Graphical abstract

   graphic file with name fx1.jpg
   [63]Open in a new tab

Highlights

     * •
       Cigarette smoke exposure (CSE) induces unique gut microbial and
       metabolomic signatures
     * •
       Gut microbial dysbiosis induced by CSE is essential for its
       tumor-promoting effects
     * •
       Fecal microbiome transfer from CSE mice into smoke-naïve mice
       drives increased tumor growth
     * •
       Targeting the dysbiotic gut microbiome ameliorates tumor-promoting
       effects of CSE
     __________________________________________________________________

   Biological sciences; Immunology; Microbiome; Cancer; Metabolomics

Introduction

   Cigarette smoke exposure (CSE) has been extensively investigated as a
   risk factor for cancer development.[64]^1^,[65]^2 Potential carcinogens
   detected in cigarette smoke[66]^3 have been postulated to aid cancer
   initiation and progression through genotoxic effects and reactive
   oxygen species generation.[67]^4^,[68]^5 Recent studies have also
   explored the direct effects on cell proliferation through binding to
   nicotinic acetylcholine receptors (nAChRs).[69]^6 However, the
   molecular mechanisms underlying the tumor-promoting effects of CSE
   remain elusive, and there is a dearth of clinically actionable targets.

   We have witnessed the recent emergence of the gut microbiome as a key
   regulator of human physiology, and its perturbations linked to ailments
   such as gastrointestinal diseases,[70]^7 neurological diseases,[71]^8
   behavioral conditions,[72]^9 and
   malignancies.[73]^10^,[74]^11^,[75]^12^,[76]^13^,[77]^14^,[78]^15
   Recent research has demonstrated that smokers, too, harbor a dysbiotic
   microbiome.[79]^16^,[80]^17^,[81]^18 However, the implications of this
   changed gut microbiota have not yet been explored. We, along with
   others, have shown that the gut microbiome significantly drives cancer
   progression by affecting the anti-tumor immune
   response.[82]^14^,[83]^19 Targeting the microbiome as a strategy to
   enhance the efficacy of immunotherapy in multiple cancers is being
   explored.[84]^20 In light of these findings, we hypothesized that the
   gut microbiome might be the missing link between smoking and cancer
   progression. Herein, we elucidate a novel mechanism for
   smoking-dependent cancer progression via the gut microbiome through
   modulation of the adaptive immune system. We characterize the unique
   microbiome and metabolome potentiating this smoking-dependent cancer
   progression and present an actionable target to mitigate the harmful
   effects of cigarette smoke.

Results

The gut microbiome is required and sufficient to mediate
cigarette-smoke-induced cancer progression

   We first evaluated the effect of gut microbiome depletion on cigarette
   smoke exposure (CSE)-induced tumor progression. CSE was given to
   cancer-naive C57BL/6J wild-type (WT) mice with and without a
   broad-spectrum antibiotic cocktail to deplete the gut microbiome
   followed by subcutaneous cancer cell implantation ([85]Figure 1A). CSE
   promoted tumor growth in KPC (pancreatic cancer), MC-38 (colon cancer),
   and MB-49 (bladder cancer) models in the presence of gut microbiome
   ([86]Figures 1B–1D, left panels; [87]Figure S1A). Interestingly, in
   gut-microbiome-depleted states, CSE failed to promote tumor growth
   ([88]Figures 1B–1D, right panels). In another set of experiments, when
   the smoke analog nicotine-derived nitrosamine ketone (NNK;
   N-methyl-N-(4-oxo-4-pyridin-3-ylbutyl)nitrous amide), one of the most
   potent carcinogens in cigarette smoke,[89]^21 was administered
   intraperitoneally, gut microbiome depletion was still able to nullify
   any tumor-promoting effects across multiple cancer models
   ([90]Figures S1B–S1D).

Figure 1.

