Abstract

   Immunotherapy has revolutionized cancer treatment by leveraging the
   patient's immune system, yet its efficacy is often hampered by the
   immunosuppressive tumor microenvironment (TME). Adenosine, a key player
   in this milieu, suppresses immune cell activity via cAMP signaling.
   Here, an innovative strategy to remodel the TME using a genetically
   engineered strain of Escherichia coli Nissle 1917 that expresses
   adenosine deaminase on its surface under hypoxic conditions is
   presented. This engineered probiotic targets tumors, converts
   immunosuppressive adenosine to inosine, and enhances anti‐tumor immune
   responses. In vivo, the engineered probiotic significantly improved
   immune cell infiltration and demonstrated synergistic effects with
   low‐dose doxorubicin in both subcutaneous and orthotopic mouse
   colorectal cancer model. Furthermore, the engineered probiotic
   modulated the TME, promoting a shift from M2‐like to M1‐like
   macrophages and increasing effector T cell populations. These findings
   highlight the potential of using engineered probiotics for metabolic
   modulation of the TME, offering a novel approach for enhancing cancer
   immunotherapy.

   Keywords: adenosine deaminase, chemo‐immunotherapy, engineered
   probiotic, metabolic modulation, tumor microenvironment
     __________________________________________________________________

   This study presents engineered probiotics that convert
   immunosuppressive adenosine into inosine in tumors, alleviating
   immunosuppression and enhancing antitumor immunity. Demonstrating
   efficacy in murine ectopic and orthotopic tumor models, this approach
   offers a promising probiotic‐based cancer immunotherapy strategy,
   highlighting the potential of microbial therapeutics in oncology.

   graphic file with name ADVS-12-2411813-g003.jpg

1. Introduction

   Immunotherapy has emerged as a promising approach for cancer treatment
   by harnessing the patient's own immune system. This therapeutic
   strategy has shown remarkable success in various cancer types and
   improved survival rates for patients.^[ [46]^1 , [47]^2 , [48]^3 ^]
   However, the immunotherapeutic efficacy can be comprised by the
   immunosuppressive tumor environment.^[ [49]^4 , [50]^5 , [51]^6 ^] The
   adenosine metabolite in tumor milieu plays important roles in the
   establishment of such an immunosuppressive microenvironment. In tumor,
   the released ATP from damaged cell was converted to adenosine by the
   cell surface ecto‐nucleotidases, such as CD39 and CD73, resulting in
   the accumulation of extracellular adenosine.^[ [52]^7 , [53]^8 ^] The
   extracellular adenosine then binds to the adenosine receptors on immune
   cells and activates downstream cyclic adenosine 3′,5′‐monophosphate
   (cAMP) signaling, which inhibits the proliferation, survival,
   activation of immune cells, and facilitates immune evasion by tumors.^[
   [54]^9 ^] Therefore, targeting the adenosinergic pathway is a plausible
   strategy to remodel the immunosuppressive tumor microenvironment (TME),
   and enhancing cancer immunotherapy.^[ [55]^10 , [56]^11 , [57]^12 ,
   [58]^13 , [59]^14 ^] Recently, several studies demonstrated that a
   repurposed drug, PEGylated adenosine deaminase (PEG‐ADA) could deplete
   extracellular adenosine by enzymatic converting adenosine to inosine,
   which subsequently stimulates T cell proliferation, promotes tumor
   infiltration and inhibits tumor growth, indicative of the potential of
   adenosine deaminase (ADA) as a promising therapeutic enzyme for cancer
   treatment.^[ [60]^15 , [61]^16 , [62]^17 ^] However, relatively high
   concentration of PEG‐ADA must be administrated due to the lack of tumor
   specificity, resulting in low therapeutic efficacy and even side
   effect.^[ [63]^18 ^] Therefore, alternative means to locally and
   continuously increase the effective concentration of ADA in tumors are
   highly desirable.

   Bacteria have attracted increasing attention in cancer treatment,
   benefiting from their inherent features, such as self‐propelled
   motility, hypoxia tropism, and preferential tumor colonization.^[
   [64]^19 ^] Furthermore, the bacteria could also trigger anti‐tumor
   immune responses due to their intrinsic immunostimulatory nature.
   Another attractive characteristic of bacteria is their capability to be
   modified chemically and genetically. In the past few years, a variety
   of attenuated pathogenic bacteria and probiotics, such as Listeria,
   Clostridium, Salmonella and Escherichia coli Nissle 1917 (EcN) have
   been engineered and extensively used in cancer treatments.^[ [65]^20 ,
   [66]^21 , [67]^22 , [68]^23 , [69]^24 , [70]^25 ^] Majority of these
   studies focused on the use of bacteria to deliver and release chemical
   or biological payloads.^[ [71]^26 , [72]^27 , [73]^28 , [74]^29 ,
   [75]^30 , [76]^31 , [77]^32 ^] However, engineered bacteria for
   metabolic modulation of immunosuppressive TME has been rarely
   reported.^[ [78]^33 ^]

   In this study, we designed and constructed a therapeutic EcN strain,
   which is capable of reshape the immunosuppressive TME through
   extracellular adenosine depletion. In this EcN strain, the ADA gene and
   the gene encoding a bacterial autotransporter domain (AIDA‐I) were
   fused and were expressed under the control of a hypoxia‐responsive fnrS
   promoter (P [fnrS] ). The constructed EcN strains (EcN_ADA)
   spatiotemporally targets tumors, expresses and displays ADA on
   bacterial surface under hypoxic tumor condition. The surface‐exposed
   ADA then converts extracellular adenosine into inosine in situ to
   alleviate the immunosuppressive effects, thereby leading to
   intratumoral immune cells activation. In a subcutaneous colorectal
   cancer (CRC) mouse model, we demonstrated that EcN_ADA strain
   substantially improved immune cell infiltration and had marked
   synergistic effects with low dose of chemotherapeutic in the clearance
   of solid tumors. Furthermore, the EcN_ADA strain also ameliorated
   intestinal inflammation and tumor growth in an orthotopic CRC mouse
   model. The results demonstrate the potential to utilize synthetic
   biology technique for TME metabolite modulation. We anticipate that the
   bioengineered probiotics EcN_ADA could provide a unique strategy for
   synergizing chemotherapeutics for combinational cancer treatment.

2. Results and Discussion

2.1. Adenosine Deaminase Reverses Adenosine‐Mediated Immunosuppression of
Immune Cells In Vitro

   ADA is one of the key enzymes involved in purine metabolism. There are
   mainly two isoforms in humans: a predominantly intracellular monomeric
   ADA1 (35 kDa) and a serum dimeric ADA2 (100 kDa).^[ [79]^34 ^] Since
   the protein with lower molecular weight is more compatible for
   bacterial surface display, the monomeric ADA1 was chosen for further
   studies. In vitro catalytic results showed that recombinant human ADA1
   had a high activity toward adenosine with a K [cat] value of 143 and
   138 s^−1 at pH 7.4 and 6.8, respectively (Figure [80]S1, Supporting
   Information), which are consistent with previous studies.^[ [81]^35 ^]
   Extensive investigations have demonstrated that high concentrations of
   extracellular adenosine inhibit the activation and effector functions
   of various immune cell populations.^[ [82]^36 , [83]^37 , [84]^38 ^]
   Therefore, we subsequently sought to investigate whether ADA could
   reverse adenosine‐mediated immunosuppression of CD8^+ T cells and
   macrophages in vitro.

   To investigate the impact of ADA on CD8^+ T cells activation, isolated
   mouse spleen CD8^+ T cells were stimulated with anti‐CD3 and anti‐CD28
   antibodies in the presence of PBS, 100 nM ADA, 1 mM adenosine, 1 mM
   inosine, and 1 mM adenosine + 100 nM ADA (Figure [85] 1a). After 24 h
   incubation, flow cytometry analysis was applied to access IFN‐γ and
   CD25 activation marker in CD8^+ T cells. As shown in Figure [86]1b,c,
   and Figure [87]S2 (Supporting Information), 100 nM ADA alone exhibited
   negligible effects on CD8^+ T cells activation. Whereas, CD8^+ T cells
   activation was significantly suppressed in the presence of 1 mM
   adenosine, as evidenced by substantially lower CD25 and IFN‐γ
   expression compared with the control group. Conversely, 1 mM inosine
   enhanced the activation of CD8^+ T cells, leading to elevated IFN‐γ
   levels. Administration of 1 mM adenosine + 100 nM ADA also resulted in
   augmented IFN‐γ production comparable to the inosine‐treated group,
   suggesting that 100 nM ADA efficiently catalyzed the conversion of 1 mM
   adenosine to inosine. Moreover, previous study have indicated that
   inosine could serve as an alternative carbon source, supporting the
   survival and function of effector CD8^+ T cells under glucose‐depleted
   conditions.^[ [88]^39 ^] Therefore, we subsequently examined whether
   the enzymatically catalyzed product by ADA could similarly sustain the
   function of effector CD8^+ T cells. After activation with anti‐CD3 and
   anti‐CD28 antibodies for 24 h, the effector CD8^+ T cells were
   transferred into glucose‐free medium and cultured for an additional
   four days under various treatments, including PBS buffer, 100 nM ADA,
   1 mM adenosine, 1 mM inosine, and 1 mM adenosine + 100 nM ADA
   (Figure [89]1d). Flow cytometry data showed that 100 nM ADA or 1 mM
   adenosine treatment had similar expression levels of IFN‐γ, TNF‐α and
   Granzyme B in CD8^+ T cells compared with the control group. Although
   not statistically significant, a clear trend toward higher expression
   of IFN‐γ, TNF‐α and Granzyme B was observed in the presence of 1 mM
   inosine and 1 mM adenosine + 100 nM ADA (Figure [90]1e–g;
   Figure [91]S3, Supporting Information). These results indicated that
   the enzymatically generated inosine by ADA had the capacity to sustain
   the function of effector CD8^+ T cells in the glucose‐depleted
   conditions.

