MyD88-dependent BCG immunotherapy reduces tumor and regulates tumor microenvironment in bladder cancer murine model

Bacillus Calmette-Guerin (BCG) is the only FDA approved first line therapy for patients with nonmuscle invasive bladder cancer. The purpose of this study is to better understand the role of innate immune pathways involved in BCG immunotherapy against murine bladder tumor. We first characterized the immunological profile induced by the MB49 mouse urothelial carcinoma cell line. MB49 cells were not able to activate an inflammatory response (TNF-α, IL-6, CXCL-10 or IFN-β) after the stimulus with different agonists or BCG infection, unlike macrophages. Although MB49 cells are not able to induce an efficient immune response, BCG treatment could activate other cells in the tumor microenvironment (TME). We evaluated BCG intratumoral treatment in animals deficient for different innate immune molecules (STING−/−, cGAS−/−, TLR2−/−, TLR3−/−, TLR4−/−, TLR7−/−, TLR9−/−, TLR3/7/9−/−, MyD88−/−, IL-1R−/−, Caspase1/11−/−, Gasdermin-D−/− and IFNAR−/−) using the MB49 subcutaneous mouse model. Only MyD88−/− partially responded to BCG treatment compared to wild type (WT) mice, suggesting a role played by this adaptor molecule. Additionally, BCG intratumoral treatment regulates cellular infiltrate in TME with an increase of inflammatory macrophages, neutrophils and CD8+ T lymphocytes, suggesting an immune response activation that favors tumor remission in WT mice but not in MyD88−/−. The experiments using MB49 cells infected with BCG and co-cultured with macrophages also demonstrated that MyD88 is essential for an efficient immune response. Our data suggests that BCG immunotherapy depends partially on the MyD88-related innate immune pathway.

does not seem to be essential in the initial control of BCG infection, but it is required to activate a robust Th1 response [9][10][11][12] . Therefore, TLRs play an important role in the response to BCG immunotherapy and many TLR agonists have been considered as candidates for bladder cancer therapy 5,13-15 . cGAS-STING pathway is another important mechanism of the innate immunity for cytosolic DNA recognition, promoting the activation of TBK1 and phosphorylation of IRF-3, a transcription factor that induces the expression of IFN-β 16 . Mycobacterium tuberculosis and M. bovis (BCG) induce the production of IFN-β depending on the detection of c-di-AMP (cyclic-di-adenosine monophosphate) and activation of the STING pathway, which leads to the increase of autophagy in macrophages and bacterial control in mice 17 . STING is involved in the antitumor immune response playing an essential role in the recognition of cancer cells and in the activation of type I IFN-dependent cytotoxic T cell response 18,19 . The therapeutic intratumoral administration of cGAMP (cyclic dinucleotide GMP-AMP) or CDNs (cyclic di-nucleotides) suppresses tumor growth, presumably through the direct activation of STING in the tumor microenvironment (TME), leading to the activation of phagocytic cells-dependent cytotoxic T lymphocytes response [20][21][22] . STING also plays an important role in the antitumor effects associated with chemotherapeutic agents 23 and is essential for the immunotherapeutic response of radiation-induced T cells 24,25 . However, the role of the cGAS-STING pathway and IFN-β in BCG immunotherapy have been poorly investigated.
Tumor immunotherapy depends on the cell profile present in TME, which can promote a predominantly proor anti-inflammatory response. This initial innate immune response in the TME is mainly induced by dendritic cells and macrophages (tumor-associated macrophages-TAM), which may present a polarization type M1 (proinflammatory, anti-tumor) or M2 (anti-inflammatory, pro-tumor), essential for the recruitment of lymphocytes to the tumor infiltrate (tumor-infiltrating lymphocytes-TIL) and the activation of an efficient adaptive immune response 26,27 . M2 macrophages-profile predominance in bladder tumors infiltrate is related to a worse prognosis and BCG immunotherapy failure [28][29][30] . Signaling via type I IFN in tumors is crucial for the recruitment of lymphocytes in the TME and several studies seek to activate this route to potentiate the anti-tumor response 20,31,32 .