   [91]Figure 1
   [92]Open in a new tab

   The gut microbiome is required and sufficient to promote
   cigarette-smoke-induced cancer progression

   (Also refer to [93]Figure S1) (A) Schematic timeline of experimental
   design of the subcutaneous tumor model in C57BL/6J mice showing the
   pre-exposure phase of 4 weeks before tumor implantation followed by
   continued exposure of CSE with or without gut microbiome depletion.

   (B–D) Subcutaneous tumor volumes at the endpoint in mice exposed to CSE
   with or without gut microbiome depletion using (B) pancreatic cancer
   (data representative of three independent experiments), (C) colon
   cancer, and (D) bladder cancer cell lines, respectively (n = 8–12 per
   group). (Left panels (B–D): gut microbiome present; right panels (B–D):
   gut microbiome depleted state).

   (E) Pancreatic weights of LSL-Kras^G12D/+;Pdx-1-Cre (KC mice) at
   4 months of age after exposure with NNK for 2 months with or without
   gut microbiome depletion. (n = 11–12 per group).

   (F) (Top panel) Representative H&E stain of KC mice pancreas showing
   significant ductal proliferation and fibrosis in the pancreas of mice
   exposed to NNK compared to controls, gut microbiome depleted, and gut
   microbiome depleted NNK-exposed mice (scale bar, 563.6 μm); (Middle
   panel) Trichrome staining showing fibrosis in the pancreas of KC mice.
   Quantification (shown on the left) done as described in methodology
   (n = 4 mice per group; scale bar, 563.6 μm); (bottom panel)
   cytokeratin-19 staining showing ductal proliferation among various
   groups. (n = 4 per group; scale bar, 281.8 μm).

   (G) Tumor kinetics of control and CSE mice with CSE mice implanted with
   Mc-38 cell line randomized at day 14 to receive no antibiotics vs.
   broad-spectrum antibiotics. (n = 10 controls: n = 20 smoke-exposed
   before randomization at day 14).

   (H) Schematic timeline of FMT studies in C57BL/6J mice randomized to
   receive FMT from control mice or CSE/NNK-exposed mice.

   (I) Subcutaneous tumor volumes at the endpoint in recipient mice
   getting FMT from control donors vs. CSE donors (n = 12 per group). Data
   representative of two independent experiments. CSE: cigarette smoke
   exposure; NNK: 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone; FMT:
   fecal microbiome transplant. ∗ p value <0.05; ∗∗p value <0.01; ∗∗∗ p
   value <0.001. Unpaired t test or ANOVA used as appropriate for
   statistical comparison. Data presented as mean ± SEM.

   Next, we utilized genetic models with a predisposition to cancer
   formation to test our hypothesis ([94]Figure S1E). NNK exposure to KC
   (LSL-Kras^G12D/+;Pdx-1-Cre) mice, which are predisposed to spontaneous
   pancreatic intraepithelial neoplasia (PanIN) and pancreatic cancer
   development, resulted in increased mean pancreatic weights compared to
   mice receiving simultaneous NNK and broad-spectrum antibiotic treatment
   ([95]Figure 1E). Histology showed significant fibrosis and increased
   CK-19 expression in the NNK-exposed group relative to the control
   group; however, simultaneous treatment with antibiotics to NNK-exposed
   mice significantly reduced fibrosis and ductal proliferation
   ([96]Figure 1F). Similarly, A/J mice, which are highly susceptible to
   forming carcinogen-induced lung tumors, developed significantly more
   tumor nodules in the lungs of NNK-exposed mice with intact gut
   microbiomes when compared to those with depleted gut microbiomes
   ([97]Figure S1F). To confirm whether this effect could be replicated in
   the setting of well-established tumors, we employed a therapeutic model
   of gut microbiome depletion where subcutaneous tumors were allowed to
   grow for 14 days under the influence of CSE before antibiotic depletion
   was initiated. The broad-spectrum antibiotic cocktail was able to
   significantly retard tumor growth in mice receiving CSE after only
   5 days of treatment ([98]Figure 1G).

