Mice of the C57BL/6JCit (B6) strain (hereafter – WT) were bred under conventional, non-SPF conditions at the Animal Facilities of the Central Institute for Tuberculosis (Moscow, Russia), in accordance with the guidelines from the Russian Ministry of Health # 755, US Office of Laboratory Animal Welfare (OLAW) Assurance #A5502-11. Water and food were provided ad labium. Female mice 10-12 weeks of age at the beginning of experiments were used. All experimental procedures were approved by the Institutional Animal Care and Use Committee (IACUC), protocols 2, 3, 7, 10 of March 6, 2019.

Mice with ablation of IL-6 in B cells (hereafter – B-IL-6KO) were generated by crossing IL-6^flox/flox mice (23) with CD19-Cre knock-in mice (CD19^cre/cre), which, together with mice with systemic IL-6 knockout (IL-6KO) were kindly provided by Prof. S. A. Nedospasov. To generate B-IL-6KO mice, parental IL-6^flox/flox and CD19^cre/cre strains were mated, and IL-6^flox/-CD19^cre/- heterozygous littermates were backcrossed to the parental IL-6^flox/flox strain. After PCR genotyping, females bearing the IL-6^flox/floxCD19^cre/- genotype were selected for further experimentation. Mice were genotyped by the genomic PCR using tail DNA. Primers for the CD19-Cre transgene CD19.8 5′-AATGTTGTGCTGCCATGCCTC-3′, CD19.9 5’- GTCTGAAGCATTCCACCGGAA-3’; lck2 5’- AATGTTGCTGGATAGTTTTTACTGC-3’; primers for the IL-6flox transgene F: 5’- CCCACCAAGAACGATAGTCA-3’; R1: 5’- GGTATCCTCTGTGAAGTCTCC-3’; R2: 5’-AGCACTTTATTGGGCTCTATACA-3’; primers for the IL-6KO F: 5’- ACCGCTATGAAGTTCCTCTC-3’; R 5’- CCAACCAACCCTACCTAGA-3’.

Mice were infected with ~10² CFU of virulent M. tuberculosis strain H37Rv (sub-strain Pasteur) using the inhalation Exposure System (Glas-Col, Terre Haute, IN) exactly as previously described (24). At different time points post challenge (as indicated in Figure 1 ), mice were euthanized by the thiopental (Biochemie GmbH, Vienna, Austria) overdose.

At indicated time points, spleens were homogenized in 2.0 ml of sterile saline and identical lobes of the right lungs from individual mice were digested in collagenase-DNase 1 mixture as described below, except the presence of antibiotics. 10-fold serial dilutions of 0.2 ml samples were plated on Dubos agar (Difco) and incubated at 37°C for 20–22 days before CFU were counted. Survival time was monitored daily starting month 2 post-infection.

Two upper right lung lobes from individual mice were enzymatically digested exactly as described previously (25). Two bottom right lobes were used for RNA purification. Briefly, blood vessels were washed out by perfusion with 0.02% EDTA-PBS through the right ventricle and cut vena cava, lungs removed, sliced into 1-2 mm³ pieces, and incubated at 37°C for 90 min in supplemented RPMI-1640 containing 200 U/ml collagenase and 50 U/ml DNase-I (Sigma, MO). Single-cell suspensions from 4 to 5 mice were obtained individually, washed twice in HBSS containing 2% FCS and antibiotics, and 3 x 10⁵ cells per sample were used for surface phenotyping. 1 x 10⁶ lung cells from each mouse were cultured overnight (16-17 hours total) in the absence (control) or in the presence of 10µg/ml mixture of mycobacterial antigens (cultural filtrate was kindly provided by Dr. Vadim Avdienko, CIT) and GolgiPlug (BD Bioscience) for the last 12 hours of incubation for further evaluation of intracellular cytokines. Cultural filtrate was obtained by culturing M. tuberculosis as a biofilm on the surface of the protein-free Sauton’s medium for 4 weeks, followed by precipitation of mycobacteria at 3,000g for 20 min, concentration of the protein content in the liquid phase using Amicon cell device, and bringing the final protein concentration to 1mg/ml with sterile saline.