The understanding of the possible mechanisms involved in BCG immunotherapy can guide the selection of specific therapies for each patient and bring new alternatives to replace or associate with existing therapies. Thus, this study aimed to investigate the involvement of the cGAS-STING pathway, different TLRs (TLR2, TLR3, TLR4, TLR7 and TLR9) and MyD88 adaptor protein in the immunotherapy with BCG to treat MB49 syngeneic mouse bladder cancer. Even though MB49 tumor cells did not efficiently activate an immune response, BCG presence in bladder cancer could induce the appropriate response from other immune cells present in the TME. We evaluate BCG intratumoral treatment in a subcutaneous MB49 tumor model using different knockout Additionally, this study show the importance of MyD88 regulating cellular infiltrate and inflammatory profile in TME during BCG treatment. In summary, our report suggests the involvement of MyD88 and not a single TLR tested here in BCG immunotherapy against bladder cancer.

Results
BCG treatment reduces subcutaneous bladder tumor. The effect of BCG treatment on mice subcutaneous administered with MB49 bladder tumors has already been demonstrated in several studies. The different BCG strains are characterized by phenotypic and immunogenic differences with a variable of virulence levels, leading to different anti-tumor response 33 . We compared BCG Moreau and BCG Pasteur and both strains show very similar effects in tumor volume reduction ( Supplementary Fig. S1). So, we decided to use BCG Moreau because the strain is the most commonly used in Brazil and the virulence is considered intermediary 34 . Tumor regression effect is more evident when the treatment starts right after the tumor injection than after the tumor grow 35 . In addition, bladder tumors development depends on the hormone testosterone, which may explain why bladder cancer is more prevalent in men than women and MB49 tumor grows rapidly in male than in female mice 36 . After some trials, we established the heterotopic syngeneic model by the subcutaneous injection of MB49 bladder cancer cells (5 × 10 5 cells) in female wild type C57BL/6 (WT) and different knockout mice (KO). After 24 h post-tumor injection, animals were treated intratumorally with BCG (8 × 10 6 CFU) or PBS (mock) every seven days (total three doses) (Fig. 1A). After 22 days post-tumor injection we euthanized the animals, dissected the tumors and observed that BCG treatment robustly reduced tumor size compared to mock (Fig. 1B). BCG treatment shows partial or total reduction in tumor volume reinforcing the efficacy of the immunotherapy (Fig. 1C,D).
Innate immune signaling pathways are impaired in MB49 cells. The innate immune pathways are essential for the activation of adaptive immunity and to provide an efficient immune response against the tumor. However, we know that many tumor cell lines are deficient for some of these pathways. For example, it has already been demonstrated that cGAS-STING pathway is deficient in different tumor cell lines due to the inhibition of cGAS and/or STING expression by a process of DNA hypermethylation [37][38][39][40] . In order to evaluate the activation of some different innate immune signaling pathways in MB49 cells, we compared MB49 to C57BL/6 bone marrow-derived macrophages (BMDMs) that are well known for its capacity to strongly respond after stimulation with different agonists and BCG infection. Although we verified the production of inflammatory cytokines such as TNF-α ( Fig. 2A) and IL-6 ( Fig. 2B) in BMDMs activated with Pam3, LPS, CpG or BCG, the same was not detected in MB49 cells which did not respond to any of these agonists. This result suggests that TLR2, TLR4 and TLR9 receptor innate immune pathways are impaired in MB49 cells. In addition, we also evaluate the activation of pathways that recognize nucleic acids, such as cGAS-STING, TLR3 and RIG-I. Surprisingly, when MB49 was www.nature.com/scientificreports/ activated with PolyI:C via TLR3 and/or RIG-I these cells were capable to produce large amounts of CXCL-10 and to express high levels of IFN-β (Fig. 2C,D). The cGAS-STING pathway was not activated in MB49 in the presence of dsDNA, but responded to direct stimulus of cGAMP producing CXCL10 and expressing IFN-β, which suggests that cGAS is not active in this tumor cell line (Fig. 2C,D). BCG infection (MOI 10, 20 and 40) was unable to directly activate MB49 to produces any of the cytokines measured ( Fig. 2A-D S2B). Since MB49 cells were infected but not directly activating the innate immune response, we hypothesized that the tumor cells could be activating other immune cells present in TME and generating a systemic response. To evaluate this hypothesis, we proceeded with some experiments co-culturing previously infected MB49 cells with spleen cells from mice that received subcutaneous implantation of MB49 tumors and treated with BCG (treated-T) or control PBS (not treated-NT). These animals were euthanized 21 days after tumor injection to remove the spleens and prepare the total splenocytes. MB49 cells used for co-culture were previously infected (BCG MOI 40) for 24 h and washed to remove free BCG in the supernatant, followed by the addition of the splenocytes and maintained in co-culture for 24 or 48 h. MB49 cells and splenocytes separated (not in co-culture) were infected at the same time as the co-culture and maintained as controls. The amount of intracellular BCG in MB49, spleen cells or co-cultured cells were very similar even in the spleen cells from BCG treated mice ( Supplementary Fig. S2C). Regarding cytokine production, our data reinforces that MB49 does www.nature.com/scientificreports/ not respond to BCG stimulation ( Fig. 3A-C). Instead, we observed the production of TNF-α, IL-6 and IFN-γ in response to BCG infection both in the splenocytes only and in the co-cultured MB49 with splenocytes. The cytokine response detected was stronger in spleen cells from mice that were intratumorally treated with BCG. In contrast, we detected higher levels of CXCL10 and NO only in co-culture of infected MB49 and splenocytes from BGC treated mice (Fig. 3D,E). Cells in co-culture without BCG infection were not activated (data not shown). The co-culture data indicates that tumor treatment with BCG leads the spleen cells from these mice to a more responsive state. The immune response activated from BCG in the spleen cells could favor a systemic antitumor response in mice.