Figure 1.

   Figure 1
   [92]Open in a new tab

   Adenosine deaminase alleviates adenosine‐mediated suppression of immune
   cells in vitro. a) Schematic illustration of the impacts of purine
   nucleotides on the activation of CD8^+ T cells. b,c) Flow cytometry
   analysis of intracellular INF‐γ and cell surface CD25^+ marker
   expression levels in CD8^+ T cells after the indicated treatments. d)
   Schematic illustration of purine nucleotides as alternative carbon
   source to sustain the function of effector CD8^+ T cells in
   glucose‐depleted medium. e–g) Flow cytometry analysis of the persistent
   expression of INF‐γ, TNF‐α, and Granzyme B in effector CD8^+ T cells
   after the indicated treatments in glucose‐depleted condition. Error
   bars indicate standard deviation of three independent experiments (n =
   3). h) Schematic diagram illustration of the impacts of purine
   nucleotides on phenotype switch of M1 and M2 macrophages. Flow
   cytometry analysis of the expression levels of F4/80, CD86 and CD206
   markers on M1‐like i,j) and M2‐like macrophages k,l) after the
   indicated treatments. Error bars indicate standard deviation of five
   independent experiments (n = 5). Differences between groups were
   assessed using one‐way ANOVAs, followed by Bonferroni's correction for
   multiple comparisons. Significance levels were indicated as follows: *p
   < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

   Adenosine receptor activation promotes the macrophages switching from a
   pro‐inflammatory M1 to an anti‐inflammatory M2 phenotype. Therefore, we
   further examined the impact of ADA on the adenosine‐mediated M1‐M2
   macrophage phenotype switch. The mouse bone‐marrow‐derived macrophages
   (BMDMs) were isolated and polarized into M1 and M2 macrophages. The
   obtained M1‐like macrophages were subsequently incubated with PBS,
   100 nm ADA, 1 mM adenosine, 1 mM inosine, or 1 mM adenosine + 100 nm
   ADA for 24 h (Figure [93]1h). Exposure to 1 mM adenosine induced a
   transition in macrophage phenotype from M1 to M2, characterized by a
   significant reduction in CD86 levels and elevation of CD206 biomarker
   expression. In contrast, 1 mM adenosine + 100 nm ADA treatment has no
   effects on M1 phenotype (Figure [94]1i, j; Figure [95]S4, Supporting
   Information). For M2‐like macrophages, 1 mM inosine and 1 mM adenosine
   + 100 nm ADA treatment resulted in an obvious increase of CD80 levels
   and reduction of CD206, indicative of a distinct shift toward
   pro‐inflammatory M1 phenotype (Figure [96]1k,l; Figure [97]S5,
   Supporting Information). Collectedly, all these results demonstrated
   that ADA could effectively catalyzed adenosine into inosine, reversing
   the adenosine‐mediated immunosuppression on CD8^+ T cells and
   macrophages in vitro.

2.2. Peritumoral Injection of Adenosine Deaminase Promotes Immune Cell
Infiltration into Solid Tumor

   Previous studies demonstrated that D296 is a key residue for the
   catalytic function of ADA.^[ [98]^40 ^] Therefore, an ADA^D296A variant
   was constructed as a negative control. Deficiency in the catalytic
   activity of ADA^D296A toward adenosne was confirmed by HPLC analysis.
   (Figure [99]S6, Supporting Information). Initially, in vivo experiment
   using C57BL/6 subcutaneous tumor models (MC38) were conducted to
   examine the catalytic activity of ADA in the TME. After intratumoral
   injection of 50 µg ADA or ADA^D296A, the inosine concentration in mice
   tumor significantly increased in the ADA‐treated group within 24 h,
   while no significant change was observed in the ADA^D296A‐treated group
   (Figure [100] 2a–d). The in vivo experiment confirmed the catalytic
   function of ADA in the TME.

Figure 2.

   Figure 2
   [101]Open in a new tab

   Adenosine deaminase in combination with doxorubicin exhibits
   synergistic anti‐tumor effect in mouse subcutaneous tumor model. A)
   Schematic illustration of ADA activity evaluation in a mouse MC38
   subcutaneous tumor model. Intratumoral concentrations of adenosine b)
   and inosine c), along with the inosine‐to‐adenosine ratio d), were
   measured in MC38 tumor‐bearing mice after different treatments and
   plotted (n = 3). e) Schematic illustration of the combinational therapy
   of adenosine deaminase and doxorubicin in a mouse MC38 subcutaneous
   tumor model. Body weight f) and tumor volume changes g) of MC38
   tumor‐bearing mice after different treatments were plotted (n = 6).
   Representative tumor figures h) and tumor weights i) of MC38 tumors in
   mice from different treatment groups (n = 3). Error bars indicate
   standard deviation of three mice in each experimental group.
   Differences between groups were assessed using one‐way ANOVAs, followed
   by Bonferroni's correction for multiple comparisons. Significance
   levels were indicated as follows: *p < 0.05, **p < 0.01, ***p < 0.001,
   ****p < 0.0001. j) Immunohistochemical staining of tumor sections
   reveals infiltration of M1 (green) and M2 (green) macrophages, CD8^+ T
   cells, CD4^+ T cells, and NK cells. The black triangles indicate the
   infiltrated immune cells. Scale bar: 50 µm.

   Doxorubicin (DOX), a clinically used chemotherapy drug for CRC and an
   inducer of immunogenic cell death, was combined with ADA to enhance the
   efficacy of chemo‐immunotherapy. The potential synergistic effect of
   ADA and DOX was initially validated at the cellular level. Equal
   numbers of polarized M1 macrophages and MC38 tumor cells were
   co‐cultured with 100 µM adenosine under three conditions: DOX alone,
   ADA alone, and ADA + DOX, for 4 h. Fluorescence microscopy revealed
   that, in the presence of 100 nM ADA and 50 µM DOX, tumor cell
   phagocytosis by M1 macrophages increased significantly to ≈80%,
   surpassing the levels observed with either DOX or ADA alone
   (Figure [102]S7, Supporting Information). These results provide strong
   evidence for the synergistic effect of ADA and DOX.

   We next sought to evaluate the clinical efficacy of the recombinant ADA
   when paired with the chemotherapy agent DOX in cancer therapy using a
   subcutaneous tumor mouse model. To determine the optimal dosages of ADA
   and DOX, C57BL/6 mice bearing MC38 tumors in the flank region were
   treated with PBS, DOX, ADA, and ADA + DOX every three days. At 70 µg
   DOX, tumor suppression was effective in both DOX and DOX + ADA groups,
   with no significant difference observed, indicating that DOX alone
   overshadowed the combination's synergy. To reduce DOX's cardiotoxicity,
   its dose was lowered to 20 µg in subsequent experiments to enhance
   synergy while minimizing toxicity (Figure [103]S8, Supporting
   Information). The 100 µg ADA group showed no antitumor effect, likely
   due to poor tumor targeting and loss during peritumoral injection. To
   address this, the ADA dose was increased to 300 µg in subsequent
   studies.

   In the formal experiments, C57BL/6 mice bearing MC38 tumors in the
   flank region was treated with PBS, DOX, ADA^D296A, ADA, ADA^D296A + DOX
   and ADA + DOX every three days. Notably, ADA^D296A and ADA (300 µg)
   were administered via peritumoral injection, while low‐dose DOX (20 µg)
   was delivered intravenously via mice tail vein. On the 36th day post
   tumor implantation, the mice were sacrificed for tumor collection
   (Figure [104]2e). No significant changes in mice body weight were
   observed during the treatments, indicative that the treatments had no
   substantial toxicity in the mice (Figure [105]2f). Administration of
   either DOX or ADA^D296A alone did not exhibit significant effects on
   tumor growth compared to the PBS‐treated mice. By contrast, ADA and
   ADA^D296A + DOX administration substantially attenuated tumor growth.
   In particular, the combination of ADA and DOX even resulted in complete
   suppression of tumor growth, suggesting synergistic anti‐tumor effects
   between ADA and DOX (Figure [106]2g–i). The collected tumor tissues
   were further subjected to immunofluorescence staining. A notable
   increase in M1 macrophage abundance were observed across all the
   treated group. In particular, the ADA + DOX group exhibited the highest
   M1 macrophage density alongside the lowest M2 macrophage count,
   indicative of the M2‐to‐M1 switch within the TME and enhancement of
   anti‐tumor immune response. Furthermore, notable augmentation of CD4^+,
   CD8^+ T cells and natural killer (NK) cells infiltration within the TME
   was evident in both the ADA and ADA + DOX groups, with the ADA + DOX
   combination therapy demonstrating the most pronounced enhancement of
   immune cell infiltration (Figure [107]2j; Figure [108]S9a–d, Supporting
   Information). These data collectively suggested that peritumoral ADA
   administration facilitates immune cell infiltration into the TME, while
   ADA + DOX combination therapy provides highly potent synergistic
   anti‐tumor effects.

