A metabolically engineered bacterium controls autoimmunity and inflammation by remodeling the pro-inflammatory microenvironment

ABSTRACT Immunotherapy has led to impressive advances in the treatment of autoimmune and pro-inflammatory disorders; yet, its clinical outcomes remain limited by a variety of factors including the pro-inflammatory microenvironment (IME). Discovering effective immunomodulatory agents, and the mechanisms by which they control disease, will lead to innovative strategies for enhancing the effectiveness of current immunotherapeutic approaches. We have metabolically engineered an attenuated bacterial strain (i.e., Brucella melitensis 16M ∆vjbR, Bm∆vjbR::tnaA) to produce indole, a tryptophan metabolite that controls the fate and function of regulatory T (Treg) cells. We demonstrated that treatment with Bm∆vjbR::tnaA polarized macrophages (Mφ) which produced anti-inflammatory cytokines (e.g., IL-10) and promoted Treg function; moreover, when combined with adoptive cell transfer (ACT) of Treg cells, a single treatment with our engineered bacterial strain dramatically reduced the incidence and score of autoimmune arthritis and decreased joint damage. These findings show how a metabolically engineered bacterium can constitute a powerful vehicle for improving the efficacy of immunotherapy, defeating autoimmunity, and reducing inflammation by remodeling the IME and augmenting Treg cell function.


Introduction
Recent advances in cancer immunotherapy have not only revolutionized the field of tumor immunology, -1-3 but also rejuvenated research into exploring novel strategies for autoimmune immunotherapy. Numerous studies have demonstrated promising results using immunotherapy to treat autoimmune and pro-inflammatory disorders. [4][5][6] Nevertheless, despite advances in the field of autoimmune-immunotherapy, including regulatory T cell (T reg )-based therapy, the efficacy and benefits of these approaches remain less satisfactory due to limited in vivo T reg expansion and persistence, the pro-inflammatory microenvironment (IME), and insufficient T reg trafficking to inflamed sites. In addition, immune dysregulation of the IME contributes to disease. 7 Immunotherapeutic combinations may produce greater efficacy, and thus strategies that circumvent these barriers are urgently needed.
To address this need, we developed and tested an intervention that combined two innovations. First, although a therapeutic role for live attenuated bacterial vaccines in addressing infectious diseases is undeniable and appreciated in cancer immunotherapy, the use of bacterial agents to manage autoimmune and pro-inflammatory diseases remains limited. [8][9][10] Second, a growing number of microbiota-specific products and metabolites have novel immunologic properties that constitute one mechanism whereby the microbiota influence host health and disease. [11][12][13][14] We therefore developed a metabolically engineered bacterial vaccine strain to produce immunomodulatory metabolites that improve autoimmunity and inflammation.