Single-cell suspensions were obtained from individual mice and cell phenotypes were analyzed by flow cytometry as described previously (25) using the FACS Calibur machines (BD Biosciences). The gating strategy is displayed in Figure S1 . The following labeled monoclonal antibodies were used: CD4-PerCp (dilution 1:300), CD4-AF488 (dilution 1:200), or CD4-PE (dilution 1:200) (all GK1.5, BioLegend), CXCR5-Fitc (L138D7, BioLegend, dilution 1:100), CD19-AF647 (6D5, BioLegend, dilution 1:400), B220-APC (RA3-6B2, BioLegend, dilution 1:400), F4/80-AF-488 (BM8, BioLegend, dilution 1:200), Ly6G-PE (1A8, BD Biosciences, dilution 1:200), CD44-FITC (IM7, BioLegend, dilution 1:150), CD62L-PE (MEL-14, BioLegend, dilution 1:200), CD11b-biotine (M1/70, dilution 1:100) with subsequent SAv-PerCP (BioLegend, dilution 1:500) staining. For intracellular staining of the lung cells, BD Fixation/Permeablization Kit (BD Biosciences) was used due to the recommendations. Briefly, after 16-17 hours of in vitro cultivation in the presence of mycobacterial antigens, cells were harvested, stained first for surface antigens, then fixed with Cytofix/cytoperm buffer for 30 minutes and then stained for intracellular cytokines with IFN-γ-PE (BD Biosciences dilution 1:150) and IL-17-PerCp (TC11-18H10.1, BioLegend, dilution 1:150) antibodies. Results are presented as mean ± SD.

Left lungs were frozen at the regimen of –20°C to –60°C temperature gradient in the electronic Cryotome (ThermoShandon, UK), and serial 10μm-thick sections were made across the widest area. For visualization of lung pathology, sections were fixed with ice-cold acetone and stained with hematoxylin-eosin. For IHC, lung cryosections were fixed with 1% PFA, blocked with 10% donkey serum, and incubated with rabbit anti-mouse primary anti-iNOS (NovusBio) pAB, rat anti-mouse primary B220 (IgG2a, clone RA3-6B2, eBioscience), rabbit anti-mouse primary anti-ARG pAB (NovusBio), rabbit anti-mouse primary anti-CD3 (clone SP7, ThermoScientific) for 1 h at room temperature. Afterwards, preparations were incubated with secondary donkey anti-rabbit IgG-Cy3, donkey anti-rat IgG-AF488 or anti-rabbit IgG-AF488 pAB for 1 h at the room temperature. Slides were preserved using ProLong Gold anti-fade reagent with DAPI (Invitrogen-Life Technologies) before visualization using the Zeiss Axioskop40 microscope and AxioCam MRc5 AxioVisio 4 camera (Carl Zeiss, Berlin, Germany).

Identical lobes of the right lungs from 4-5 individual mice per group per time point were isolated and immediately put into RNA lysis buffer. Total RNA was isolated using the commercial SV Total RNA Isolation System (Promega). Reverse transcription of mRNA was performed using oligo(dT) primers, dNTP mix, M-MLV RT and RNasin^® (Promega). Quantitative real-time RT-PCR (qPCR) of cDNA was performed using qPCRmix-HS SYBR (Evrogen) and cfx-96 Real-Time PCR Detection System (BioRad). The following primers were used:

Relative expression levels were calculated by normalizing levels for genes of interest to that of hprt using the 2^–ΔΔCt method.

Statistical analysis was performed using GraphPadPrism7 software. Representative data from one of two identical experiments are displayed, except survival curves for which combined results from two independent experiments were summarized. The log-rank test for survival and One- or Two-way Anova with Tukey post-test for multiple comparison for other experiments were used. P < 0.05 was considered statistically significant.

In early studies, it was shown that in mice with genetically disrupted il6 gene in all cells (systemic IL-6KO) i. v. challenge with virulent M. tuberculosis resulted in a more rapid mortality compared to the wild type mice (15). Similar work utilizing the aerosol infection route reported an increase in the lung mycobacterial counts in IL-6KO mice; however, mortality was not assessed in this study (16). Thus, we evaluated the influence of systemic and B-cell-specific IL-6 deficiency on the most prominent features of chronic mycobacterial infections, the life span of mice, and mycobacterial multiplication in their organs. As shown in Figure 1A , both systemic and B-cell-restricted IL-6 elimination significantly decreased survival time of the infected animals compared to controls, although selective IL-6 defect in B-cells resulted in a less severe phenotype. Unexpectedly, no influence of KO mutations on lung ( Figure 1B ) and spleen ( Figure 1C ) CFU counts was observed, suggesting that a more rapid mortality in IL-6-deficient animals was not due to higher mycobacterial loads in their organs.