   To further evaluate if smoking-induced gut microbial dysbiosis is
   sufficient to cause smoke-induced tumor progression, we performed fecal
   microbiota transplants (FMTs) from tumor-naive control (smoke-free) WT
   donor mice and cigarette-smoke-exposed WT donor mice into recipient WT
   mice with depleted gut microbiome ([99]Figure 1H). These recipient mice
   with reconstituted gut microbiomes were then challenged with
   subcutaneous KPC pancreatic cancer tumors. Mice receiving FMT from
   smoke-exposed mice had significantly higher mean tumor volumes than
   mice receiving FMT from control mice ([100]Figure 1I). To further
   validate this finding, FMT was performed with NNK-exposed mice instead
   of CSE mice with similar results ([101]Figure S1G). We measured serum
   cotinine concentrations (a byproduct of nicotine metabolism, which has
   been shown to correlate with smoke exposure[102]^22) in the donor mice
   as well as recipient mice post-FMT to ensure that smoke metabolites
   were not being passively transferred through the FMT. Serum cotinine
   levels were significantly elevated only in the CSE-exposed donor mice
   (25.8 ng/mL), consistent with the levels reported in the literature for
   mice undergoing smoke exposure.[103]^5 Mice receiving FMT from smoke
   donors had relatively undetectable levels (<0.1 ng/mL), making the
   transmission of nicotine through donor stool unlikely
   ([104]Figure S1H). In addition, antibiotics also did not affect serum
   cotinine levels ([105]Figure S1I). A recent report has indicated that
   gut microbes can metabolize and degrade nicotine.[106]^23 To verify
   whether antibiotics treatment was affecting nicotine levels in the gut,
   we analyzed the nicotine levels in ileal contents. Although mice
   exposed to cigarette smoke had higher levels of nicotine in the ileal
   contents, these remained unchanged in mice where gut microbiome was
   sterilized with antibiotics, indicating that nicotine metabolism was
   not affected by gut microbiome in our model ([107]Figure S1J). These
   findings indicate that the gut microbiome is required and sufficient to
   induce smoking-dependent tumor progression.

Modulation of adaptive and innate immune response drives the
smoking-induced-dysbiosis-mediated cancer progression

   We next characterized the immune tumor microenvironment (TME) of
   subcutaneous tumors from control and smoke-exposed mice with or without
   gut microbiome depletion through flow cytometry. Smoke-exposed mice had
   a globally attenuated anti-tumor immune response with decreased CD3^+
   lymphocytes, CD4^+ T cells, CD8^+ T cells, and CD11c+ MHC II +
   dendritic cells. On the other hand, tumor-promoting moieties, such as
   myeloid-derived suppressor cells (MDSCs [CD45^+ CD11b+ Ly6G+]), were
   upregulated. The M1 (CD45+F4/80+MHCII+CD206-) to M2
   (CD45+F4/80+MHCII-CD206+) macrophage ratio was tilted significantly in
   favor of the M2 phenotype ([108]Figures 2A–2C and [109]S2A–S2C). Gut
   microbial depletion was able to rescue the anti-tumor immune machinery
   with downregulation of MDSCs and upregulation of antigen-presenting
   dendritic cells, tumor-infiltrating CD4^+ T cells, and CD8^+ T cells;
   however, the macrophage polarization could not be reversed
   ([110]Figure S2C). Intriguingly, this immunomodulation of the TME was
   also evident in mice receiving FMT from NNK-exposed donors.
   Immunophenotyping of tumors implanted in NNK FMT recipient mice
   revealed reduced infiltration of CD3^+ T cells, CD4^+ T cells, and
   CD8^+ T cells, along with significantly elevated MDSCs
   ([111]Figures 2D–2F and [112]S2D). Th17 T cells (CD3^+CD4+IL17+), known
   to potentiate an immunosuppressive phenotype in the tumor
   microenvironment, increased with NNK FMT but did not reach statistical
   significance ([113]Figure S2E). Moreover, tumor growth inhibition seen
   with therapeutic antibiotic administration to smoke-exposed mice was
   accompanied by significantly increased CD8^+ T cell infiltration. MDSCs
   (CD45^+CD11b+Ly6G+) infiltration in the TME simultaneously decreased;
   however, it did not reach statistical significance ([114]Figures S2F
   and S2G). These findings indicate that gut microbiome depletion rescues
   the anti-tumor adaptive immune machinery in the TME.