Successful BCG treatment against subcutaneous bladder tumor is independent of cGAS-STING pathway and TLR3 receptor. In order to evaluate the innate immune pathways related to BCG tumor treatment, we tested the MB49 syngeneic tumor model in different KO mice. The cGAS-STING pathway is activated by double-stranded DNA in the cytoplasm and is well known for its importance in antiviral response, inflammation and cancer. cGAS works in concert with STING adaptor molecule to trigger an innate immune response 41 . BCG intratumoral treatment in cGAS −/− and STING −/− mice presents similar results as C57BL/6 WT animals, discarding the importance of this pathway in BCG tumor treatment ( Fig. 4A-C). TRIF molecule is essential for TLR3 signaling pathway, a sensor for double-stranded RNA, important to mediate regulation in the pathogenesis of mycobacterial infection 42 . TLR3 −/− mice did not show difference in BCG tumor treatment www.nature.com/scientificreports/ compared to WT (Fig. 4D,E). Therefore, the effect of BCG treatment on bladder tumors is independent of TLR3 and the cGAS-STING pathway.  5D). All four TLRs KO mice responded similarly to WT regarding BCG tumor treatment (Fig. 5E). In this study, TRIF-or MyD88-dependent single TLRs did not show any relevance for BCG tumor treatment. Therefore, our strategy was to investigate the synergistic effect of a triple deficient TLR3/7/9 −/− animals eliminating the endosomal TLRs signaling or MyD88 −/− deficient mice used to abrogate completely all the receptors pathways related to MyD88. TLR3/7/9 −/− mice showed a delay in tumor regression and only a partial tumor volume reduction compared to WT mice (Fig. 6A  www.nature.com/scientificreports/ in BCG treatment does not involve specifically the activation of the inflammasome pathway ( Supplementary  Fig. S3A-D). Tumor growth experiments using interferon receptor knockout mice (IFNAR −/− ) were performed and we observe that BCG tumor treatment is independent on type I IFN signaling ( Supplementary Fig. S3E,F).
The results suggest that there is a synergistic effect of several receptors that signal via the MyD88-dependent pathway favoring the action of BCG in tumor treatment.

BCG treatment alters the immune cells infiltrate profile in TME.
In order to evaluate the importance of MyD88 to influence TME, we compared tumors from C57BL/6 WT and MyD88 −/− mice after two BCG intratumoral treatments (15 days). Tumors from euthanized mice were dissected, dissociated using collagenase IV and the cell infiltrate was analyzed by flow cytometry using specific markers to different immune cells. Intratumoral BCG treatment in WT mice regulates the cellular infiltrate in TME with a significant increase The TME was also evaluated concerning the activation status of macrophages. A significant increment in the population of inflammatory macrophages (M1) was observed in tumors treated with BCG only from WT mice but not MyD88 −/− , a change not observed for type 2 macrophages (M2), which presented a very low percentage of CD163 + cells in macrophages from all mouse groups tested (Fig. 8A,B). We also detected an upregulation of iNOS mRNA expression unlike arginase (Fig. 8C,D) after two BCG treatments in tumors from WT mice. All the analysis from TME indicates an increase in cell infiltrate and cell activation in WT mice, but not in MyD88 −/− , favorable to tumor remission.
