the saliva of mice
   in the antibiotic group and the control group.

Characterization of the Microbiome in the Saliva of Oral Tumor-bearing Mice

   We further sought to determine the taxa in murine saliva that
   contribute to tumor formation; intergroup differences at the species
   level in each group were analyzed by the linear discriminant analysis
   (LDA) effect size (LEfSe) method [92]^30. The salivary microbial
   communities in mice with oral tumors were first compared to those in
   mice without oral tumors. A histogram of the LDA scores showed
   differential abundance in the saliva between tumor-bearing mice and
   tumor-free mice. LEfSe revealed 23 species that might ex-plain
   differences between the tumor-bearing and tumor-free mice (Figure
   [93]3). Among those, Lactobacillus animalis was the most depleted
   species in the tumor-bearing group. This species was also found
   associated with OSCC using MaAsLin2 (Supplementary [94]Figure S2 and
   [95]Table S1). Interestingly, we found that the same a bacterial
   species L. animalis was also more enriched in the antibiotic treatment
   group (Supplementary [96]Figure S3), indicating that antibiotics may
   reduce the incidence of oral cancer due to the increase in L. animalis.

Figure 3.

   [97]Figure 3
   [98]Open in a new tab

   Characterization of microbiomes in the saliva of OSCC-bearing mice.
   Linear discriminant anal-ysis (LDA) effect size (LEfSe) analysis
   performed on the microbial community relative abundance data in the
   saliva microbiota of the control group's OSCC versus OSCC-free mice.

Changes in the Metabolic Pathways of the Oral Tumor-associated Microbiome

   Bacterial metabolites are important regulators and may exert important
   influences on the biological functions of host cells. Here, we further
   analyzed the changes in the metabolic pathways of the microbiome in the
   saliva of tumor-bearing mice. Kyoto Encyclopedia of Genes and Genomes
   database (KEGG) pathway enrichment analysis of differentially microbial
   communities from OSCC-bearing mice (T1-T3) vs. OSCC-free mice (N1-N3)
   was performed. The results showed that the amino acid (such as
   methionine, lysine, iso-leucine, and histidine) and polyamine (such as
   arginine, ornithine, and nonspermidine) biosynthesis pathways were most
   enriched in OSCC-bearing mice (Figure [99]4).

Figure 4.

   [100]Figure 4
   [101]Open in a new tab

   Predicted functional composition of metagenomes based on 16S rRNA gene
   sequencing data of murine salivary samples. Heat map of differentially
   abundant Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways
   identified in the six specimens. The values of color in the heat map
   represent the normalized relative abundance of KEGG pathways.

Oral Tumor-associated Microbiome Altered Polyamine Biosynthesis May Promote
the Expression of Spermine Oxidase (SMOX) in Murine Oral Cancerous Tissues

   The analysis of metabolic pathways in the salivary microbiome showed
   that the biosynthesis of polyamines was increased in the saliva
   microbiome of mice with oral cancer, so we were interested in whether
   the expression of polyamine metabolism-related enzymes in mouse oral
   cancer tissues is also changed. Spermine oxidase (SMOX) is a polyamine
   metabolizing enzyme that catalyzes the conversion of spermine to
   spermidine and is known to mediate Helicobacter pylori-induced gastric
   cancer [102]^34. Here, we carried out immunohistochemistry (IHC) for
   SMOX on murine oral mucosal tissues. We found that SMOX immunostaining
   increased with the progression of oral cancer; it was barely expressed
   in normal tongue epidermal tissues, but showed moderate staining in
   precancerous lesions and strong staining in cancerous tissues (Figure
   [103]5). We also found that the SMOX IHC was more densely stained in
   the marginal areas of the tongue cancer tissues, suggesting that cells
   located on the tumor surface may be exposed to external bacteria to
   induce SMOX expression. We also found that SMOX IHC staining in
   leukoplakia tissues from antibiotic-treated mice was weaker than in
   tissues from control mice (Supplementary [104]Figure S4).

Figure 5.

   [105]Figure 5
   [106]Open in a new tab

   Immunohistochemical staining of spermine oxidase (SMOX) in mouse normal
   tongue and tongue cancer tissues. The rectangular images on the upper
   panel are enlarged and displayed in the lower panel. The stars,
   arrowheads, and arrows indicate normal, precancer, and cancer tissues,
   respectively. Scale bars in upper panel = 500 µm, scale bars in lower
   panel = 100 µm.

Discussion

   Increasing our understanding of the role of the microbiota in
   carcinogenesis will give new perspectives for future cancer treatment
   and prevention strategies. In this study, we used a mouse model of oral
   cancer induced by 4-NQO/arecoline to provide evidence that the
   dysbiosis of the microbiota affects the incidence of oral cancer. This
   is the first report on the impact of antibiotics-induced dysbiosis or
   depletion of the oral microbiome on oral carcinogenesis and finds a
   cancer-promoting role of the oral microbiome. We determined the
   characteristics of the oral microbiota and its associated metabolic
   pathways of tumor-bearing mice and proved that tumor-related
   microorganisms might upregulate the expression of polyamine
   metabolizing enzyme SMOX in the host cells.