Production of IFN-γ by CD4⁺ T cells is considered critical for protective immunity against M. tuberculosis (2). Since there is evidence that in IL-6KO mice general production of IFN-γ is decreased (15, 16), we evaluated how the B-cell-produced IL-6 influences IFN-γ production in the lungs. After culturing overnight in vitro in the presence of mycobacterial antigens, IFN-γ response in CD4⁺ T lung cells from infected WT and B-IL-6KO mice was assessed by flow cytometry (intracellular staining). Additionally, we evaluated the overall expression level of the ifng gene in the lung tissue by qPCR. The proportion of CD4⁺ T cells infiltrating lungs increased significantly during the course of infection in mice of both strains, and their percentage did not differ between strains at weeks 3 and 7 post-infection ( Figure 2A ). However, in B-IL-6KO mice, at week 3 post-infection the proportion of IFN-γ-positive lung CD4⁺ cells was significantly lower than in control animals ( Figure 2B ). In addition, at this time point the expression level of ifng gene in the lung was also decreased in B-IL-6KO mice ( Figure 2C ). These differences disappeared by the week 7 post challenge ( Figure 2 ).

IL-6 is important for the development and maintenance of a balanced Th17 immune response (26), whose role in TB is ambiguous, although several studies suggested the importance of Th17 population for the effective immune response against TB (27, 28). Given that in mice bearing either genetically disrupted il6, or lacking the gp130 receptor in T cells, development of IL-17-producing Th17 cells is decreased (29), we measured whether IL-6 deficiency in B-cells could affect the mycobacteria-specific Th17 population. Lung cells from mice infected for 3 and 7 weeks were cultured overnight in the presence of mycobacterial antigens and IL-17 production by CD4⁺ T cells was assessed by flow cytometry. Even though the overall frequency of mycobacteria-specific CD4⁺IL-17⁺ cells was lower than 1%, the CD4⁺IL-17⁺ population was significantly smaller in B-IL-6KO mice at both time points ( Figure 2D ). However, the expression level of the il17 gene in the lung was equal in the two mouse strains at week 3 or even slightly increased in B-IL-6KO mice at week 7 post infection ( Figure 2E ).

IL-6 is crucial for differentiation of follicular CXCR5⁺ Tfh, which are required for the “correct” co-localization of T cells and macrophages in the manner optimal for TB control (22). Given that during TB course B cells migrate to the lung and form BCFs that possess all prominent features of BCFs of secondary lymphoid organs, including the presence of CD4⁺CXCR5⁺ Tfh cells (22, 30–32), we assumed that the absence of IL-6 in B cells may decrease the content of these regulatory T-cells during infection. Indeed, both in lungs and spleens of B-IL-6KO infected mice the proportion of CXCR5⁺ CD4⁺ Tfh-cells was significantly lower than in their WT counterparts ( Figures 2F, G ). Moreover, even before infection, both total and B-cell-specific IL-6 deficiency resulted in a significantly diminished size of this important T-cell population in spleen ( Figure 2H , such an experiment is technically infeasible with non-infected lung tissue). In addition, at the early stage of infection lung cells of B-IL-6KO mice expressed significantly less il21 mRNA ( Figure 2I ). This gene encodes IL-21, the cytokine intensively produced by Tfh cells (33, 34).

Since IL-6 may originate from many different cell sources, we next assessed what is the B cell input into the total IL-6 production in the lungs of infected mice. At week 3 post challenge, B cells comprise not more than 10 percent of all lung cells ( Figure 3A ). However, in B-IL-6KO mice the total amount of il6 mRNA in the whole lung tissue appeared to be 10-fold lower compared to WT mice, suggesting that B-cells are one of the major sources of IL-6 in tuberculous lungs early after challenge ( Figure 3B ). This ratio in the il6 gene expression was retained at least up to week 7 post infection, although the overall expression level significantly decreased by this time point ( Figure 3B ), despite substantial growth of the lung B-cell content, which also partially depended upon B-cell-provided IL-6 ( Figure 3A ). We limited our IL-6 measurements by the total mRNA assessment in freshly isolated lung tissue since many different types of lung cells produce this cytokine and require different culture conditions for optimal responses. Thus, lung T-cells from mice infected with M. tuberculosis or Pseudomonas aeruginosa readily secrete IL-6, TNF-α and IFN-γ after stimulation in vitro with specific antigens in the presence of splenic APC, but the response is markedly inhibited by lung macrophage-produced prostaglandin E2 (our unpublished observations). However, we expect to resolve these limitations by rigorous cell sorting and measure IL-6 production by different cells at the protein level in our further studies.

An important aspect of IFN-γ- and IL-6-mediated signaling is activation of two transcription factors, Stat-1 and Stat3, respectively (35, 36). We have measured the expression levels of corresponding genes and found that at week 3 of infection a decreased level of the ifng and il6 expression is indeed accompanied by the lower expression of stat1 and stat3 in the lungs of B-IL-6KO compare to the WT animals ( Figure 3C ).