Macrophages are essential for the immune response during BCG tumor treatment. In order to confirm the importance of macrophages we performed co-culture experiments with macrophages-depleted spleen cells from mice previously treated intratumorally with BCG. After standard intratumoral treatment with BCG for 3 weeks, C57BL/6 WT mice were euthanized for spleen cells suspension preparation as we explained previously. Cells were submitted to a depletion protocol using magnetic beads to negatively select macrophages previously labeled with anti-PE antibody. Cells remaining after depletion were quantified by flow cytometry and we determined a reduction of 94% in macrophages population among the spleen cells. We compared the co-culture of MB49 with spleen cells before and after macrophages depletion. The results show a reduction on TNF-α and IL-6 cytokines after depletion and reinforce that macrophages are essential for the inflammatory response www.nature.com/scientificreports/ in BCG treatment. Dendritic cells and T lymphocytes present among the spleen cells may be responsible for the small portion of inflammatory cytokines that continue to be produced even after depletion (Fig. 9).

MB49 infected with BCG activates a MyD88-dependent inflammatory response in macrophages.
Tumor development depends on the modulation of the immune response in the TME, considering that M1 macrophages work to control tumors while M2 macrophages prevents inflammation favoring tumor growth. To confirm if BCG-infected tumor cells depend on MyD88 in the macrophages to polarize an inflammatory profile, we performed in vitro co-culture using infected MB49 with BMDMs from C57BL/6 WT or MyD88 −/− mice. MB49 cells used for co-culture were previously infected (BCG MOI 40) for 24 h followed by the addition of WT or MyD88 −/− BMDMs and maintained in co-culture for 24 or 48 h. BMDMs from WT or MyD88 −/− (not in co-culture) were infected at the same time as the co-culture and maintained for controls. Only WT BMDMs in co-culture with BCG-infected MB49 cells were able to produce significant levels of inflammatory mediators such as TNF-α, IL-6, IL-1β, NO and iNOS mRNA expression compared to BMDMs from MyD88 −/− mice. We also detected the production of IL-10 in WT BMDMs, probably acting to counterbalance the inflammatory response. MyD88 −/− BMDMs did not show a significant cytokine response in the presence of infected MB49 cells (Fig. 10A-F). MB49 and BMDMs in co-culture without BCG infection were not properly activated (data not shown). BCG depends on the MyD88 pathway in WT macrophages infected and also in WT BMDMs in co-culture with infected MB49 to activate the inflammatory status of these cells. The mechanism of cell activation in the TME by MB49 infected with BCG needs to be better understood to improve immunotherapy strategies in order to activate an effective anti-tumor response. www.nature.com/scientificreports/

Discussion
The negative correlation between tuberculosis and the incidence of cancer in patients was first related by Pearl in 1929 43 opening new perspectives for the development of a immunotherapy using microorganisms to treat bladder cancer 44 . Different microorganisms 45 and viruses have also been tested for cancer treatment 46,47 . Recently, herpes simplex virus type 1 (HSV-1)-based oncolytic viruses engineered to express granulocyte-macrophage colony-stimulating factor (GM-CSF) was approved to treat advanced melanoma 48 . Immunotherapy efficiency depends not only on the microorganism itself but especially on the individual's immune system. Many studies try to explain the role of the immune response in BCG immunotherapy. The new concept of "Trained Immunity" brings evidence of how the innate immune response is essential for the heterologous effect of BCG treatment against cancer 49 . BCG treatment has already been experimentally tested for other types of cancer such as leukemia, melanoma and breast cancer 50,51 . In this study, we investigate the role of important host innate immune molecules in BCG treatment for tumor reduction in a subcutaneous murine bladder cancer model (MB49). BCG treatment works for tumors already grown (data not shown), but with a more prominent effect in early treatment right after implantation of subcutaneous tumors as showed before by Gan 35 and confirmed by