Figure 2.

   [115]Figure 2
   [116]Open in a new tab

   Modulation of adaptive immune response drives the smoking-induced
   dysbiosis-mediated cancer progression

   (Also refer to [117]Figures S2 and [118]S3) (A–C) Graphical
   representation of tumor flow-cytometric analysis (left) and bar graphs
   (right) of control, CSE, antibiotics alone, and CSE + antibiotics group
   showing percentage infiltration of CD3CD4+ T cells (A), CD3CD8+ T cells
   (B), and CD11b+Ly6G + MDSC (C) cells of single cells in the tumor
   microenvironment, respectively, showing CSE-induced immunosuppressive
   changes reversed with gut microbiome depletion (n = 7–10 per group).

   (D–F) Tumor flow-cytometric analysis showing immunosuppressive changes
   characterized by decreased CD3CD4+ T cells (D), CD3CD8+ T cells (E),
   and increased CD11b+Ly6G + MDSC cells (F), respectively, induced in
   mice that received FMT from NNK-exposed mice vs. controls (n = 7–8 per
   group).

   (G) Tumor volumes in WT and Rag1-KO mice exposed to CSE with or without
   gut microbiome depletion with broad-spectrum antibiotics (n = 7–10 per
   group).

   (H) Tumor volumes in WT and Rag1-KO mice receiving FMT from smoke
   donors (smoke FMT) vs. control donors (control FMT) (n = 9–12 for WT
   mice, n = 8–9 for Rag1-KO mice).

   (I) Tumor volumes in WT (n = 14–15) and CD8-KO mice (n = 7–8) exposed
   to CSE.

   (J) Tumor kinetics in WT CSE mice with or without Ly6G-depleting
   monoclonal antibody with appropriate controls (n = 10 per group). ∗p
   value <0.05; ∗∗p value <0.01; ∗∗∗p value <0.001. Unpaired t test or
   ANOVA used as appropriate for statistical comparison. Data presented as
   mean ± SEM.

   To confirm the importance of the adaptive immune system, we first
   utilized the Rag1-KO mouse model, which lacks a functional adaptive
   immune system.[119]^24 We found that CSE did not promote tumor growth
   in Rag1-KO mice; additionally, in these mice, gut microbiome depletion
   failed to decrease the tumor burden ([120]Figure 2G). Administration of
   NNK instead of CSE to Rag1-KO mice yielded similar results
   ([121]Figure S2H). This indicates that the adaptive immune system is
   essential for smoking-induced gut microbial dysbiosis to affect tumor
   progression.

   To eliminate the possibility that potential differences in the baseline
   microbiome of Rag1-KO mice, when compared to the WT mice, could be
   confounding our results, we reconstituted the gut-microbiome-depleted
   Rag1-KO mice with FMT from either smoke-exposed or smoke-free WT mice.
   The tumor-promoting effect of the smoke-exposed dysbiotic gut
   microbiome was abrogated in Rag1-KO mice ([122]Figure 2H). To
   specifically delineate the role of the CD8 adaptive immune arm, we
   evaluated the effect of smoking in CD8-KO mice and found CSE did not
   promote tumor growth, signifying that the effector immune cells are
   essential links for the dysbiotic microbiome to mediate its effects
   ([123]Figure 2I). To test whether antibiotics treatment led to
   increased cytotoxicity in CD8^+ T cells, we performed an ex vivo
   cytotoxicity study by incubating splenic CD8^+ T cells, isolated from
   tumor-bearing mice, with calcein-AM-labeled KPC cells (see [124]STAR
   Methods). CD8^+ T cells isolated from CSE mice treated with antibiotics
   had significantly greater cytotoxicity against KPC cells ex vivo, as
   compared to CSE mice without antibiotics ([125]Figure S3A).
   Additionally, Ki67 staining of tumors obtained from CSE + antibiotics
   mice demonstrated less cancer cell proliferation as compared to CSE
   mice ([126]Figure S3B), indicating that increased CD8^+ T cell
   infiltration and cytotoxicity may be leading to decreased cancer cell
   proliferation. As smoke-exposed tumors showed increased accumulation of
   CD11b+Ly6G + cells, which have been previously shown to decrease
   cytotoxic CD8^+ T cell infiltration in the TME,[127]^25 we hypothesized
   that these cells were responsible for decreased effector CD8 T cell
   tumor infiltration in smoke-exposed mice. Specifically targeting these
   cells using anti-Ly6G monoclonal antibody (Invivomab Catalog# BE0075-1,
   Clone 1A8), with cell depletion confirmed on flow cytometry of
   splenocytes ([128]Figure S3C), prevented the tumor-promoting effects of
   smoking ([129]Figure 2J).