   In this study, we used a mixture of streptomycin, colistin, and
   ampicillin as a broad-spectrum antibiotic regimen that targeted both
   Gram-positive and Gram-negative bacteria, in order to clarify the
   microbiota of oral carcinogenesis. Broad-spectrum antibiotic treatment
   has been used for the purpose of very significantly ablating the gut
   microbiota [107]^13. Most animal studies using antibiotics to deplete
   the microbiota have illustrated the role of the microbiome in promoting
   different types of cancer. For instance, Ma et al. found that depleting
   gut commensal bacteria enhanced the primary bile acids in the liver,
   causing natural killer T cells to accumulate and thereby preventing
   liver tumorigenesis [108]^35. Antibiotic-mediated gut microbiome
   depletion also significantly reduced the tumor burden in several
   genetically engineered and xenograft mouse models, such as pancreatic
   cancer, colon cancer, lung cancer, and melanoma [109]^36^,[110]^37. On
   the other hand, microbial antitumoral effects have also been reported
   [111]^38. Bacterial toxins and metabolites can prevent tumor growth and
   activate the immune system, leading to reduced tumor development
   [112]^38. In oral cancer, using germ-free animal models to evaluate the
   role of the microbiome in 4-NQO-induced oral cancer, that it did not
   differ in tumor incidence, diversity, and size, the authors also
   suggested that the dose of the carcinogen may affect the ability to
   draw conclusions [113]^15. Stashenko et al. recently reported that
   directly transplanting the oral microbiome of donor mice (whether
   normal mice or oral cancer patients) into germ-free mice can promote
   4-NQO-induced oral tumors [114]^16. This observation is consistent with
   our results and proves that oral microbes can promote the formation of
   oral cancer. Moreover, we found that antibiotic treatment has no
   significant effect on the richness of oral microbes, but oral tumors
   seem to reduce the richness of oral microbes. A decrease in the
   richness and diversity of microbes has been observed in the cancer
   tissues and saliva of clinical human oral cancer patients [115]^39, and
   it has also been found in the saliva specimens of patients with oral
   precancerous lesions [116]^40.

   The present study demonstrated, for the first time, that L. animalis
   was more enriched in the saliva of tumor-free mice and the antibiotic
   treatment groups, suggesting that this bacterial species may be
   associated with the prevention of oral cancer development. The
   anti-cancer activity of Lactobacillus strains has received considerable
   attention as a beneficial microbiota. For example, culture supernatants
   of L. acidophilus exhibited a significant inhibitory effect on the
   proliferation of human breast cancer cells and tumor growth in
   xenograft mouse models [117]^41. The treatment with L. bulgaricus
   decreased the tumor volumes in colitis-associated colon cancer and
   attenuated intestinal inflammation by suppressing the production of
   IL-6, TNF-α, IL-17, IL-23, and IL-1β [118]^42. Lactobacillus-driven
   down-regulation of the Wnt/β-catenin pathway-related genes was
   accompanied by tumor growth inhibition, which also suggests that this
   probiotic can be used as a clinical supplement for the prevention and
   treatment of colon cancers [119]^43. In the oral cavity, Lactobacillus
   also acts against oral pathogens associated with periodontitis and
   caries and even induces apoptosis in oral cancer cells
   [120]^44^,[121]^45. Although we have identified potential oral
   microorganisms associated with oral tumorigenesis in mice tumor models,
   the impact of these bacterial strains on the formation of human oral
   cancers needs to be further elucidated in the future.

   Bacterial metabolites are important regulators of host cell function
   and may play a critical role in the physiological and pathological
   conditions of the host. In the present study, we found that the
   biosynthesis of polyamines was increased in the saliva microbiome of
   mice with oral cancer. Moreover, the expression of the polyamine
   metabolizing enzyme SMOX increased with the degree of oral tissue
   malignant transformation, and there are also more expressions on the
   surface of oral mucosal tissues that may be frequently exposed to
   foreign bacteria. A recent study through the integration of
   metabolomics analysis and transcriptomic data also found that the
   polyamine pathway is significantly dysregulated in human OSCC,
   including the upregulation of SMOX [122]^46. Although our data
   suggested that the increase in SMOX in oral cancer cells may be due to
   frequent exposure to bacteria with polyamine metabolism, the effect of
   microorganisms in the up-regulation of SMOX needs to be further
   investigated. The polyamines are produced through a continuous process
   that starts with the conversion of L-ornithine to putrescine, and then
   putrescine is metabolized to spermidine and spermine. SMOX specifically
   back-converts spermine to spermidine, generating H[2]O[2] in the
   process, which leads to DNA damage and subsequent tumorigenesis
   [123]^18. The induction of SMOX by Helicobacter pylori infection
   mediates H. pylori-induced gastric inflammation, DNA damage, and
   activation of β-catenin signaling in gastric cancer [124]^34.
   Microbiota and their metabolites participate in the carcinogenic
   process through different pathological mechanisms, including induction
   of epithelial cell gene mutations and regulation of the tumor
   microenvironment [125]^47^,[126]^48. Further investigating the direct
   and indirect mechanisms by which the oral microbiota can influence the
   metabolism in oral keratinocytes as well as the interplay between the
   host and microbiota metabolic pathways will help with the development
   of oral cancer treatment and prevention by regulating the microbiota.

Supplementary Material

   Supplementary figures and table.
   [127]Click here for additional data file.^ (4.9MB, pdf)

Acknowledgments