our data. The weekly treatment with BCG in C57BL/6 WT mice leads the tumors practically to disappear after 3 treatments. Several studies have shown that intratumoral BCG injection is the most effective route for cancer immunotherapy. BCG attaches via fibronectin and α5β1 integrin to the surface of the bladder cancer cells and also to benign urothelial cells 6,52-54 . The bacteria is internalized and induces different levels of surface molecules, such as MHC-II, CD1, CD80 and ICAM-1 55,56 . The recognition and processing of BCG by tumor cells are reported as essential for subsequent induction of the antitumor response. However, many tumor cell lines are deficient or lost Figure 8. The inflammatory macrophages profile present in TME from tumors treated with BCG. Fifteen days syngeneic tumors from C57BL/6 WT or MyD88 −/− mice (same as used in Fig. 7) were evaluated concerning activation status of macrophages. Cell infiltrate was analyzed by flow cytometry using specific markers to address M1 (CD11b + F4/80 + CD80 +) (A) and M2 (CD11b + F4/80 + CD163 +) (B) macrophages profiles. The graphs percentages are relative to CD11b + F4/80+ (Macrophages) entire population. The mRNA expression in the same group of tumors was also analyzed by qPCR for the following targets: iNOS (C) and Arginase (D). All qPCR results were relative to 18S mRNA as normalizer. Untreated (mock) tumors from each group were used as control. Data represents ΔΔCT mean and standard deviation from tumors in two independent experiments. *Statistically significant compared to the respective untreated control (mock), *P < 0.05, ***P ≤ 0.001. # Statistically significant comparing BCG treated C57BL/6 WT and MyD88 −/− , ### P ≤ 0.001. www.nature.com/scientificreports/ www.nature.com/scientificreports/ the ability to activate important immune pathways after many passages in culture. The ability to internalize BCG and initiate an immune response varies according to each tumor cell line 57 . Initially, we proposed to investigate whether MB49 cell line was able to respond to BCG infection or agonists specific for different innate immune pathways such as Pam3CSK4 (TLR2), ultrapure LPS (TLR4), CpG-ODN (TLR9), dsDNA90 (cGAS-STING) or cGAMP (STING). MB49 cells were not able to activate any inflammatory response (TNF-α, IL-6, CXCL-10 or IFN-β) unlike macrophages, revealing an impaired response related to some TLRs receptors and also to cGAS-STING pathway. However, MB49 was activated by poly I:C via TLR3 producing large amounts of CXCL-10 and expressing higher levels of IFN-β. Additionally, the cGAS-STING pathway was not activated by dsDNA in MB49 cells, but responded to the direct stimulus of cGAMP suggesting that cGAS is not functional in this tumor cell line. The expression of cGAS and/or STING is recurrently suppressed by DNA hypermethylation in a variety of cancer cell lines, leading to an inability to activate the production of STING-dependent cytokines, important to suppress tumor development [37][38][39][40] . We did not detect direct MB49 activation even with high BCG doses (MOI 10, 20 and 40). MB49 cell line itself does not seem to be an effective inducer of the innate immune system in response to BCG and other agonists. We confirmed that BCG was able to infect MB49 cells; however, the different BCG MOI used did not induce significant cell death that could justify the low levels of cytokines released. Our hypothesis was that BCG did not directly stimulate the innate immune response in MB49, but the infected tumor cells could be important to activate other phagocytic cells present in TME leading to a systemic response. After euthanized the mice, we isolated spleen cells from treated and untreated animals and co-cultured with previously infected MB49 cells (MB49 + BCG). The results reinforce the inability of MB49 to produce inflammatory cytokines in response to high BCG CFU (MOI 40). However, we detected an increased inflammatory response (TNF-α, IL-6, IFN-γ, CXCL10 and NO) in the co-culture of MB49 infected with BCG and spleen cells from treated mice. Even though the intrinsic innate immune response from MB49 cells does not seem to be important, BCG immunotherapy can activate the extrinsic innate responses from the immune cells present in TME leading to a subsequent BCGspecific systemic adaptive immunity 55,56,58 .