Smoke-induced dysbiosis is characterized by unique microbial and metabolomic
signatures amenable to selective targeting

   Gut microbial profiling by 16S rRNA sequencing revealed that the CSE
   group clustered separately compared to the control group upon principal
   coordinate analysis (PCoA) ([130]Figure S4A). Linear regression
   analysis showed significant enrichment of gram-negative genera
   (Bacteroides and Akkermansia) and certain gram-positive genera
   (Turicibacter, Acetatifactor, and Blautia) in the CSE group
   ([131]Figure S4B). Similarly, NNK exposure induced a significant
   difference in beta diversity ([132]Figure S4C), and interestingly,
   gram-negative genera like Bacteroides and Akkermansia were enriched
   upon NNK exposure as well ([133]Figure S4D).

   When the gut microbiome composition of mice bearing subcutaneous tumors
   was evaluated, the CSE group again demonstrated a distinct gut
   microbial signature. In KPC subcutaneous tumor-bearing mice, bacteria
   belonging to Akkermansia spp., Turicibacter spp., Faecalibaculum spp.,
   and Pseudomonas spp. were the most enriched in the CSE group
   ([134]Figure 3A). Furthermore, CSE led to a significantly different gut
   microbial composition in mice bearing subcutaneous MC-38 colon cancer
   ([135]Figure S4E) and MB-49 bladder tumors ([136]Figure S4G).
   Interestingly, despite a different tumor microenvironment, bacteria
   from genus Akkermansia were the enriched bacteria in the gut in the CSE
   group in both tumor types, similar to KPC pancreatic tumors
   ([137]Figures S3F and [138]S4H). Additionally, tumor-bearing mice that
   were receiving smoke FMT had significant enrichments of bacteria from
   the gram-negative genera Prevotella, Parabacteroides, and
   Alloprevotella compared to mice that received control FMT
   ([139]Figure 3B).

Figure 3.

   [140]Figure 3
   [141]Open in a new tab

   Smoke-induced dysbiosis is characterized by unique microbial and
   metabolomic signatures amenable to selective targeting

   (Also refer to [142]Figures S4 and [143]S5) (A) Fecal samples were
   obtained from subcutaneous KPC tumor-bearing control and CSE mice and
   analyzed using 16s rRNA amplicon sequencing for metagenomic
   characterization. (Left panel) The significantly enriched microbes
   based on linear model analysis in the CSE (green bars) and control (red
   bars) groups are shown. (Right panel) PCoA plot comparing beta
   diversity between control (red) and CSE (green) groups. Distance was
   calculated using Bray-Curtis analysis, and PERMANOVA was used for
   statistical significance.

   (B) Fecal samples obtained from C57BL/6J mice receiving FMT from
   control or CSE donors and implanted with KPC pancreatic tumors were
   similarly interrogated through 16 s rRNA sequencing. Linear model (left
   panel) and Bray-Curtis analysis (right panel) are shown.