Results
For the bacterial vector, we selected an attenuated strain of Brucella melitensis that harbors a deletion in vjbR, a master regulator of virulence (Bm∆vjbR). 15 Like other Gram-negative organisms, Brucella strains express a lipopolysaccharide (LPS) lacking endotoxin activity. Importantly, Bm∆vjbR has been shown to be safe in immunocompetent and immunocompromised mice, 16 goats, 17 sheep, 18 and non-human primates. 19 We have also shown that Bm∆vjbR can combat cancer in a murine model by remodeling the tumor microenvironment (TME) to a pro-inflammatory state. 15 Moreover, when Bm∆vjbR treatment was combined with adoptive cell transfer (ACT) of tumor antigen (Ag)-specific CD8 + T cells, tumor growth and proliferation were dramatically impaired. 15 Conversely, in the current studies, we engineered Bm∆vjbR to express tryptophanase (tnaA); i.e., BmΔvjbR::tnaA, to produce the tryptophan metabolite indole, a molecule that modulates the fate and function of T regs . 20 We have reported that indole, when used at a range of physiologic concentrations, suppresses several inflammatory characteristics in immune and nonimmune cells, 21 and also augments T reg differentiation. 22 Consistent with our previous reports, we demonstrated that indole suppressed TNF-αproduction in CD11b + spleen cells after E. coli LPS (eLPS) and heat-inactivated Salmonella Typhimurium [HKST] stimulation ( Figure 1a & Figure 1b) and dampened their activation by suppressing Akt and ERK signaling pathways in response to microbial agonists (eLPS and HKST) (Fig. S1a). In addition, indole augmented the differentiation of naive CD4 + CD25 − T cells into induced T regs (iT regs ) measured by FoxP3 in vitro in a dose dependent manner (Figure 1c & Figure  1d). These findings were consistent with our earlier reports, 15 and were comparable to results from studies using the microbiota metabolite butyrate, albeit with distinct dose-dependency. 23 Based on these findings, we hypothesized that indole would ameliorate immune-mediated inflammation in autoimmune and pro-inflammatory diseases.
Thus, we examined whether indole reduces autoimmune responses in a murine collagen-induced arthritis (CIA) model. We delivered 20 mg/kg indole in corn-oil to C57BL/6 mice (N = 5) having CIA and assessed the arthritis score in these mice compared to untreated control (PBS; N = 5). First, we showed that the severity of CIA was significantly attenuated in indole treated mice, which exhibited clinical scores of 0.8 ± 0.2 (means ± SEM) at (Day 50), compared to 1.6 ± 0.5 in controls ( Figure 1e). However, a single dose of indole only showed a slight decrease in inflammation. Similarly, single dose treatment did not induce significant alterations in the infiltration of T reg cells as assessed by confocal microscopy (Figure 1f & Figure 1g). These findings were in striking contrast to our ex vivo experimental findings, which indicated that indole significantly promoted the expansion of CD4 + FoxP3 + T reg cells and enhanced their activation by increased expression of PD-1 an immunosuppressive molecule, compared to the controls (p < .001), in cells derived from the mouse lymph nodes (LNs) and spleen ( Figure 1h & Figure 1i). Based on our in vivo and ex vivo findings, we hypothesized that the sustained delivery of indole in a bacterial vector may greatly improve the durability of the molecule's immunomodulatory effects, resulting in an attenuation of autoimmunity and inflammation in CIA. To test this hypothesis, we engineered a safe live-attenuated bacterial strain (i.e., BmΔvjbR::tnaA) to constitutively produce indole (Figure 2a-c). First, we observed that the engineered bacterial strain survived mainly in the spleen for 7 days and liver and kidney for 3 dpi (Figure 2d). The bacteria did not penetrate the joints of the mice (Figure 2d). Second, we observed low-level immunogenicity by detection of anti-Brucella IgG antibodies from 3 to 21 dpi of the bacteria (Figure 2e). Moreover, we also determined that the colonies of BmΔvjbR::tnaA bacterial strain recovered from mice challenged with collageninduced arthritis (CIA) at 7 dpi could still produce indole ( Figure 2f). Further, cytokine array profiling analyses showed that BmΔvjbR::tnaA induced the expression of IL-10 in macrophages (Mφ) ( Figure  3a & Fig. S2a), which promotes the activities of T reg cells and reduces autoimmunity and inflammation. Strikingly, BmΔvjbR::tnaA also significantly (p < .01) reduced the expression of additional proinflammatory cytokines like IL-6, IL-1β and TNFα in macrophages (Mφ) compared to BmΔvjbR parental strain (Figure 3a). We also found that BmΔvjbR::tnaA, when co-cultured with bone marrow-derived Mφ (BMDMs), not only significantly reduced the total CD4 + T cells (p < .001) but also reduced the production of the pro-inflammatory cytokines such as TNF-α and IFN-γ (p < .001) compared to the BmΔvjbR parental strain ( Figure  3b). Moreover, BmΔvjbR::tnaA promoted the expansion of T reg cells and significantly enhanced their activity as assessed by IL-10 production (p < .001) and PD-1 expression (p < .01) (Fig. S2a). Third, in the CIA model (Fig S1b), a significant reduction in arthritis score and incidence was observed following treatment with BmΔvjbR:: tnaA. This amelioration of autoimmunity and inflammation was further augmented when BmΔvjbR::tnaA treatment was combined with ACT of T reg cells ( Figure 3c). Fourth, we observed significantly (p < .01) reduced numbers of infiltrating inflammatory cells into the joints of mice treated with BmΔvjbR::tnaA. This effect was further enhanced by BmΔvjbR::tnaA treatment followed by the ACT of T reg cells compared to the controls (p < .001) (Figure 3d). Finally, mice treated with BmΔvjbR::tnaA showed reduced infiltrates in the joint and intact cartilages as evidenced by H&E analysis and Safranin O (Saf-O) staining of knee cross-sections (60 days post collagen administration). Notably, these findings were further attenuated by addition of ACT of T reg cells (Figure 3d). There was also a significant reduction in the total CD4 + T cell proportion in the joints and a significant increase in T reg cell proportion in mice treated with BmΔvjbR::tnaA compared to controls (p < .001; Figure 3e). Since it is unlikely that BmΔvjbR::tnaA bacteria will be used prior to the onset of arthritis in the clinic, we conducted an experiment to assess the impact of our strategy in treatment of established CIA by starting the administration of bacteria and T reg cells, 3 weeks post collagen administration. Our results indicate that BmΔvjbR::tnaA combined with ACT of T reg cells significantly dampens the progression of the disease but is not sufficient to completely cure the disease (Fig S3a and S3b).
To identify the mechanism by which BmΔvjbR:: tnaA might be acting, we conducted a multiparametric CyTEK analysis from the cells isolated from the spleen, LNs and/or the knee and ankle joints of control, ACT with T reg cells only, or ACT with T reg cells plus BmΔvjbR::tnaA groups. We found that the BmΔvjbR::tnaA reduced the proportion of B cells (Figure 3f, Figure 3g, Fig. S2b, and Table  S1) in addition to promoting T reg cell expansion. Overall, our results indicate that BmΔvjbR::tnaA remodels the IME and facilitates the expansion and suppressive function of T reg cells as depicted in the illustration ( Figure 4). Our results also indicate that BmΔvjbR::tnaA may modulate B cellmediated immunity in our murine model of CIA to alleviate symptoms of arthritis in these mice.