2.3. Construction and Characterization of Engineered Probiotic E. coli for
Adenosine‐conversion

   To achieve the desired anti‐tumor effect, relatively high doses of ADA
   must be targeted administrated to the solid tumor site. However,
   peritumoral injection of ADA is impracticable for clinic use. This
   highlights the need to develop alternative approaches to locally and
   continuously deliver ADA to tumor. Previous studies showed that a
   probiotic E. coli Nissle 1917 (EcN) could selectively colonize and
   thrive in hypoxic tumor site.^[ [109]^20 ^] Consequently, we sought to
   engineer the probiotic as a synthetic biology‐based cellular therapy
   for intratumoral ADA production. We envisioned that ADA display on the
   bacterial surface would optimize the catalysis of extracellular
   adenosine. Therefore, we initially evaluate the feasibility of
   displaying ADA on EcN surface utilizing a bacterial surface display
   system that we had previously reported.^[ [110]^31 ^] In brief, the
   human ADA1 (40 kDa) was fused to the N‐terminus of a bacterial AIDA‐I
   autotransporter domain.^[ [111]^25 ^] The gene encoding the fusion
   protein was under control of a lacUV5 promoter (P[lacUV5]) for
   constitutive expression (Figure [112] 3a). The constructed plasmid was
   subsequently transformed into EcN to prepare the EcN_lac_ADA strain.
   Immunoblotting and flow cytometry analysis revealed that the ADA‐AIDA‐I
   fusion protein was successfully expressed on the EcN surface and
   oriented toward the cell exterior (Figure [113]3b,c), which was further
   verified by the fluorescence bacterial staining analysis
   (Figure [114]3d). Bacterial growth curves revealed that EcN_lac_ADA
   exhibits similar a growth rate compared to that of EcN, indicating that
   the genetic modification did not adversely affect the bacterial
   proliferation (Figure [115]S10, Supporting Information). We next sought
   to assess the enzymatic activity of bacterial surface‐displayed ADA. In
   brief, equal amount of EcN or EcN_lac_ADA (OD[600] = 0.25) were
   incubated with 1 mM adenosine in 10 mM HEPES buffer at 37 °C. The
   adenosine and inosine contents in the buffer were measured by HPLC
   every hour. Notably, adenosine was completely converted to inosine by
   EcN_lac_ADA within 5 h, whereas only 20% of adenosine was catalyzed by
   wild‐type EcN (Figure [116]3e,f; Figure [117]S11, Supporting
   Information). These data unequivocally underscore the remarkable
   adenosine conversion proficiency of the surface‐displayed ADA.

Figure 3.

   Figure 3
   [118]Open in a new tab

   Construction and characterization of engineered probiotic. a) Schematic
   illustration of E. coli Nissle 1917 (EcN) surface‐display system of
   adenosine deaminase (ADA) with a lacUV5 promoter (P[lacUV5]). b)
   Western blot analysis to verify the expression of ADA‐AIDA fusion
   protein. Flow cytometry c) and immunofluorescence imaging d) analysis
   of the surface‐displayed ADA on engineered E. coli Nissle 1917
   (EcN_lac_ADA). e) HPLC profiles of adenosine after 5 h incubation with
   EcN_lac_ADA. The elution times for adenosine and inosine are indicated.
   f) Quantification of adenosine conversion by wild‐type EcN and
   EcN_lac_ADA within 5 h. g) Schematic illustration of EcN
   surface‐display system of ADA with a hypoxia‐controlled fnrS promoter
   (P[fnrS]). h) Western blot analysis to verify the expression of
   ADA‐AIDA fusion protein under anaerobic condition. Flow cytometry i)
   and immunofluorescence imaging j) analysis of hypoxia‐induced ADA
   surface display on engineered E. coli Nissle 1917 (EcN_ADA). k) HPLC
   profiles of adenosine after 5 h incubation with EcN_ADA under aerobic
   and anaerobic conditions. The elution times for adenosine and inosine
   are indicated. l) Quantification of adenosine conversion by EcN_ADA
   under anaerobic or aerobic condition within 7 h.

   When utilized for tumor treatment, constitutive expression of ADA by
   EcN_lac_ADA could potentially have adverse effect on normal tissues or
   provoke immune responses against the bacteria. In addition, sustained
   expression would also impose a metabolic burden on the bacteria,
   potentially compromising their viability and persistence during the
   treatment.^[ [119]^41 ^] The constitutive lacUV5 promoter was
   subsequently replaced with a hypoxia‐inducible fnrS promoter
   (P[fnrS])^[ [120]^42 ^] to construct the EcN_ADA strain. This
   modification should enable the engineered EcN_ADA strain to selectively
   express ADA within the hypoxic regions of tumors, thereby enhancing
   treatment specificity and reducing adverse effects (Figure [121]3g).
   Tumor tissues exhibit lower oxygen levels (0.1%–2%, −200 to −300 mV)
   than normal tissues (5%–10%, −150 to −250 mV) due to rapid cell
   proliferation and poor blood supply.^[ [122]^43 ^] Therefore, we used
   1% oxygen to mimic hypoxic conditions in tumors. Immunoblotting results
   showed that the ADA‐AIDA‐I fusion protein was only expressed in hypoxia
   conditions rather than normoxia (Figure [123]3h). Similarly, flow
   cytometry analysis revealed that the bacterial surface‐displayed ADA in
   the EcN_ADA strain was only detectable under hypoxic conditions,
   although the level was relatively lower compared to that of the
   EcN_lac_ADA strain (Figure [124]3i). Moreover, bacterial
   immunofluorescence staining further confirmed the successful surface
   display of ADA under hypoxic conditions rather than normoxic conditions
   (Figure [125]3j). In consistent, the EcN_ADA strain could completely
   catalyzed 1 mM adenosine into inosine within 7 h under hypoxic
   condition, whereas only ≈25% adenosine could be converted under
   normoxic condition (Figure [126]3k,l). Furthermore, the impact of the
   promoter switch on enzymatic activity was also evaluated by comparing
   the catalytic efficiency of EcN_ADA and EcN_lac_ADA in converting 1 mM
   adenosine under 1% oxygen condition. HPLC analysis showed that
   EcN_lac_ADA achieved almost 100% conversion at 300 min, compared to
   ≈70% for EcN_ADA, indicative of slightly reduced catalytic activity
   (Figure [127]S12, Supporting Information). Overall, these findings
   demonstrate that the engineered EcN_ADA strain expresses and presents
   ADA on its surface in response to hypoxia, resulting in effective
   catalysis of adenosine into inosine in vitro.

2.4. Engineered Probiotic E. coli Synergizes with Chemotherapy to Facilitate
the Regression of Mouse Subcutaneous Tumor

   To assess the compatibility of EcN_ADA with DOX in therapy, we first
   investigated the impact of DOX on the engineered probiotics.
   Intriguingly, experimental results showed that 400 µM DOX had no
   significant effect on bacterial ADA expression and catalytic activity
   (Figure [128]S13, Supporting Information). Additionally, the catalytic
   activity of bacterial surface‐displayed ADA and purified ADA enzyme was
   compared in the presence of DOX. Enzyme‐linked immunosorbent assay
   (ELISA) results showed that ≈0.2 nmol of ADA enzyme was diplayed on
   10^8 CFU of EcN_ADA. Based on this, the adenosine conversion rates of
   bacterial surface‐displayed ADA and purified ADA were calculated as 167
   and 70 min^−1, respectively (Figure [129]S14, Supporting Information).

   We subsequently investigated whether the EcN_ADA could synergize with
   low‐dose DOX chemotherapy for cancer treatment using a subcutaneous
   tumor mouse model. In addition, a mutant EcN_ADA^D296A strain without
   catalytic activity was also constructed as a negative control, and
   deficiency in the catalytic activity of EcN_ADA^D296A strain toward
   adenosne was confirmed (Figure [130]S15, Supporting Information). To
   determine the optimal dosage of engineered probiotics for treatment, a
   dose of 10⁶ CFU of bacteria was administered intravenously to mice, but
   this resulted in significant mortality within 10 days. While reducing
   the bacterial dose to 10⁵ CFU eliminated mortality throughout the
   treatment period (Figure [131]S16, Supporting Information). Notably, a
   single intravenous administration of 10⁵ CFU engineered bacteria
   persisted in mouse tumors for over 6 days (Figure [132]S17, Supporting
   Information). Therefore, a bacterial dose of 10⁵ CFU was established as
   the optimal dosage for both therapeutic efficacy and safety in
   subsequent animal studies. In brief, C57BL/6 mice bearing MC38 tumors
   the in the flank region were treated with mono‐ or combinational
   therapy, including PBS, DOX, EcN, EcN_ADA^D296A, EcN_ADA, EcN + DOX,
   EcN_ADA^D296A + DOX and EcN_ADA + DOX. Typically, 20 µg of DOX in a
   total volume of 100 µL was administered via intravenous injection for
   each mouse every three days. Tail vein injections of bacteria were
   conducted at a concentration of 1 × 10^6 CFU mL^−1 in PBS, with a total
   volume of 100 µL injected per mouse every five days. On the 30th day
   post tumor implantation, the mice were sacrificed for tumor collection
   (Figure [133] 4a). Throughout the treatment period, no significant
   decreases in mice body weight were observed in any experimental groups,
   implying that both mono‐ and combination therapy exhibited no
   significant toxicity on the mice (Figure [134]4b). As shown in
   Figure [135]4c, administration of DOX, EcN and EcN_ADA^D296A alone
   exhibited no significant effects on tumor growth compared to the
   control group. In contrast, administration of EcN_ADA resulted in a
   substantial slowdown in tumor growth, with the final tumor weight being
   only ≈50% of that in the control group. Intriguingly, the combination
   therapy of EcN + DOX or EcN_ADA^D296A + DOX also slowed down the tumor
   growth rate, despite the lack of significant impact on tumor growth
   observed with EcN, EcN_ADA^D296A or DOX alone, implying a synergistic
   effect between the probiotic‐based therapy and chemotherapy.
   Remarkably, the administration of Ec N_ADA + DOX led to a pronounced
   suppression in tumor growth rate and resulted in the lightest tumor
   weights (Figure [136]4d,e). After mice sacrifice, various tissues were
   collected, homogenized and analyzed for retained bacteria. The results
   showed that the engineered bacteria primarily accumulated at the tumor
   site, with minimal presence in the heart, liver, spleen, lungs, and
   kidneys, indicative of the targeted colonization ability of the
   engineered bacteria in tumors. Notably, both the engineered
   EcN_ADA^D296A and EcN_ADA colonized and persisted in tumors to a
   similar extent as EcN, indicating that genetic manipulation did not
   alter the tumor colonization features of EcN (Figure [137]S18,
   Supporting Information).

Figure 4.