Scientific Reports
The effect of BCG in urothelial cells or antigen-presenting cells (APCs) depend on the recognition of bacterial PAMPs by extracellular TLRs 10,49,59 or the mycobacterial DNA sensing by endosomal TLRs or STING 60,61 . The majority of the studies suggest the relevance of these immune pathways according to the cytokines measured in urine or peripheral blood from patients after BCG treatment [62][63][64] . We specifically investigated the role of different TLRs and the cGAS-STING pathway using KO mice in BCG tumor treatment using MB49 syngeneic tumor model. Mycobacterium tuberculosis and M. bovis activate a cGAS-independent STING pathway dependent on the detection of c-di-AMP (cyclic-di-adenosine monophosphate) in the cytoplasm 17 . Type I IFN production in tumors promoted by the activation of the cGAS-STING pathway 22 in response to specific agonists 20,21,65 and even fractionated radiotherapy 24,25,66 has been shown to enhance the response not only in the local tumor but also recruiting lymphocytes and promoting the systemic response in the abscopal tumor. In vitro, MB49 cells and macrophages showed significantly amount of CXCL10 and IFN-β expression in response to cGAMP and poly:IC. Even though cancer cells containing double-stranded DNA or cyclic di-nucleotides (CDNs) could not direct activate intrinsic STING, they are able to stimulate the extrinsic STING pathway present in phagocytic cells and to promote antigen cross-presentation 21 . Knowing the importance of type I IFN in the antitumor response, we decided to investigate the effect of BCG treatment in mice deficient for cGAS, STING and TLR3, molecules involved in IFN-β production. cGAS −/− , STING −/− and TLR3 −/− mice showed similar tumor reduction as C57BL/6 WT animals after three treatments with intratumoral BCG, discarding the importance of these receptors in BCG-induced tumor regression. In order to confirm if IFN is involved in BCG anti-tumor immunotherapy we performed experiments using IFNAR −/− mice and observed that BCG tumor treatment is independent on IFN signaling. Recent study reported the production of a recombinant BCG expressing a STING agonist (c-di-AMP), capable of strongly activate inflammatory macrophages (M1 profile) and induce trained immunity when compared to wild type BCG 67 . The use of agonists, especially those that activates type I IFN, in concert with BCG could potentiates the activation of STING and TLR3 68,69 pathways to improve the effect of the immunotherapy.
The majority of TLRs depends on the MyD88 adaptor molecule, except for TLR3 that recruits TRIF and TLR4 that depends on both MyD88 and TRIF. MyD88 activates MAPKs and NF-κB whereas TRIF activates IRF3 leading to type I IFN production 13 . Herein, we decided to investigate the role of four different MyD88dependent single TLRs (TLR2 −/− , TLR4 −/− , TLR7 −/− and TLR9 −/− ) in BCG tumor treatment. All four TLRs KO mice responded similar to WT after three intratumoral BCG treatment. TLR4 −/− , TLR7 −/− and TLR9 −/− untreated tumors presented significantly small volume compared to WT untreated tumors indicating that these TLRs could be involved in normal tumor development. The results did not show any relevance for the single TLRs tested here in BCG tumor treatment. Therefore, our strategy was to eliminate a set of TLRs related to endosomal pathway using triple deficient mice (TLR3/7/9 −/− ) or MyD88 −/− deficient animals. TLR3/7/9 −/− mice showed a delay in tumor regression and partial tumor volume reduction compared to WT mice. The lack of MyD88 molecule completely eliminated the effect of BCG in tumor volume reduction. MyD88 drives a wide role as an adapter molecule for TLRs and also interleukin-1 receptor (IL-1R) family 70 . The importance of MyD88 to the inflammasome complex activation can be related to the first signal after TLRs induction or downstream IL-1R signaling 71 . Our complementary data using IL-1R −/− , caspase1/11 −/− and Gasdermin-D −/− mice demonstrated that MyD88 role in BCG treatment does not involve specifically the activation of the inflammasome complex or IL-1R. All results with different innate immune pathways suggest that MyD88 is essential for BCG-tumor treatment depending on a synergistic effect involving several MyD88-dependent TLRs working simultaneously and/or a TLR-independent effect.