Discussion
Several strategies have been employed to enhance the efficacy of live attenuated bacterial vaccines. 13,14 However, bacterial vaccines producing immunomodulatory metabolites that alter the immunological tolerance and the IME have not been previously reported. Our work features both conceptual and methodological innovations. First, this work provides the first description of a live attenuated vaccine whose metabolism has been reprogrammed to amplify anti-autoimmune/inflammation activity.
Second, this study provides the first description of how combining a single dose of an engineered live attenuated bacterial vaccine strain with the ACT of T reg cells can achieve potent therapeutic outcomes. Previous studies have demonstrated the efficacy of the administration of a single dose of polyclonal T reg cells in controlling CIA in DBA/1 mouse model. 24 Moreover, Sun et al have also Intracellular FoxP3 staining (green) was used to denote Treg cells. g, Quantitative analysis of H&E, Saf-O and Treg cell infiltration from confocal microscopy sections of Control (Ctrl) and indole-treated mice. h, Flow cytometric dot-plot analysis of PD-1 and FoxP3 in ex vivo activated CD4+ T cells isolated from LNs and spleen of C57BL/6 mice. Exposure to indole drives these cells towards higher Treg cell phenotype by increased FoxP3 expression. Vehicle (control) indicates corn-oil which was used as a solvent for indole. i, Graphical representation of FoxP3 derived from the flow cytometric dot-plots of CD4+ T cells exposed to indole. Graphical representation of PD-1+ FoxP3+ T cells (%) from the flow cytometric dot-plots. Data represent means ± SD. Student's t-test or Tukey's multiple comparisons test was applied for statistical analysis. *, **, ***: significance at p < 0.05, 0.01, 0.001. demonstrated the efficacy of antigen specific T reg cells in suppressing inflammation of CIA in C57BL/ 6 mouse model. 25 However, these studies are limited by the durability of anti-inflammatory responses of antigen specific T reg cells and the possibility of reversion to pathogenic effector T-cell type. Moreover, the studies did not show the efficacy of therapeutic adoptive T reg cell therapy in controlling CIA in mice. Although previous studies have reported the efficacy of ACT of T reg cells in autoimmune diseases 26 our study is the first report showing the enhanced efficacy of combinatorial adoptive immunotherapy in C57BL/6 CIA mouse model. We have also shown that ACT of T reg cells  not only prevents CIA in mice, but also controls the progression of the disease when administered subsequent to the onset of the disease. More importantly, our study also provides an overview of the immune landscape in the CIA mouse model and shows how the adoption of our combinatorial strategy breaks immunological tolerance and ameliorates CIA.
Third, we demonstrate how the genetic tractability of BmΔvjbR can be exploited to engineer vaccines with novel properties. Although in this work, we focused on metabolic reprogramming of the bacterium, we envision that future iterations of the BmΔvjbR::tnaA vaccine will include engineered auto antigens (Ags) to boost persistent auto Agspecific T reg cell responses. Fourth, this work showed that a single administration of indole did not succeed as a stand-alone agent. However, this limitation was circumvented by indole delivery using our engineered bacterial vector. BmΔvjbR:: tnaA may offer improved pharmacodynamics for indole by sustained production levels and effects in vivo versus indole alone. Moreover, we showed that the separate individual beneficial effects of BmΔvjbR and indole act synergistically in vivo.
Finally, the natural localization of BmΔvjbR to leukocytes enabled BmΔvjbR::tnaA to provide improved targeted indole delivery and a higher local effective indole concentration in contrast to the untargeted and transient nature of indole as a single agent. In sum, our work presents an attractive new avenue for the development of anti-autoimmunity/inflammation vaccines.