   (C) Heatmap representation of all differential fecal metabolites (BH
   FDR <0.25) between tumor-bearing control (n = 10) and CSE (n = 11)
   mice. A total of 194 differential metabolites were obtained upon
   untargeted LC-MS analysis.

   (D) Significantly altered metabolic pathways were obtained using
   pathway enrichment analysis of significantly altered fecal metabolites
   upon CSE using MetaboAnalyst 5.0. Only the metabolites annotated in
   HMDB were used for the analysis (88/194 differential metabolites).

   (E) Subcutaneous tumor volumes at endpoint after CSE with selective
   targeting of gut microbiome using oral neomycin gavage (200 mg/kg) or
   complete gut microbiome depletion using broad-spectrum antibiotics
   cocktail (n = 8–12 per group).

   (F) Flow cytometric analysis of tumors showing % of CD3^+CD8^+ of
   single cells after selective targeting of gut microbiome using oral
   neomycin gavage (200 mg/kg) or broad-spectrum antibiotic cocktail (n =
   7–8 per group). ∗p value <0.05, ∗∗p value <0.01. CSE, cigarette smoke
   exposure; PCoA, principle coordinate analysis; PERMANOVA, permutational
   analysis of variance; FMT, fecal microbiome transplant; LC-MS, liquid
   chromatography-mass spectrometry; BH, Benjamini-Hochberg; FDR, false
   discovery rate; HMDB, human metabolome database. Unpaired t test or
   ANOVA used, as appropriate, for statistical comparison. Data presented
   as mean ± SEM.

   To evaluate the gut microbial metabolomic signature induced by CSE, we
   performed untargeted liquid chromatography-mass spectrometry (LC-MS)
   metabolomics on fecal samples from tumor-bearing control and CSE mice.
   A total of 781 unique metabolites were identified. Unbiased
   hierarchical clustering using a scale-normalized concentration of
   discriminatory metabolites with a false discovery rate <0.25 in control
   and smoke-exposed groups showed that the two groups had distinct
   signatures, with 194 metabolites having a significantly differential
   abundance ([144]Table S1; [145]Figure 3C). Qualitative metabolic set
   enrichment analysis of metabolites that were significantly altered in
   the CSE group compared to the controls revealed that arginine
   biosynthesis and arachidonic acid metabolism were the top metabolic
   pathways that were altered due to CSE ([146]Figure 3D).

   Since CSE led to the enrichment of both gram-positive and gram-negative
   bacterial species, we utilized selective targeting of these microbes
   using narrow-spectrum antibiotics. Neomycin, which has a predominantly
   gram-negative spectrum, significantly reduced tumor burden in CSE mice.
   In contrast, vancomycin (exclusively gram-positive spectrum) showed
   only partial tumor reduction ([147]Figure S5A). Additionally, neomycin
   demonstrated an anti-tumor effect equivalent to the broad-spectrum
   antibiotic cocktail ([148]Figure 3E). Neomycin treatment partially
   ameliorated the metabolomic changes induced due to CSE
   ([149]Figures S5B and S5C). Finally, selective targeting with neomycin
   also reverses the immunosuppressive changes in the TME, as evidenced by
   increased CD8^+ T cell tumor infiltration ([150]Figure 3F).

Discussion

   We report one of the first studies demonstrating the role of gut
   microbiome in propagating pro-tumorigenic effects of CSE across
   multiple smoking-dependent cancers. CSE was associated with an
   immunosuppressive TME, and its effect was neutralized when the gut
   microbiome was depleted. FMT experiments confirmed the tumor-promoting
   and immunosuppressive effects of this CSE-modulated gut microbiome. Our
   results are in line with a recent study by Bai et al., where cigarette
   smoke expedited colon cancer growth in an azoxymethane-induced colitis
   model through enrichment of Eggerthella lenta.[151]^26 Prior
   pre-clinical investigations have focused on alterations in cell
   signaling pathways, such as nicotinic receptor stimulation or
   genetic/epigenetic changes in proto-oncogenes and tumor suppressor
   genes, as mechanistic explanations for smoking-induced carcinogenesis.
   However, in our study, FMT experiments suggest that CSE-induced gut
   microbial dysbiosis can independently exert tumor-promoting effects.
   Moreover, ablation of the gut microbiome in mice receiving CSE led to
   abrogation of tumor growth, whereas the levels of smoke metabolites in
   serum (cotinine) or stool (polyaromatic hydrocarbons) were unaffected.
   This suggests that even in the presence of circulating CSE components,
   the gut microbiome plays an essential role in promoting tumor growth.