   Figure 4
   [138]Open in a new tab

   Combining engineered probiotics and doxorubicin inhibit solid tumor
   growth in mouse subcutaneous tumor model. a) Schematic illustration of
   the combinational therapy of engineered probiotics and doxorubicin in a
   mouse MC38 subcutaneous tumor model. Body weight b) and tumor volume
   changes c) of MC38 tumor‐bearing mice after different treatments were
   plotted (n = 6). Representative of tumor figures d) and tumor weights
   e) of MC38 tumors in mice from different treatment groups (n = 6).
   Error bars indicate standard deviation of three mice in each
   experimental group (n = 6). f) Schematic illustration of the
   experimental procedure to quantify adenosine and inosine in murine
   tumor tissues using the solid‐phase microextraction (SPME). g)
   Quantified inosine/adenosine ratios in murine tumor tissues in
   different experimental groups. h) Immunohistochemical staining of tumor
   sections reveals infiltration of M1 (green) and M2 (green) macrophages,
   CD8+ T cells, CD4+ T cells, and NK cells. The black triangles indicate
   the infiltrated immune cells. Scale bar: 50 µm. i–k) Secreted cytokine
   levels of INF‐γ i), TNF‐α j), IL12‐p70 k) in murine serum in different
   experimental groups. Error bars indicate standard deviation of three
   mice in each experimental group (n = 6).Differences between groups were
   assessed using one‐way ANOVAs, followed by Bonferroni's correction for
   multiple comparisons. Significance levels were indicated as follows: *p
   < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

   In vivo hypoxic‐induced expression of ADA by the engineered probiotics
   was confirmed by myc‐FITC antibody staining of the tumor sections.
   Green fluorescence from the myc‐tag linked to ADA or ADA^D296 was
   observed in tumor treated with EcN_ADA, EcN_ADA + DOX, EcN_ADA^D296A or
   EcN_ADA^D296A + DOX, but not in tumor treated with EcN, confirming in
   situ expression of ADA or ADA^D296A by the engineered probiotics
   (Figure [139]S19, Supporting Information). Moreover,
   immunohistochemistry (IHC) analysis of the mice heart, liver, spleen,
   lungs, and kidneys showed no significant damage in the EcN + DOX group
   compared to others, further confirming the safety of the combinational
   therapy (Figure [140]S20, Supporting Information).

   To investigate the metabolic changes in the TME after various
   treatments, we employed hydrophilic‐lipophilic balanced solid‐phase
   microextraction (HLB‐SPME) method to measure the levels of adenosine
   and inosine in solid tumors (Figure [141]4f). The calculated
   concentration of adenosine in TME was 0.04 µM in the control group,
   which is in consistent with the values reported previously.^[ [142]^44
   ^] Unexpectedly, no significant change on the adenosine levels were
   observed among all the treated groups (Figure [143]S21, Supporting
   Information). In contrast, substantially increased levels of inosine
   were detected in EcN_ADA and EcN_ADA + DOX‐treated group, indicative of
   successful intratumoral enzymatic conversion of adenosine to inosine
   (Figure [144]4g; Figure [145]S22, Supporting Information). One possible
   explanation for this discrepancy could be the rapid turnover of
   adenosine within the TME, wherein newly generated adenosine replenishes
   the converted adenosine, resulting in a relatively stable
   concentration.

   Immunofluorescence analysis of tumor sections showed that most
   tumor‐associated macrophages in the control group exhibited an M2‐like
   phenotype, consistent with an immunosuppressive condition.
   Administration of DOX, EcN, EcN_ADA^D296A, EcN + DOX or EcN_ADA^D296A +
   DOX led to a slight increase the number of M1‐like macrophage within
   tumors. In contrast, treatment with EcN_ADA and EcN_ADA + DOX
   substantially elevated the M1/M2 macrophage ratio in the TME,
   suggesting a shift toward a more immunostimulatory TME (Figure [146]4h;
   Figure [147]S23, Supporting Information). Moreover, immunohistochemical
   analysis demonstrated a significant increase in infiltration of CD4^+
   cells, CD8^+ cells, and NK cells in tumors from mice treated with
   EcN_ADA and EcN_ADA + DOX compared to other experimental groups.
   Particularly, the co‐administration of EcN_ADA and DOX resulted in the
   highest infiltration of these immune cell populations, highlighting the
   synergistic immune‐stimulatory effects of EcN_ADA and DOX combination
   therapy (Figure [148]S24a–d, Supporting Information). In consistent,
   ELISA analysis of the mouse serum demonstrated that the EcN_ADA + DOX
   treatment resulted in the highest levels of INF‐γ, TNF‐α, and IL‐12p70
   cytokines compared to other treated groups (Figure [149]4i–k). These
   data further confirmed the activation of immune cells such as T cells
   and macrophages, which are known to produce these cytokines upon
   activation.

2.5. Single‐Cell RNA (scRNA) Sequencing Profiling of Mouse Subucaneous Tumor

   To further investigate how EcN_ADA enhances anti‐tumor immune responses
   in mouse subcutaneous tumor, the collected tumors were subjected to
   single‐cell RNA sequencing (scRNA) analysis. After quality control and
   batch correction, all the identified cells were classified into 5
   different cell populations with a reference‐based type annotation
   method followed by manual curation of top marker genes. These
   populations include tumor cells, fibroblast, macrophage, myeloid cells,
   and T cells (Figure [150]S25, Supporting Information). Globally, the
   combinational treatment of EcN_ADA and EcN_ADA + DOX induced
   substantial changes in the population of both myeloid cells and
   lymphocytes, leading to a notably higher abundance of T cells and
   myeloid cells (Figure [151]S26, Supporting Information). Sub‐setting
   and re‐clustering of all immune cells further resulted in 12 clusters,
   which were classified into six immune cell subtypes after annotation,
   including T cells, NK cells, DC cells, monocytes, macrophages, and
   neutrophils (Figure [152] 5a; Figure [153]S27, Supporting Information).
   Similarly, the results also revealed significant increase in the
   population of macrophages, DC cells and T cells after EcN_ADA or
   EcN_ADA + DOX treatment (Figure [154]5b). KEGG enrichment analysis on
   the differentially expressed genes (DEGs) of immune cells between
   EcN_ADA and EcN_ADA^D296A‐treated groups revealed increased activation
   of cytokine‐cytokine interaction and inflammation regulatory pathways,
   such as NF‐κB, NOD‐like receptor and Toll‐like receptor signaling
   pathways in EcN_ADA‐treated tumors, indicative of induced inflammation
   responses in EcN_ADA‐treated tumors (Figure [155]5c). In particular,
   the upregulation of chemokine and cytokine genes such as Cxcl1, Cxcl2,
   Cxcl3, Ccl3, Ccl4, Ccl5, Il1a, Il1b and Il1rn indicates a significant
   enhancement in immune cell recruitment within EcN_ADA‐treated tumors
   (Figure [156]5d,e). The unexpected downregulation of Cxcl9 may be due
   to the reduced expression of Spi1, a key transcription factor that
   regulates Cxcl9 expression.^[ [157]^45 ^] Overall, the results are in
   consistent with the enhancement in immune cells infiltration in tumor
   section analysis.

Figure 5.

   Figure 5
   [158]Open in a new tab

   Single‐cell RNA (scRNA) sequencing analysis of the mice tumor tissues
   after treatment. a) t‐distributed stochastic neighbor embedding (t‐SNE)
   representation of all the identified cell populations in mice tumors.
   Labels have been added on the basis of the expression of marker genes.
   NK, natural killer cell. DC, dendritic cell. b) t‐SNE plots are shown
   separated by different treatment conditions. Macrophage are highlighted
   in purple, DC cells are highlighted in red and T cells are highlighted
   in orange. c) KEGG pathway enrichment analysis of differentially
   expressed genes (DEGs) of all the immune cells between EcN_ADA and
   EcN_ADA^D296A‐treated group. Heatmap d) and volcano plot e) showing the
   differentially expressed genes in cytokine‐cytokine receptor
   interaction pathway of immune cells between EcN_ADA and
   EcN_ADA^D296A‐treated groups. f) t‐SNE plot of all identified T cell
   populations, with labels based on marker gene expression. The exhausted
   T cell cluster is highlighted in red, and the effector T cell cluster
   is highlighted in yellow. g) The proportions of the four populations of
   T cells in mice tumor tissues after different treatments. h) Heatmap
   depicting the differentially expressed genes in citric acid cycle
   pathway of T cells between EcN_ADA and EcN_ADA^D296A‐treated group. i)
   t‐SNE plot of all identified macrophage populations, with labels based
   on marker gene expression. The M1 macrophage cluster is highlighted in
   blue, and the M2 macrophage cluster is highlighted in purple. j) The
   proportions of the four populations of macrophage in mice tumor tissues
   after different treatments. k) Heatmap depicting the differentially
   expressed genes in NF‐κB pathway of macrophage between EcN_ADA and
   EcN_ADA^D296A‐treated group. l) Schematic illustration of downstream
   genes involved in tumor proliferation and migration in the adenosine
   receptor (A2aR) signaling pathway. m) Heatmap depicting the
   differentially expressed genes in A2aR signaling pathway of tumor cells
   between EcN_ADA and EcN_ADA^D296A‐treated group. n) The average
   transcriptional levels of Cd73 and Cd39 genes in tumor cells from
   EcN_ADA and EcN_ADA^D296A‐treated group. Statistical analyses were
   performed using the Gehan‐Wilcoxon test followed by Bonferroni's
   correction for multiple comparisons for all survival analyses. ****p
   < 0.0001.