CIA induction
CIA was induced as described previously with minor modifications. 27 Briefly, male C57BL/6 mice were injected with an emulsion of 100 µl of chick type II collagen (Chondrex; 100 µg) in Complete Freund's Adjuvant (CFA; Chondrex) using a glass tuberculin syringe with 26-gauge needle. The mice were then assessed for development of joint inflammation and clinical arthritis score until Day 60 as described previously. 28

Indole treatment in CIA mouse
CIA was induced in male C57BL/6 mice. On Day 7 after the CIA induction, mice were intraperitoneally (i.p). injected with 20 mg/kg indole or corn oil (vehicle control).

Bacterial culture
Bm∆vjbR or BmΔvjbR::tnaA were cultivated and prepared for experimentation as previously described. 15

Engineering indole-producingBmΔvjbR::tnaA strain
To generate an indole producing attenuated Bm∆vjbR strain, we cloned an Escherichia coli (E. coli) tnaA gene into a broad range bacteria expression plasmid (pBBR1MCS6Y) 29 and transferred the plasmid to Bm∆vjbR.

Indole Detection and quantification
The indole production by Bm∆vjbR::tnaA was detected by liquid chromatography-mass spectrometry (LC-MS). After 24 hr cultivation of the bacteria in Tryptic Soy Broth (TSB) medium at 37°C, the metabolites were extracted using ice-cold methanol for LC-MS assay. Liquid chromatography tandem mass spectrometry analysis was performed on a TSQ Altis triple quadrupole mass spectrometer also shown. e, Cells from knee and ankle joints were collected from CIA-induced mouse groups (Ctrl, BmΔvjbR, and BmΔvjbR::tnaA combined with ACT of Treg cells). These cells were then stained and quantified by flow cytometry using markers for CD4+ T cells (Thermo Scientific, Waltham, MA) coupled to a binary pump HPLC (Vanquish, Thermo Scientific). The indole concentration was measured using an indole assay kit following the manufacturer's protocol (Sigma-Aldrich).
For treatment of established CIA, mice were injected with 5.0 × 10 7 live BmΔvjbR::tnaA or indole/PBS control on Day 21 post injection of CFA +Collagen followed by T reg cell injection on Day 25.  The illustration depicts the process of amelioration of CIA in C57BL/6 mice by the engineered BmΔvjbR::tnaA bacterial strain. C57BL/6 mice may harbor a microbiome deficient in production of the metabolite indole. When challenged with the CIA, these mice rapidly develop arthritis due to dysfunctional Tregs and rapid expansion of CD4+ effector T cells. However, when BmΔvjbR::tnaA bacteria are administered to mice bearing CIA, the Treg activity is substantially increased and the pathogenic activity of CD4+ effector T cells is compromised, which results in amelioration of joint inflammation and improved symptoms. This activity is greatly augmented by an ACT of T regs . and joints were homogenized and plated on Tryptic Soy Agar (TSA) plates supplemented with chloramphenicol. The CFU was enumerated after 3-day post-cultivation of the bacteria.