   Exposure to smoke, or FMT from CSE mice, led to significant
   upregulation of granulocytic MDSCs in the TME, and this
   immunosuppressive phenotype was reversed with gut microbiome depletion.
   In addition, MDSC-specific marker antibody depletion was able to
   reverse the tumor-promoting effects of smoking, implicating a possible
   mechanistic link between the dysbiotic gut microbiome induced by CSE
   and myeloid cell infiltration. Interestingly, Pushalkar et al.
   similarly demonstrated that the gut microbiome can modulate MDSC
   infiltration in the TME in KC genetic model of pancreatic
   cancer.[152]^19 Another dominant immunosuppressive population in
   pancreatic cancer TMEs is tumor-associated macrophages.[153]^27 Prior
   work from Kumar et al. indicates that smoke exposure can induce
   differentiation of Ly6G + MDSCs into tumor-associated macrophages
   (TAMs), which secrete epidermal growth factor receptor (EGFR) ligands
   to accelerate PanIN development in the KC mouse model.[154]^28
   Similarly, we found increased M2-polarized macrophages (F4/80+, MHC
   II−, CD206 +) in the TME upon CSE; however, this could not be reversed
   upon gut microbial ablation, indicating that macrophage polarization
   likely does not explain the effects of gut microbial targeting on tumor
   growth. Additionally, CSE significantly impacted adaptive immune
   response in the TME and decreased CD8 T cell infiltration. These
   findings are concordant with recent human studies in esophageal, head,
   and neck squamous cell cancer, where smokers had decreased CD8 T cell
   infiltration in the TME.[155]^29^,[156]^30 Treatment with antibiotics
   was able to increase CD8^+ T cell infiltration in the TME as well as
   CD8^+ T cell cytotoxicity on ex vivo analysis. The loss of
   pro-tumorigenic effects of CSE in Rag1-KO and CD8-KO mice, along with
   the failure of smoke-FMT to increase tumor growth in Rag1-KO mice,
   suggests the need for functional adaptive immunity for CSE dysbiosis to
   promote tumor growth.

   Characterization of the CSE dysbiotic gut microbiome through 16s rRNA
   sequencing revealed tumor-specific enrichment patterns. However,
   certain microbes, such as Akkermansia spp. and the Clostridiales vadin
   BB60 group, were consistently elevated across tumor types. Akkermansia
   has been favorably associated with immunomodulatory responses in
   obesity and homeostatic immunity but also has been pathologically
   indicated in disease states like multiple
   sclerosis.[157]^31^,[158]^32^,[159]^33^,[160]^34 The gut microbiome is
   known to exert its effects in a context-specific manner.[161]^10 It is
   possible that under the influence of CSE, Akkermansia assumes a
   pro-tumorigenic phenotype. However, further experiments are needed to
   substantiate such conclusions. We did observe significant anti-tumor
   activity with neomycin treatment that might indicate a specific role
   for gram-negative dysbiosis in smoke-mediated tumor progression. There
   have been recent reports regarding accumulation of nicotine in the
   intestinal lumen and its degradation by gut bacteria in the context of
   NASH cirrhosis.[162]^23 Interestingly, in our study, although nicotine
   levels in the luminal contents of ileum increased with smoke exposure,
   we did not observe any differences in nicotine levels upon antibiotics
   treatment. This combined with unchanged serum cotinine levels in
   smoke-exposed mice upon antibiotics treatment may indicate that in our
   model, nicotine metabolism by gut bacterium did not play a significant
   role. This discrepancy may reflect the different biological context,
   when comparing NASH cirrhosis to implanted cancer models. However,
   further studies are needed to establish whether nicotine metabolism by
   gut microbiome is essential for tumor-promoting effects of cigarette
   smoke.