To find out whether the conditioned IL-6 knockout influences TB pathogenesis, we assessed lung histopathology in B-IL-6KO and WT mice at weeks 3 and 7 post infection. There was little difference between B-IL-6KO and WT mice regarding the dynamics of TB inflammation. As shown in Figure S2 , at week 3 post infection TB foci displayed no regular structure, whilst at week 7 we observed the development of circled granulomata with macrophage-occupied centers surrounded by lymphocyte cuffs in mice of both strains. However, synoptic pictures reflecting ratios between the areas occupied by diffuse pneumonia, condensed TB foci, and unaffected breathing tissue differed between mice of the two strains. As shown in Figures 4A, C and Figure S2 , in the WT mice zones of diffuse inflammation occupied a larger proportion of the lung tissue compare to the B-IL-6KO mice at week 3 post challenge. As this was an unexpected observation, we decided to evaluate how lung-infiltrating B- and T -lymphocytes interact within the inflamed tissue. To this end, we performed the IHC staining of B- and T-cells residing within the zones of condensed and diffuse inflammation in infected lungs of the WT and B-IL-6KO mice. As shown in Figures 4B, D , within the lymphocyte cuffs surrounding blood vessels and bronchi B- and T-lymphocytes were located in tight vicinity in mice of both strains, making possible their physical and functional interactions, e.g., antigen presentation by B-cells. However, within the zones of diffuse pneumonia such co-localization (quite often – geminates) was substantially more frequent in the WT mice ( Figure 4E , left). Furthermore, B-cells were practically lacking within the condensed inflammatory foci in lung parenchyma of B-IL-6KO animals ( Figure 4E , right). A substantial increase of the area of T- and B-cell contacts in the WT mice possibly provide additional opportunities for the development of acquired immunity in the lung.

Since we obtained a histological evidence that the B-cell-produced IL-6 is involved in early lung cells interaction after TB challenge, we decided to check a possible influence of selective IL-6 deficiency on the expression of the gene for another member of the gp130 cytokine family, IL-11, highly expressed in TB susceptible mice (37). This was prompted by our earlier in vivo observation that IL-11 regulates both self and IL-6 production in infected lungs (38). However, no such dual effect was observed for B-cell-specific IL-6: the level of il11 gene expression was similar in mice of the two strains ( Figure S3 ).

Infected B-IL-6KO and WT mice did not differ by the sizes of neutrophil ( Figure 5A ) and macrophage ( Figure 5B ) populations residing in their lungs. Since there is a growing body of evidence indicating that IL-6 participates in macrophage polarization toward the anti-inflammatory M2 phenotype (39–42), which is less effective for TB protection (43, 44), we tested a few parameters distinguishing M1 and M2 macrophages. Between weeks 3 and 7 of infection, in the lungs of B-IL-6KO and WT mice we observed a marked (~8-fold) and equal increase in expression of the inos gene ( Figure 5С ) encoding inducible NO synthase (iNOS), the enzyme playing a pivotal role in the mycobacterial killing in mice (45) and a key marker of the M1 macrophage subset. This was supported by IHC staining, visualizing high amounts of iNOS in the central zones of TB lung granulomata in mice of both strains ( Figure 5D ), whilst at week 3 post challenge iNOS-positive cells were almost undetectable in the lung tissue and absent in intact lungs ( Figure S4 ). Surprisingly, whilst in B-IL-6KO lungs the level of arg1 gene expression remained stable, it was significantly down-regulated in the WT mice between weeks 3 and 7 ( Figure 5C ), despite a higher il6 gene expression level in their lungs at both time points. A more prominent production of arginase instead of inducible NO synthase is often regarded as a property of M2 macrophages. We observed a stable level of expression of the corresponding arg gene in B-IL-6KO mice, whilst in WT controls its expression sharply dropped and became significantly lower at week 7 post challenge ( Figure 5E ). A suggestively higher expression of the fizz gene involved in the M2 polarization pathway was also observed in B-IL-6KO mice at week 7 post infection, while the expression of ym1 was equal in the two strains ( Figure 5E ). We also tried to assess arginase expression in lungs of mice of the two strains by IHC staining. As shown in Figure S5 , both in WT and IL-6KO mice arginase-positive cells were present in the inflamed lung tissue in similar quantities. Thus, there was only weak evidence of the involvement of B-cell-derived IL-6 in M2 polarization, however, irrespective to the presence or absence of IL-6 produced by B-cells, M1 pathway was much more prominent during the early stage of TB infection.