   We subsequently sought to investigate the effects of EcN_ADA treatment
   on T cells and macrophages individually. The sub‐setting of total T
   cells yielded four clusters, with a typical exhausted T cell cluster
   (T[EX], cluster 0) and an effector T cell cluster (T[EFF], cluster 1)
   (Figure [159]5f; Figure [160]S28, Supporting Information). Notably,
   differential gene expression analysis showed a pronounced exhausted T
   cells reduction and a substantially increase in the antitumor effector
   T cells within tumors from EcN_ADA + DOX‐treated mice relative to other
   experimental groups (Figure [161]5g). KEGG enrichment analysis of total
   T cells between EcN_ADA and EcN_ADA^D296A‐treated tumors also revealed
   increases in pathways related to anti‐tumor immune responses, such as
   cytokine‐cytokine receptor interaction and natural killer cell mediated
   cytotoxicity (Figure [162]S29a–c,Supporting Information). In
   particular, the tricarboxylic acid (TCA) cycle‐related genes exhibited
   significant upregulation in the EcN_ADA‐treated compared to the
   EcN_ADA^D296A‐treated group (Figure [163]5h), indicative of T cell
   metabolic reprogramming and activation.^[ [164]^46 , [165]^47 ^]
   Indeed, flow cytometry analysis showed a significant increase in CD3+ T
   cell infiltration in tumors treated with EcN_ADA + DOX
   (Figure [166]S30, Supporting Information), indicating that the therapy
   promotes T cell infiltration into the TME.

   The probiotic‐based therapy also affected intratumoral macrophage
   populations (Figure [167]5i; Figure [168]S31, Supporting Information).
   Differential gene expression analysis unveiled that tumor from mice
   treated with EcN_ADA or EcN_ADA + DOX contained significantly fewer
   immunosuppressive M2‐like macrophages characterized by high expression
   levels of hallmark genes such as Arg1 or Mrc1 (clusters 2 and 3).
   Instead, there was a notable shift toward M1‐like macrophages
   predominantly expressing antitumoral effector molecules, such as Il1b,
   Tnf and Cxcl2 (cluster 0 and 1; Figure [169]5j). In consistent, KEGG
   enrichment analysis further corroborated the M2 to M1 macrophage
   phenotypic switch in tumors treated with EcN_ADA, evidenced by the
   elevation in pathways associated with M1 macrophage polarization,
   including the typical NF‐κB signaling pathway^[ [170]^48 ^]
   (Figure [171]5k; Figure [172]S32a–c, Supporting Information). Flow
   cytometry analysis showed a significant increase in the M1/M2
   macrophage ratio in tumors treated with EcN_ADA + DOX (Figure [173]S33,
   Supporting Information), supporting the notion that the combination
   therapy promotes macrophage reprogramming from the M2 to M1 phenotype
   within the TME.

   In addition to immune cell inhibition, adenosine could also bind to the
   A2aR receptor in tumor cells and activate the downstream PI3K‐AKT
   pathway, thereby promoting tumor cell proliferation and migration^[
   [174]^49 , [175]^50 ^] (Figure [176]5l). Given the efficient adenosine
   conversion by EcN_ADA treatment, we also examined the impact of EcN_ADA
   treatment on the A2aR signaling pathway in tumor cells. After
   sub‐setting, the transcriptional levels of genes associated with the
   A2aR signaling pathway in all tumor cells were assessed. Overall,
   EcN_ADA treatment significantly downregulated genes related to the A2aR
   signaling pathway in tumor cells, including Gnaq, Plcb2, Pap1b in
   comparison with the EcN_ADA^D296A‐treated group (Figure [177]5m).
   Importantly, three key genes essential for tumor proliferation and
   metastasis, namely Ccnd1, Vegfa, and Hif1a, were also downregulated
   after EcN_ADA treatment (Figure [178]S34, Supporting Information).
   Collectively, all these results suggested that the EcN_ADA treatment
   has the potential to inhibit tumor cell proliferation and metastasis by
   reducing the activity of the A2aR signaling pathway. Intriguingly, we
   observed a significant upregulation in the transcriptional levels of
   the two ecto‐nucleotidases Cd73 and Cd39, responsible for generating
   extracellular adenosine in the EcN_ADA‐treated group compared to the
   EcN_ADA^D296A‐treated group (Figure [179]5n). This upregulation may be
   attributed to a compensatory feedback effect following the removal of
   extracellular adenosine.

2.6. Engineered Probiotic Together with Low‐Dose Chemotherapeutics Inhibits
Tumor Growth in a Mouse Orthotopic CRC Model

   The oral administration of probiotics presents a preferable and
   effective alternative to intravenous delivery, providing sustained
   benefits in mitigating systemic toxicities. Oral administration of 10^9
   CFU engineered EcN_ADA persisted in mouse intestine for over 6 days,
   suggesting that the engineered probiotics hold the potential to provide
   long‐term therapeutic effects within the gut (Figure [180]S35,
   Supporting Information). Inspired by the anti‐tumor efficacy of the
   engineered probiotic, we proceeded to evaluate its therapeutic
   potential using an orthotopic CRC mouse model. In brief, 3‐week‐old
   C57BL/6c mice underwent a 10‐week continuous induction period involving
   the initial administration of the carcinogenic agent azoxymethane (AOM)
   and the colitis‐inducing compound dextran sodium sulfate salt (DSS). To
   mitigate the risk of severe colitis and mortality associated with
   prolonged DSS exposure, mice were alternatively given a diet consisting
   of 2.5% DSS and water on a weekly basis.^[ [181]^51 ^] After confirming
   successful model establishment via colonoscopy in the fourth week, the
   mice were randomly assigned to eight groups. During the following six
   weeks, each group received various treatments, including PBS, DOX, EcN,
   EcN_ADA^D296A, EcN_ADA, EcN + DOX, EcN_ADA^D296A + DOX, and EcN_ADA +
   DOX. Among these treatments, 20 µg of DOX in a total volume of 100 µL
   was administered via intravenous injection through the tail vein every
   three days. Meanwhile, 100 µL bacterial suspension in PBS with a
   concentration of 1 × 10^10 CFU mL^−1 was administrated via oral gavage
   every five days. At week 11, another colonoscopy was conducted to
   evaluate the progression of intestinal inflammation and colorectal
   tumor of mice in each experimental group and followed by euthanasia
   (Figure [182] 6a). Given the close correlation between the severity of
   colonic lesions and colon length, the colons of the mice were
   subsequently collected and measured. Notably, mice in the untreated
   group exhibited significant colonic lesions, characterized by a
   considerable reduction in colon length to ≈22.8 ± 1.3 cm. In contrast,
   mice receiving the EcN_ADA + DOX treatment had the longest colon
   lengths among all the experimental groups with an average length of
   30.1 ± 3.1 cm, indicative of substantial alleviation of colonic lesions
   post‐treatment (Figure [183]6b,c). Colonoscopy analysis showed that
   majority of mice treated with AOM/DSS exhibited colon inflammation or
   small‐sized tumors by week 4, indicating the successful establishment
   of the orthotopic CRC model. By week 11, colonoscopy examination
   revealed the presence of more and large tumors as well as severe
   colonic inflammation in all experimental groups except the EcN_ADA +
   DOX‐treated group, implying that the combinational therapy could
   relieve intestinal inflammation and inhibit the growth of orthotopic
   colorectal tumor (Figure [184]6d; Figures [185]S36 and [186]S37 and
   Movie [187]S1, Supporting Information).

Figure 6.

   Figure 6
   [188]Open in a new tab

   Combining engineered probiotics with low‐dose doxorubicin inhibit tumor
   growth in mouse orthotopic colorectal cancer model. a) Schematic
   illustration of the mouse orthotopic colorectal cancer model
   establishment and combinational therapy procedure. Mice body weights
   are recorded every three days. The days for the colonoscopy analysis
   are indicated. b) Representative colon figures of mice after different
   treatments. c) Statistical analysis of mice colon lengths from
   different treatment groups (n = 3). Error bars indicate standard
   deviation of three mice in each experimental group. Two‐tailed
   Student's t‐test was used to determine statistical significance. *p
   < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. d) Colonoscopy
   imaging shows the condition of colorectal carcinoma in mice at week 4
   (baseline) and week 11 (post‐treatment) across different treatment
   groups. Tumors are marked with white arrows, and ear tag IDs are used
   to distinguish mice within each group: Control (16, 27, 33), DOX (6,
   30, 39), EcN (38, 36, 51), EcN_ADA^D296A (20, 34, 37), EcN_ADA (12, 8,
   25), EcN + DOX (19, 2, 15), EcN_ADA^D296A + DOX (18, 7, 28), and
   EcN_ADA + DOX (22, 46, 1). e) Hematoxylin and eosin (H&E) staining of
   rolled small intestine sections showing villous architecture integrity
   and crypt depth in mice from different treatment groups. f) Ki67
   staining of rolled small intestine sections showing the oncogenic state
   of mice from different treatment groups.

   In consistent, histopathologic analysis of the mouse colon sections
   revealed significant epithelial barrier damage and crypt abscesses in
   untreated mice, those receiving monotherapy, and those treated with
   EcN_ADA^D296A + DOX. Notably, a substantial portion of the observed
   intestinal mucosal epithelial layer was shed, indicative of severe
   colonic inflammation or the presence of colonic tumors. In contrast,
   colon tissues from mice treated with EcN_ADA + DOX largely maintained
   the integrity of the epithelial barrier (Figure [189]6e).
   Immunohistochemically analysis of small intestine sections showed that
   mice treated with EcN_ADA + DOX had significantly longer villi and
   decreased expression of Ki‐67, a proliferation biomarker for tumor,
   compared to other experimental groups, indicative of an improvement in
   intestinal lesions (Figure [190]6f; Figure [191]S38, Supporting
   Information). Furthermore, staining with FITC anti‐Myc antibodies
   confirmed the presence of bioengineered bacteria surrounding the villi,
   indicating successful colonization of the small intestine villous walls
   by the orally administered bacteria (Figure [192]S39, Supporting
   Information). Notably, high levels of INF‐γ, TNF‐α, and IL‐12p70
   cytokines were detected in all treated groups, which were likely caused
   by severe colitis induced by prolonged DSS administration
   (Figure [193]S40, Supporting Information).