After the internalization of BCG by tumor or phagocytic cells or even the phagocytosis of infected tumor cells by macrophages and dendritic cells, the antigens are presented to T cells via MHC-II 55,56 . T cells recognizes BCG-specific antigens and possibly tumor-specific antigens as well. The importance of CD4+ and CD8+ T www.nature.com/scientificreports/ lymphocytes in BCG treatment was demonstrated in experiments using lymphocyte-deficient animals 58,72,73 . BCG-specific responses are important to target the cell infiltrate profile in the tumor by recruiting effector cells like CD8 + cytotoxic T-cells (CTLs), macrophages, neutrophils, natural killer (NK) cells and others 7,74 . BCG immunotherapy affects cellular infiltrate in TME and also activates the inflammatory response with a polarization of type 1 macrophages (M1) 30,75 , production of inflammatory cytokines [76][77][78][79] and nitric oxide 80 . We therefore investigated the immune cells infiltrate in WT and MyD88 −/− mice after two BCG treatments. BCG treatment in WT mice significantly increase the presence of macrophages, neutrophils, CD8+ T lymphocytes and NKT cells in TME. Our results confirm that intratumoral BCG has the potential to increase the inflammatory macrophage population (M1) and to diminish even more the anti-inflammatory profile (M2). We also detected upregulation of iNOS mRNA expression unlike arginase in tumors from WT mice. TME cell infiltrate in MyD88 −/− mice did not alter after BCG treatment. Additionally, we performed co-culture experiments with MB49 and macrophagesdepleted spleen cells and the results show a reduction on inflammatory cytokines after depletion, reinforcing that macrophages are essential for the immune response during BCG tumor treatment. Finally, we confirm the importance of MyD88 in macrophages showing that BCG induces the release of inflammatory cytokines (TNF-α, IL-6, IL-1β) and nitric oxide by WT macrophages in co-culture with infected tumor cells, but not in MyD88 −/− BMDMs. Tumor-infiltrating immune cell subpopulations dictate the immunotherapy outcome 81 and M2 profile is associated with poor prognosis of bladder cancer [28][29][30] . The production of nitric oxide after treatment with BCG and its cytotoxic effect was already demonstrated 80,82,83 . On the other hand, BCG also releases regulatory cytokine such as IL-10 allowing a counterbalance of the inflammatory response favoring the tumor control without so much damage caused by an exacerbated immune response [84][85][86] . Overall, our results suggest the essential role of MyD88 to generate an efficient BCG anti-tumor response. This research consolidates the knowledge related to the immune response induced by BCG in the bladder tumor model and may contributes to the improvement of cancer immunotherapy.

Bone marrow-derived macrophages (BMDMs). To generate BMDMs, bone marrow cells from mice
were removed from the femurs and tibias of the animals and cultured in DMEM (Gibco) with 10% FBS, 1% HEPES, penicillin G sodium (100 U/ml), streptomycin sulfate (100 μg/ml), and 20% L929 cell conditioned medium in petri dishes (1 × 10 7 cells). The cells were cultured at 37 °C in an atmosphere of 5% CO 2 . After 4 days, 10 ml of complete fresh medium was added. At day seven in culture, the cells had completely differentiated into macrophages. Macrophages were seeded in 6-well plates (5 × 10 5 cells/well), 24-well plates (2 × 10 5 cells/well) or 96-well plates (1 × 10 5 cells/well) and used for in vitro studies as previously described 90 . ), a stable metabolite of NO, was measured using Griess reagent (1% sulfanilamide and 0.1% naphthyl ethylenediamine dihydrochloride in 2.5% phosphoric acid). Briefly, 50 μl of cell culture supernatants were mixed with 50 μl of Griess reagent in 96-wells plate. Subsequently, the mixture was incubated protected from light at room temperature for 20 min and the absorbance was measured at 550 nm in a microplate reader. Fresh culture medium was used as blank. The quantity of nitrite was determined from a sodium nitrite (NaNO 2 ) standard curve as previously shown 90 .

Spleen cells culture.
Lactate dehydrogenase release assay. MB49 cells were seeded into 24-well plates (2 × 10 5 cells/well) and infected with BCG (MOI 10, 20 and 40). After 24 h of infection, supernatants were harvested for analysis of lactate dehydrogenase (LDH) release by dying cells. The cultured cells were lysed using M-PER Mammalian Protein Extraction Reagent (Thermo Fisher Scientific). Total LDH was determined by supernatant plus protein cell lysate released LDH. Dying cells were calculated by the ratio: supernatant LDH/total LDH. LDH was quantified using the CytoTox 96 LDH-release kit (Promega, Madison, WI), according to the manufacturer's instructions as previously demonstrated 91 .
Evaluation of cellular infiltrate in TME by flow cytometry. Mice were euthanized 15 days after tumor cells injection. The resulting tumor was dissected, minced and incubated in 200 U/ml collagenase IV (Gibco) shaking at 1000 rpm at 37 °C for 1 h. Tumor dissociated cells were filtered using cell strainer, centrifuged at 1200 rpm for 5 min and the pellet was suspended in DMEM medium. Tumor microenvironment (TME) cells were analyzed by flow cytometry (1 × 10 6 cells/well). Different antibodies mix (anti-mouse) were used for evaluation of myeloid cells, lymphocytes, activation status or M1/M2 markers. Briefly, cells were incubated for 20 min with anti-mouse CD16/32 to block Fc receptors (BD Bioscience, Franklin Lakes, NJ, USA) in FACS buffer (PBS, www.nature.com/scientificreports/