   CSE also led to alterations in composition of stool metabolites that
   are known to exert immunomodulatory effects. CSE was associated with
   significantly altered arachidonic acid metabolism, which was partially
   reversed through neomycin treatment. The arachidonic acid pathway can
   significantly affect the induction of inflammation in both benign and
   malignant disease states. Prostaglandin E2 can induce profound
   immunosuppression via various mechanisms, including infiltration of
   MDSCs in the TME,[163]^35 whereas leukotriene B4 is a potent myeloid
   chemoattractant.[164]^36^,[165]^37 We also observed significant
   alterations in the arginine biosynthesis pathway on CSE. L-arginine is
   essential for T lymphocytes to mount an anti-tumor immune
   response,[166]^38 and depletion of extracellular arginine by
   MDSC-expressed arginase I can lead to immunosuppression in the tumor
   microenvironment.[167]^39 Interestingly, our data indicate
   downregulation of L-arginine upon CSE and subsequent increase upon
   neomycin treatment. Bai et al.[168]^26 identified alterations in
   bile-acid metabolism that could explain the effects of smoking-induced
   dysbiosis on colon cancer progression. In their experiments,
   taurodeoxycholic acid (TCDA), a secondary bile acid, was upregulated
   upon smoke exposure, and the authors posited that this could promote
   cancer growth through the MAP/ERK signaling pathway. In our stool
   metabolomic profiling, we did not observe an enrichment of bile acid
   metabolism in CSE mice. The highly context-specific composition and
   phenotypic effects might explain the differences in the microbes and
   their metabolites identified in our studies. For example, in our study,
   antibiotics treatment alone decreased the tumor burden in the KPC model
   but did not affect the tumor burden in MC-38 or MB-49 models.
   Regardless, antibiotics treatment consistently decreased the
   tumor-promoting effects of CSE in all these models.

   In summary, we demonstrate the induction of a unique metabolic and
   microbial signature upon smoke exposure that creates an
   immunosuppressive tumor microenvironment and favors cancer progression.

Limitations of the study

   Our study does have certain limitations. It is unclear how
   smoking-induced dysbiosis is established and what mechanisms underlie
   the interaction between the dysbiotic gut microbiome and anti-cancer
   immune response. Whether these effects are mediated by specific
   microbes or a group of microbes sharing a specific functional phenotype
   needs to be clarified. For example, we did notice a preponderance of
   gram-negative bacteria upon CSE; however, the mechanisms by which these
   bacteria interact with and modulate the innate and/or adaptive immune
   system has not been explored in our study. Additionally, the role of
   immunomodulatory metabolic pathways altered upon CSE needs to be
   investigated.

Resource availability

Lead contact

   Further information and requests for resources should be directed to
   and will be fulfilled by the lead contact, Vikas Dudeja
   (vdudeja@uabmc.edu).

Materials availability

   No new unique reagents were generated in this study.

Data and code availability

     * •
       Data: the metabolomics data raw files can be accessed through NIH
       metabolomics workbench using Data track ID 4229 and Study ID is
       ST002823. The 16 S rRNA raw sequencing files can be accessed using
       the NCBI Bio project accession number [169]PRJNA1007862. The
       following link can be used to access the data after the release
       date. [170]https://www.ncbi.nlm.nih.gov/sra/PRJNA1007862.
     * •
       Code: the code used for processing and analyzing the 16s rRNA data
       can be accessed at
       [171]https://github.com/asorgen/Smoking_Cancer_Gut_Dysbiosis_Analys
       is_2023.git. All analysis was done using R
       ([172]https://www.r-project.org/).
     * •
       All other data are available upon request from the [173]lead
       contact.

Acknowledgments