   The impact of EcN_ADA on intestinal microbiota was evaluated in light
   of its therapeutic potential. Species abundance analysis of the
   intestinal microbiota across the PBS, EcN_ADA^D296A, and EcN_ADA
   treatment groups revealed notable differences (Figure [194]S41a,
   Supporting Information). The EcN_ADA treatment significantly increased
   Parabacteroides abundance and showed trends of elevated Bacteroides
   acidifaciens, Bacteroides sartorii, and Lactobacillus johnsonii, all of
   which are associated with intestinal health (Figure [195]S42b,
   Supporting Information). Meanwhile, no significant changes in Chao1 and
   Shannon indices were observed, indicating that oral administration of
   engineered bacteria did not significantly affect overall microbial
   diversity or richness (Figure [196]S42c,d, Supporting Information).
   Collectively, all these results demonstrated that the combination
   therapy of EcN_ADA and low‐dose DOX could efficiently inhibit tumor
   growth in an orthotopic CRC mouse model.

3. Conclusion

   Immunotherapy has revolutionized cancer treatment, leveraging the
   patient's immune system to combat tumors. Despite its success, the
   immunosuppressive TME, largely mediated by adenosine metabolites, often
   hinders its efficacy. Targeting the adenosinergic pathway to mitigate
   this suppression holds great promise. In this study, we constructed a
   probiotic strain of E. coli Nissle 1917 (EcN) to express human ADA on
   its surface under hypoxic conditions typical of TME. This approach
   capitalizes on the unique capabilities of bacteria, particularly their
   ability to selectively colonize hypoxic tumor environments,
   self‐propel, and trigger anti‐tumor immune responses due to their
   intrinsic immunostimulatory nature. The engineered strain of EcN
   (EcN_ADA) not only enhances the local concentration of ADA, converting
   immunosuppressive adenosine into inosine and thus alleviating immune
   suppression, but also synergizes effectively with chemotherapeutic
   agents such as DOX. Our preclinical studies have demonstrated that
   EcN_ADA, particularly when used in combination with low‐dose DOX,
   markedly improves immune cell infiltration into tumors, reprograms
   macrophages from an anti‐inflammatory (M2) to a pro‐inflammatory (M1)
   phenotype, and substantially suppresses tumor growth in both
   subcutaneous and orthotopic mouse tumor models. While this study
   primarily validated the therapeutic efficacy of EcN_ADA using the MC38
   CRC model. Nonetheless, the strategy of utilizing probiotics to convert
   adenosine into inosine within the TME holds promise for treating a wide
   range of solid tumors, particularly those with high CD73 and CD39
   expression. These findings suggest that the synthetic biology approach
   of using engineered probiotics can effectively remodel the TME,
   enhancing the efficacy of existing cancer therapies and potentially
   leading to better clinical outcomes. Moreover, the adaptability of
   bacteria for genetic and chemical modifications opens up numerous
   possibilities for further enhancements. Future research should focus on
   optimizing the delivery and persistence of these engineered bacteria
   within the TME. Ensuring stable and controlled expression of
   therapeutic enzymes in response to specific tumor conditions remains a
   critical challenge. Additionally, combining this bacterial therapy with
   other immunomodulatory treatments, such as checkpoint inhibitors, could
   further enhance antitumor responses and improve patient outcomes.

   In summary, the development of engineered EcN_ADA represents a novel
   approach in cancer treatment, leveraging the natural properties of
   bacteria and the principles of synthetic biology to overcome the
   challenges of tumor‐induced immunosuppression. We believed that this
   strategy holds significant potential to enhance the effectiveness of
   cancer immunotherapies.

4. Experimental Section

Protein Expression and Purification

   Codon optimized full‐length ADA genes from Homo sapiens,
   Bifidobacterium pseudolongum and Lactobacillus johnsonii were
   synthesized by Genewiz and sub‐cloned into pET‐47b vector using
   restriction sites of BamHI and SacI, yielding pET47b‐ADA, pET47b‐BpADA
   and pET47b‐LjADA. Plasmid of inactive mutant ADA^D296A was generated
   via site‐directed mutagenesis using wild‐type pET47b‐ADA plasmid as DNA
   template. Primers for DNA amplification and site‐directed mutagenesis
   were listed in Table [197]S1 (Supporting Information). The constructed
   plasmids were transformed into BL21(DE3) cells for expression after
   sequencing. For ADA, BpADA and LjADA protein expression, transformed
   bacteria were cultured in Luria‐Bertani (LB) medium containing
   30 µg mL^−1 kanamycin. 5 mL overnight culture was transferred into 1 L
   fresh LB medium supplemented with respective antibiotics at 37 °C with
   agitation until OD[600] reaches 0.6. Isopropyl
   β‐D‐1‐thiogalactopyranoside (IPTG) was added to a final concentration
   of 0.5 mM to induce protein expression. Bacteria were further incubated
   at 25 °C for 16 h and collected by centrifugation at 4000 g for 10 min.
   Ni‐NTA buffer A (40 mM Tris‐HCl, pH 8.0, 500 mM NaCl, 10 mM imidazole,
   1 mM phenylmethanesulfonyl fluoride (PMSF) and 1 mM
   Tris(2‐carboxyethyl) phosphine (TCEP) was used to resuspend the
   bacteria cell pellet. The cell pellet was lysed by the high‐pressure
   homogenizer. The bacterial lysate was centrifuged at 20 000 g for
   30 min, then the supernatant was filtered by a 0.22 µm filter and
   loaded onto the Ni‐NTA column pre‐equilibrated by Ni‐NTA buffer A.
   Bound proteins were eluted by Ni‐NTA buffer B (40 mM Tris‐HCl, pH 8.0,
   500 mM NaCl, 500 mM imidazole, 1 mM PMSF and 1 mM TCEP). The eluted
   proteins were dialyzed against dialysis buffer (40 mM Tris‐HCl, pH 8.0,
   250 mM NaCl and 1 mM TCEP) and Precision protease was added to cleave
   the His6‐tag at 4 °C overnight. The dialyzed sample was then purified
   using a Superdex increase 200 10/300 column (Cytiva) pre‐equilibrated
   with GF buffer (20 mM Tris‐HCl, pH 7.5, 300 mM NaCl and 1 mM TCEP).

In Vitro Enzyme Activity Assay

   In vitro enzyme activity of ADA was determined by measuring the
   enzymatic generated NH[4] ^+ using the indophenol blue method as
   described previously.^[ [198]^52 ^] In brief, 10 mL of reagent A (1 mM
   adenosine in 20 mM HEPES buffer, pH 6.8) was mixed with 10 µL of
   purified enzyme (100 µM ADA) and incubated at 37 °C. An aliquot of
   50 µL of reaction sample was collected every 5 min. Subsequently, 50 µL
   of reagent B (106 mM phenol and 0.17 mM sodium nitroprusside) and 50 µL
   of reagent C (11 mM NaOCl and 125 mM NaOH) were added into the 50 µL
   reaction sample and vortexed immediately. The mixture was then
   incubated for an additional 30 min at 37 °C. Finally, reaction mixture
   was transferred to a 96‐well plate, and the absorbance was measured at
   628 nm using a plate reader (BioTek, USA).

Activation of Mouse CD8^+ T Cells

   Mouse CD8^+ T cells were enriched from spleen of 8‐week‐old female
   mouse (C57BL/6) by negative selection using an EasySep mouse Naïve
   CD8^+ T cell isolation kit (STEMCELL Technologies) according to the
   manufacturer's instructions. For CD8^+ T cells activation analysis,
   freshly isolated naïve mouse CD8^+ T cells were maintained in RPMI 1640
   medium (Gibco) supplemented with 10% FBS (Gibco) and mouse IL‐2
   (100 U mL^−1) (Pepro Tech). The cells were activated using plate‐bound
   anti‐mouse CD3 (clone 145‐2C11) (BioLegend) and anti‐mouse CD28 (clone
   37.51) (BioLegend) antibodies under the indicated experimental
   conditions for 24 h. For the analysis of surface marker CD25, live
   cells were stained with PerCP/cyanine5.5 anti‐mouse CD25 (eBioscience);
   After fixation and permeabilization, the treated CD8^+ T cells were
   stained with PerCP/Cyanine5.5 anti‐mouse IFN‐γ antibodies (BioLegend),
   followed by flow cytometry analysis. For the analysis of sustained
   effector T cell function under glucose‐depleted conditions, freshly
   isolated naive mouse CD8^+ T cells in RPMI 1640 medium supplemented
   with 10% FBS and 5 ng mL^−1 mouse IL‐2 (100 U mL^−1) and IL‐12
   (5 ng mL^−1) were pre‐activated by plate‐bound anti‐mouse CD3 and
   anti‐mouse CD28 antibodies for 24 h. Then CD8^+ T cells were then
   transferred into Glucose‐free RPMI 1640 medium (10% FBS) under the
   indicated experimental conditions and cultured for an additional four
   days. After fixation and permeabilization, the T cells were stained by
   PC5.5 anti‐mouse INF‐γ, FITC anti‐mouse TNF‐α and APC anti‐ mouse
   Granzyme B antibodies (BioLegend), followed by flow cytometry analysis.

Mouse Macrophage Polarization Analysis

   Mouse bone marrow cells from thigh bone of 8‐week‐old female mouse
   (C57BL/6) were harvested and cultured as previously described.^[
   [199]^53 ^] In brief, the mouse thigh bone was isolated and cut open.
   The bone marrow cells were washed out with PBS buffer supplemented with
   10% FBS. Red blood cells were then removed by incubating the bone
   marrow cells in red blood cell lysis buffer (Beyotime). The retained
   bone marrow cells were cultured in sodium pyruvate free‐DMEM (Gibco)
   medium supplemented with 10% FBS and 100 ng mL^−1 GM‐CSF (Sino
   biological) for another 6 days to differentiate into mature bone‐marrow
   derived‐macrophages (BMDMs). The BMDMs were either activated with
   100 ng mL^−1 IFN‐γ (Sino biological) and 100 ng mL^−1 LPS
   (Sigma–Aldrich) (M1‐like macrophages), or activated by 100 ng mL^−1
   IL‐4 (PeproTech) (M2‐like macrophages) for another 2 days. For
   macrophage polarization analysis, macrophage from indicated
   experimental conditions were stained with PE/Cyanine7 anti‐mouse F4/80,
   PE anti‐mouse CD86, and APC anti‐mouse CD206 antibodies (BioLegend),
   followed by flow cytometry analysis to distinguish between M1 and M2
   subsets.