Serum ELISA for detection of BmΔ specific IgG antibody
The CIA induced C57BL/6 mice were sacrificed at 1, 3, 7, 14, and 21 dpi of BmΔvjbR and/or BmΔvjbR::tnaA bacteria. Blood samples were collected from the mice and serum was isolated by coagulation of the blood at room temperature followed by centrifugation at 2,000 × g for 20 minutes. The serum sample was assayed for anti-BmΔvjbR IgG antibody by using mouse Brucella antibody IgG ELISA kit following manufacturer's instructions (AFG Scientific).

Cytokine responses
BMDMs were seeded in 24-well plates at a concentration of 2.0 × 10 5 cells/well in DMEM without antibiotics. After overnight culture, the cells were inoculated with Bm∆vjbR or Bm∆vjbR::tnaA bacteria at a multiplicity of infection of 20. At 24 h post-treatment, cellular supernatant was collected and analyzed for the presence of cytokines/chemokines by using a Proteome Profiler Mouse Cytokine Array Kit (R&D Systems, Inc.).

Flow cytometric analysis
Cell staining and flow cytometric analysis were performed as described previously 15 using the described labeling reagents. Briefly, surface and intracellular staining was performed on the singlecell suspensions and analyzed using LSR Fortessa cell analyzer (BD). The spleen, joints and LNs were also processed and stained similarly with antibodies listed in Table S2, and data was acquired on CyTEK aurora flow cytometer (Cytek Biosciences). For multiparametric analysis, the data were analyzed with FlowJo v10 and represented as heatmaps and tSNE plots.

Histology and immunofluorescence
Mice were humanely sacrificed on day 60 after induction of CIA, and tissue sections were analyzed as previously described. 27 Briefly, the hind foot paws and knees were removed and fixed in 10% formalin and decalcified in Formical-4 (Decal chemical, Tallman, NY). The fixed tissue sections were then stained with H&E and/or Safranin O fast green (Saf-O) stain. The H&E and Saf-O stained sections were then assessed by a semiquantitative system of 0 to 4 as described previously. Immunofluorescent staining and microscopy were performed on the deparaffinized sections by using FITC anti-mouse FoxP3 antibody (Ab) for T reg cells and DAPI as nuclear stain.

Statistical analysis
One-Way ANOVA, Student's t-test, or Tukey's multiple comparisons test was performed for statistical analysis between the groups. All analyses were performed in GraphPad Prism v9. A p value of <0.05 was considered statistically significant.

Key messages
What is already known about this subject?
• Bm∆vjbR has been shown to be safe in immunocompetent and immunocompromised animals. • Microbiota-specific products and metabolites have novel immunologic properties.
What does this study add?
• Novel bacterial platform produce immunomodulatory metabolites that alter the immunological tolerance and the IME.
How might this impact on clinical practice or future developments?
• Combining a single dose of an engineered live attenuated bacterial therapeutic with ACT of T regs can achieve potent therapeutic outcomes.
The use of engineered bacterial agents is an innovative and malleable approach to manage autoimmune and pro-inflammatory diseases.