Assessment of ADA Enzymatic Activity in Mouse Subcutaneous Tumor

   C57BL/6 mice aged six to eight weeks were subcutaneously inoculated
   with 10^6 MC38 cells. After the tumor volume reached ≈100 mm^3, the
   mice were administered a single injection of 100 nM ADA, ADA^D296A, or
   PBS. Twenty‐four hours post‐injection, the mice were euthanized, and
   tumors were excised and digested with collagenase IV (Sigma–Aldrich)
   into single‐cell suspensions. Following centrifugation, the supernatant
   was collected for HPLC analysis of adenosine and inosine
   concentrations. 100 nM lamivudine was added as an internal reference.
   Quantification was conducted using an Agilent 1290 Infinity II UHPLC
   system coupled to an Agilent 6475 Triple Quad tandem mass spectrometer
   in multiple reaction monitoring mode (MRM), operating in positive mode.
   Chromatographic separation was achieved on a ZORBAX RRHD Eclipse Plus
   C18 column (150 mm × 3 mm, 1.8 µm), with a 5 µL injection volume.
   Gradient elution utilized water with 0.1% FA as mobile phase A and
   acetonitrile as mobile phase B. The flow rate was maintained at
   300 µL min^−1. Additional instrument parameters included a nebulizer
   pressure of 28 psi, gas flow of 8.5 L min^−1, sheath gas flow of
   11.5 L min^−1, capillary voltage of 3200 V, nozzle voltage of 400 V,
   and temperatures set to 300 °C for the gas and 280 °C for the sheath
   gas. The parent ion/product ion pairs of adenosine (268.1/135.9),
   inosine (269.1/136.8) and lamivudine (230.1/94.8) were used for
   quantification, and the parent ion/product ion pairs 268.1/118.9,
   269.1/109.9 and 230.1/44.9 were used for qualification.

Macrophage Phagocytosis Assay

   Approximately 10^5 MC38 cells were seeded on a glass dish and stained
   with 10 ng mL^−1 Fluorescein Diacetate (Sigma–Aldrich) for 30 min.
   After washing with PBS to remove excess dye, 10^5 pre‐activated M1‐type
   RAW264.7 macrophages were co‐cultured with the MC38 cells under various
   drug treatments as indicated. After 2 h, APC anti‐mouse F4/80 antibody
   (BioLegend) was added to label the macrophages, and the phagocytosis
   was visualized using confocal imaging.

Bacterial Strain Construction

   The plasmid for E. coli surface‐display system was constructed as
   described previously.^[ [200]^31 ^] In brief, genes encoding the
   plasmid stabilization system Axe/Txe cassette, human ADA 1 with a
   lacUV5 promoter (P[lacUV5]), and the bacterial surface‐display system
   AIDA were synthesized by Genewiz. A pUC19 plasmid containing an
   ampicillin resistant gene (amp^R) and a high copy number origin of
   replication ColE1 was chosen as plasmid backbone. The three gene
   fragments and the pUC19 plasmid backbone were subsequently seamlessly
   assembled by Gibson assembly method to construct the plasmid pSYSU‐lac.
   For hypoxia‐induced protein expression, the lacUV5 promoter was
   replaced with the fnrS promoter by restriction enzyme digestion and
   ligation method, resulting the plasmid pSYSU‐fnrs. These plasmids were
   used to electroporate E. coli Nissle 1917. After electroporation, the
   transformed bacterial cells were selected as colonies on LB agar
   containing carbenicillin antibiotic. The plasmid maps, DNA sequences,
   and corresponding primers were available in Figure [201]S42,
   Tables [202]S2 and [203]S3 (Supporting Information).

Western‐Blot Analysis

   To investigate the ADA expression level in engineered bacteria,
   cultured bacteria in log phrase (OD[600] = 0.6) was collected. Dilute
   each group to the same OD[600] value, then add SDS‐PAGE loading buffer
   and heat in a 95 °C metal bath for 30 min for bacterial lysis.
   Bacterial total protein sample were separated by SDS‐PAGE and
   transferred to PVDF membranes. After blocking with 5% (wt/vol) non‐fat
   dry milk solution, the membranes were individually incubated with
   primary antibody, including anti‐myc (1:1000) (Proteintech) and
   anti‐GAPDH (1:2000) (Bioworld) polyclonal rabbit antibodies. After
   overnight incubation at 4 °C, the membranes were washed with TBST
   buffer and then incubated with goat anti‐rabbit HRP‐conjugated second
   antibody (1:1000) (Proteintech) for 1 h at 25 °C.

Flow Cytometry Analysis

   For the analysis of surface markers of immune cells, the cells were
   stained in PBS containing 3% (wt/vol) BSA and the appropriate
   antibodies. For intracellular cytokine staining of immune cells, the
   cells were fixed with 4% paraformaldehyde at room temperature for
   30 min, followed by permeabilization with 0.1% Triton X‐100 at room
   temperature for 10 min. The cells were then stained with the indicated
   antibodies. For the detection of bacterial ADA surface‐display, 500 µL
   of bacterial culture in log‐phrase (OD[600] = 0.7) were blocked in PBS
   buffer with 3% BSA for 20 min at 37 °C. After centrifugation at
   4000 rpm for 10 min, the bacterial pellets were resuspended in 100 µL
   PBS buffer containing FITC anti‐Myc‐ antibody (10 µg mL^−1) (Abcam) and
   incubated at room temperature for 30 min on a tube rotator.
   Subsequently, the bacterial pellets were collected by centrifugation
   and washed in 500 µL ice‐cold PBS buffer for three times, and followed
   by flow cytometer analysis. All the data were acquired on a CytoFLEX S
   flow cytometry (Beckman Coulter) and were analyzed with FlowJo 10.8.1
   software.

HPLC Quantification of Purine Nucleosides

   For analysis of adenosine‐to‐inosine conversion, the engineered
   bacteria with an OD[600] value of 0.25 were incubated with 1 mM
   adenosine in 10 mM HEPES buffer (pH 7.4) in a 37 °C shaker. An aliquot
   of 10 µL of the solution was collected every 1 h. After centrifugation
   and filtration, the concentrations of adenosine and inosine in the
   supernatant were quantitatively analyzed by HPLC (Thermo Fisher
   Ultimate 3000) equipped with a C18‐H column (4.6 × 250 mm, 5 µm, 120 Å)
   (Thermo Fisher) at a flow rate of 0.7 mL min^−1 at 25 °C. The mobile
   phase consisted of a linear gradient between 0.1% TFA in water (Buffer
   A) and 0.1% TFA in acetonitrile (Buffer B). The HPLC gradient program
   maintained 99% Buffer B for the first 5 min, then linearly decreased
   from 99% to 90% over the next 20 min (5–25 min). The peaks for inosine
   and adenosine appeared at 19 and 24 min, respectively. The chromatogram
   was recorded via UV absorption at 254 nm.

Immunofluorescence

   All the fluorescence images of the cells and bacteria were taken on a
   Carl Zeiss confocal microscope. To examine the effect of adenosine and
   inosine on M1 and M2 macrophage, polarized M1‐like or M2‐like
   macrophages were cultured in 35 mm confocal culture dishes for 24 h.
   Subsequently, the cells were incubated in the indicated experimental
   conditions for another 24 h and washed with PBS. After cell fixation
   with 4% paraformaldehyde and blocking, the cells were then stained by
   FITC anti‐mouse CD206 and APC anti‐mouse CD80 antibody (Biolegend) at
   4 °C for 12 h. The cell nuclei were stained with Hoechst 33 342
   (Beyond). To investigate the ADA expression level on bacterial surface,
   500 µL of bacterial culture in log‐phrase (OD[600] = 0.7) were blocked
   in PBS buffer with 3% BSA for 20 min at 37 °C. After centrifugation at
   4000 rpm for 10 min, the bacterial pellets were incubated with 100 µL
   PBS buffer containing FITC anti‐Myc antibody (10 µg mL^−1) at room
   temperature for 30 min on a tube rotator. Subsequently, the bacterial
   pellets were collected by centrifugation and washed in 500 µL ice‐cold
   PBS buffer for three times, and followed by fluorescence images
   collection.

Subcutaneous Tumor Models and In Vivo Studies

   The animal experiment procedures were approved by the Institutional
   Animal Care and Use Committee of Sun Yat‐sen University (approval
   number: 2024001413) and performed in accordance with the ethical
   guidelines of Sun Yat‐sen University. All the experiments were
   conducted in the Laboratory Animal Center of Sun Yat‐sen University.
   8‐week‐old SPF‐grade female C57BL/6J mice were purchased from
   GemPharmatech Co., Ltd., Nanjing, Jiangsu, China. MC38 tumor cells (2 ×
   10^6 cells in 100 µL PBS) were subcutaneously injected into the left
   mouse flank. Nine days after tumor implantation, mice with similar
   tumor sizes were randomly divided into different experimental groups.
   To evaluate the clinical efficacy of the purified ADA enzyme combined
   with the chemotherapy agent DOX in cancer therapy, the mice were
   randomly assigned to 6 groups of 6 mice each: Control, DOX, ADA^D296A,
   ADA, ADA^D296A + DOX, and ADA + DOX. Approximately 300 µg of ADAD^296A
   or ADA were administered via peritumoral injection every three day,
   while low‐dose DOX (20 µg) was delivered intravenously via the tail
   vein every three days. To investigate the synergistic anti‐cancer
   effect of EcN_ADA combined with low‐dose DOX chemotherapy, mice were
   randomly grouped into 8 groups of 6 mice each: Control, DOX, EcN,
   EcN_ADA^D296A, EcN_ADA, EcN + DOX, EcN_ADA^D296A + DOX, and EcN_ADA +
   DOX. 20 µg of DOX in a total volume of 100 µL was administered via
   intravenous injection for each mouse every three days. Tail vein
   injections of bacteria were conducted at a concentration of 1 ×
   10^6 CFU mL^−1 in PBS, with a total volume of 100 µL injected per mouse
   every five days. The length (L), width (W), and height (H) of the
   tumors and the body weight of mice were measured and recorded every 3
   day. The tumor volume (V) was calculated using the formula V (mm^3) = L
   (mm) × W (mm) × H (mm) × 0.5. The tumor‐bearing mice were euthanized at
   the indicated day after tumor implantation, and the tumors were
   stripped out for photograph and weight. For immunohistochemistry (IHC)
   analysis, the tumors were fixed with 4% paraformaldehyde at 4 °C and
   embedded in paraffin, and then the CD3^+ cells, CD8^+ cells, NK cells
   and macrophage (M1 and M2) were visualized by immunohistochemically
   staining. For quantitative assessment of resident bacteria in different
   organs, the heart, liver, spleen, lungs, and kidneys of the mice were
   harvested, weighed, and ground in sterile saline. The homogenates were
   then subjected to plate spreading (ampicillin‐resistant) to enumerate
   the remaining bacteria. For cytokine quantification, mice serum was
   collected, and the levels of INF‐γ, TNF‐α, and IL‐12p70 were measured
   using commercial ELISA kits according to the manufacturer's
   instructions.

Quantification of Adenosine and Inosine in Murine Tumor Tissues

   Adenosine and inosine were quantified using the solid‐phase
   microextraction (SPME) technique by referring to the literature.^[
   [204]^54 ^] Home‐made HLB (hydrophile‐lipophile balance) SPME fibers
   were used to extract adenosine and inosine from tumor tissues. Briefly,
   medical‐grade stainless steel wires (SSWs, 400 µm diameters) were
   sonicated in concentrated hydrochloric acid (1 mol/L) to ≈10 min. The
   HLB (hydrophile‐lipophile balance) particles (20% of the total volume)
   were combined with polyacrylonitrile (PAN, average molecular weight
   250 000, 15% w/w in DMF solvent) under continuous stirring for 24 h.
   Utilizing a dip‐coating technique. Kinetic Calibration^[ [205]^55 ^]
   was employed to quantify free adenosine and inosine in the vicinity of
   tumor cells within mice models. Lamivudine (Thermo Scientific, USA) was
   served as the internal standard owing to its structural similarity to
   adenosine and inosine. Mice were anesthetized with isoflurane, and the
   HLB‐PAN fiber was then introduced into the tumor, penetrating ≈1 cm
   deep. After the designated 30 min sampling duration, the fiber was
   carefully retracted under vortex at 550 rpm for 30 min. Quantification
   was conducted using an Agilent 1290 Infinity II UHPLC system coupled to
   an Agilent 6475 Triple Quad tandem mass spectrometer in multiple
   reaction monitoring mode (MRM), operating in positive mode.
   Chromatographic separation was achieved on a ZORBAX RRHD Eclipse Plus
   C18 column (150 mm × 3 mm, 1.8 µm), with a 5 µL injection volume.
   Gradient elution utilized water with 0.1% FA as mobile phase A and
   acetonitrile as mobile phase B. The flow rate was maintained at
   300 µL min^−1. Additional instrument parameters included a nebulizer
   pressure of 28 psi, gas flow of 8.5 L min^−1, sheath gas flow of
   11.5 L min^−1, capillary voltage of 3200 V, nozzle voltage of 400 V,
   and temperatures set to 300 °C for the gas and 280 °C for the sheath
   gas. The parent ion/product ion pairs of adenosine (268.1/135.9),
   inosine (269.1/136.8) and lamivudine (230.1/94.8) were used for
   quantification, and the parent ion/product ion pairs 268.1/118.9,
   269.1/109.9 and 230.1/44.9 were used for qualification.

In Vivo Bioluminescence Imaging of Engineered Probiotics

   To monitor the retention time of engineered probiotics in subcutaneous
   tumors or intestines of mice, the LuxCDABE reading frame fragment was
   inserted into the pSYSU‐fnrs plasmid and electroporated into EcN to
   generate EcN_Lux_ADA. A total of 10^5 CFU of EcN_Lux_ADA was
   intravenously injected into tumor‐bearing (MC38) mice, and bacterial
   accumulation was tracked daily. After isoflurane anesthesia, luminous
   signals were measured using a chemiluminescent imager (Fx6, Vilber,
   France) to determine bacterial numbers and distribution. For intestinal
   retention analysis, 10 mice were gavaged with 10^9 CFU of EcN_Lux_ADA.
   At predetermined intervals (12 h and 24 h, 1 day, 2 days, 3 days, 4
   days, and 6 days post‐gavage), one mouse was euthanized per time point.
   The intestines, along with the heart, liver, spleen, lungs, and
   kidneys, were harvested to evaluate bacterial distribution.

Single‐Cell RNA Sequencing

   To prepare the single‐cell RNA sequencing samples, tumor tissues were
   rinsed three times with Hank's balanced salt solution (HBSS) and then
   minced into fragments measuring 1–2 mm in diameter. The tissue
   fragments were digested using 2 mL of GEXSCOPE tissue dissociation
   solution (Singleron) at 37 °C for 15 min with continuous agitation.
   Following digestion, the resulting cell suspension was filtered through
   a 40‐micron sterile filter and centrifuged at 1000 rpm for 5 min. The
   supernatant was discarded, and the cells were resuspended in 1 mL of
   PBS (HyClone). The suspension was then centrifuged at 500 g for 5 min,
   and the pellet was resuspended in 1 mL of PBS. Cell viability and
   counts were determined using Trypan blue staining (Sigma) under a
   microscope. Single‐cell suspensions were prepared in PBS (HyClone) at a
   concentration of 1 × 10^5 cells/mL. The suspensions were loaded onto
   microfluidic chips, and single‐cell RNA sequencing libraries were
   constructed using the GEXSCOPE single‐cell RNA library kit (Singleron
   Biotechnologies) in accordance with the manufacturer's instructions.
   Libraries were diluted to 4 nM and sequenced using an Illumina
   NovaSeq6000 platform with 150 bp paired‐end reads. Raw sequencing data
   were processed using Singleron's internal analysis pipeline to generate
   a gene expression matrix. Briefly, reads lacking poly T sequences were
   removed, and valid cell barcodes and unique molecular identifiers
   (UMIs) were extracted. Reads were filtered to remove adapters and poly
   A tails (using fastp V1), and aligned to the reference genome from the
   Ensemble database for quantification (using STAR 2.5.3a and
   featureCounts 1.6.2). Reads, UMIs, and genes with the same cell barcode
   were grouped, and the UMI counts per gene in each cell were calculated
   for downstream analysis. The Seurat package^[ [206]^56 ^] was employed
   for cell type identification and clustering analysis of the RNA
   sequencing data. Expression matrices were imported into R using the
   read. Table function, and cell clusters were determined using the
   FindClusters function with a resolution parameter of 0.6.
   Differentially expressed genes (DEGs) between different samples or
   across clusters were identified using the FindMarkers function. GO
   enrichment analysis of gene sets was conducted using the
   clusterProfiler package to identify biological functions or pathways
   significantly associated with specific gene expressions.

Orthotopic Colorectal Cancer Model and Colonoscopy Analysis

   The animal experiment procedures were approved by the Institutional
   Animal Care and Use Committee of Sun Yat‐sen University (approval
   number: 2024001414) and performed in accordance with the ethical
   guidelines of Sun Yat‐sen University. All the experiments were
   conducted in the Laboratory Animal Center of Sun Yat‐sen University.
   C57BL/6 mice (4–6 weeks) were treated with 7.5 mg kg^−1 azoxymethane
   (AOM) (Sigma) by intraperitoneal injection at the beginning of the
   experiment. After 7 days, the mice were given 5 cycles of 2.5% DSS in
   drinking water for 7 days, followed by regular water for 14 days. Mice
   intestinal endoscopy equipment (ENDOQ) was used to detect the colon
   cancer lesions in mice on day 29 and day 78 to assess the therapeutic
   effects of different treatments. At 78 day, the mice were sacrificed,
   and blood and tissues were collected for subsequent analysis and
   experiments. The length of colon was measured to assess the severity of
   CRC. Tissue samples of colon cross‐sections and Swiss rolls were
   preserved using paraformaldehyde, subsequently encased in paraffin
   blocks, sliced into thin sections, and then subjected to H&E staining
   and ki‐67 immunofluorescent staining. For H&E staining, histological
   pathology evaluation for acute colitis and CRC was performed based on
   the severity, extent of damage, inflammation, and regeneration of the
   colon. For immunofluorescent staining, DAPI and FITC anti‐Myc antibody
   was used to visualize the distribution of engineered bacterial in the
   colon.

Conflict of Interest

   The authors declare no conflict of interest.

Author Contributions

   W.X., Z.‐W.M. and J.‐H.W. conceived and designed the study. J.‐H.W.,
   J.W., Z.Y. and J.‐Q.W. conducted the animal model experiments. C.Z.,
   Y.L. and J.‐H.W. conducted the T Cells experiments. H.W. and J.X.
   participated in solid‐phase microextraction experiment and data
   analysis. J‐H.W., J.Z. and C.X. participated in scRNA‐seq analysis.
   W.X. and J‐H.W. wrote the manuscript.

Supporting information

   Supporting Information
   [207]ADVS-12-2411813-s002.pdf^ (5.3MB, pdf)

   Supplemental Movie 1
   [208]Download video file^ (11.9MB, mp4)

Acknowledgements