Th17 cell plasticity towards a T-bet-dependent Th1 phenotype is required for bacterial control in Staphylococcus aureus infection

Staphylococcus aureus is frequently detected in patients with sepsis and thus represents a major health burden worldwide. CD4+ T helper cells are involved in the immune response to S. aureus by supporting antibody production and phagocytosis. In particular, Th1 and Th17 cells secreting IFN-γ and IL-17A, are involved in the control of systemic S. aureus infections in humans and mice. To investigate the role of T cells in severe S. aureus infections, we established a mouse sepsis model in which the kidney was identified to be the organ with the highest bacterial load and abundance of Th17 cells. In this model, IL-17A but not IFN-γ was required for bacterial control. Using Il17aCre × R26YFP mice we could show that Th17 fate cells produce Th17 and Th1 cytokines, indicating a high degree of Th17 cell plasticity. Single cell RNA-sequencing of renal Th17 fate cells uncovered their heterogeneity and identified a cluster with a Th1 expression profile within the Th17 cell population, which was absent in mice with T-bet/Tbx21-deficiency in Th17 cells (Il17aCre x R26eYFP x Tbx21-flox). Blocking Th17 to Th1 transdifferentiation in Th17 fate cells in these mice resulted in increased S. aureus tissue loads. In summary, we highlight the impact of Th17 cells in controlling systemic S. aureus infections and show that T-bet expression by Th17 cells is required for bacterial clearance. While targeting the Th17 cell immune response is an important therapeutic option in autoimmunity, silencing Th17 cells might have detrimental effects in bacterial infections.

Staphylococcus aureus is frequently detected in patients with sepsis and thus represents a major health burden worldwide. CD4 + T helper cells are involved in the immune response to S. aureus by supporting antibody production and phagocytosis. In particular, Th1 and Th17 cells secreting IFN-γ and IL-17A, are involved in the control of systemic S. aureus infections in humans and mice.
To investigate the role of T cells in severe S. aureus infections, we established a mouse sepsis model in which the kidney was identified to be the organ with the highest bacterial load and abundance of Th17 cells. In this model, IL-17A but not IFN-γ was required for bacterial control. Using Il17aCre × R26YFP mice we could show that Th17 fate cells produce Th17 and Th1 cytokines, indicating a high degree of Th17 cell plasticity. Single cell RNA-sequencing of renal Th17 fate cells uncovered their heterogeneity and identified a cluster with a Th1 expression profile within the Th17 cell population, which was absent in mice with T-bet/Tbx21-deficiency in Th17 cells (Il17aCre x R26eYFP x Tbx21-flox). Blocking Th17 to Th1 transdifferentiation in Th17 fate cells in these mice resulted in increased S. aureus tissue loads.
In summary, we highlight the impact of Th17 cells in controlling systemic S. aureus infections and show that T-bet expression by Th17 cells is required for bacterial clearance. While targeting the Th17 cell immune response is an important therapeutic option in autoimmunity, silencing Th17 cells might have detrimental effects in bacterial infections.

Introduction
Blood stream infections (BSI) and sepsis are still a major health burden, which despite aggressive antimicrobial therapies account for approximately 5 million deaths every year worldwide [1,2]. While a broad range of pathogens are able to cause BSI, Gram-positive pathogen Staphylococcus aureus is of particular clinical importance and related to severe courses of bacterial sepsis [3]. S. aureus is a human pathobiont that, while colonizing the nose of healthy individuals, can cause invasive disease, systemic inflammation and death in vulnerable individuals [4]. Unfortunately, the survival rate of S. aureus sepsis has not improved in recent decades [5]. Insights into adaptive immune responses hold promise to provide novel clues for understanding the success of S. aureus as an invasive pathogen. The immune reaction to invasive infection has been categorized in two states, excessive inflammation and subsequent immune suppression. The first response to an invading pathogen is characterized by a proinflammatory innate immune response that includes activation of the coagulation system, complement system and activation of neutrophils and platelets [6]. S. aureus infection triggers the production of antimicrobial peptides such as hBD-3 or RNase7 [7] and activation of granulopoiesis. In this context, activated antigen-presenting cells (APCs) produce IL-23 which is essential for polarization and maintenance of Th17 cells [8]. The notion that bacterial virulence factors involved in the resistance of S. aureus to the human immune system trigger the activation of different T cell subsets [9,10], indicates that the T cell immune response might play a functional role in bacterial clearance. However, it is important to preserve efficient anti-bacterial subsets while reducing subsets that promote bacterial survival in order to control S. aureus infections.
In vitro differentiation assays revealed that presence of S. aureus antigens induce the development of IL-17-expressing CD4 + T cells (Th17) [11][12][13], and in the serum of septic shock patients IL-17A levels were elevated [14], indicating that bacterial infections might trigger Th17 immune responses. However, the impact of IL-17A in controlling S. aureus infection in vivo is still controversial [9]. For example, in S. aureus infection models mice with deficient IL-17 signaling had increased mortality [15] and higher bacterial burden [16,17], while another study showed protection from tissue-injury with systemic S. aureus infection in IL-17A-deficient mice [18]. IL-17A and F have been reported to be dispensable in mice for reducing the bacterial burden in systemic infection, while mucocutaneus infection is controlled by the IL-17 immune response [8]. These data indicate differences between local tissue-specific and systemic immune responses to S. aureus infection. Interestingly, we have recently identified a pathogen-induced generation of tissue-resident memory cells of the Th17 cell subset (Trm17) in the kidney [19]. These Trm17 cells reside in the kidney after the infection has been cleared and upon reactivation by inflammatory triggers can aggravate immune-mediated kidney disease.
Several studies from the past decade indicate that T cell subsets cannot be seen as terminally differentiated cells, but that T cell polarization could be defined a temporary condition in a continuum of manifold states. This might hold true particularly for IL-17-producing Th17 cells that display significant plasticity in various models of autoimmune diseases [20][21][22][23] and interaction with commensal bacteria in the intestine [24]. However, little is known about plasticity in host defense against pathogens in systemic infection [25].
In this study, a mouse model of acute S. aureus infection facilitated the investigation of T cell immune responses to bacterial inflammation in the affected tissue, as S. aureus triggers a prominent Th17 immune response particularly in the kidney compared to other organs. We show that plasticity of renal Th17 cells generates a highly effective Th17 cell subset characterized by a Th1 cell expression profile that drives bacterial clearance in the kidney. These results show that T cell plasticity is an important factor in the host reaction to invading pathogens and highlight a T cell subset with high antibacterial capacities.

S. aureus sepsis induces a prominent Th17 immune response in the kidney
We have recently described the development of tissue-resident memory T cells with a Th17 polarization state after S. aureus infection that contribute to immune-mediated glomerular disease. Since the T cell response to S. aureus in acute infection remains unclear, we investigated the immune response to S. aureus bloodstream infection over a period of 10 days (Fig 1A). First, we measured the bacterial burden of kidneys, spleen and liver at days 0, 3, 6 and 10 after infection with S. aureus ( Fig 1B) and confirmed previous data that the kidney is the organ with the highest bacterial burden after infection [19]. After identifying the kidney as an important immunological site of acute S. aureus infection, histological investigation revealed S. aureus accumulations to be located in the tubulointerstitial area of the kidneys (Fig 1C). Renal pathology is characterized by abscess formation, a hallmark of S. aureus infection ( Fig 1D). Quantification of abscess lesions per kidney section showed an increase in the course of infection (S1A and S1B Fig). This finding was supported by CD3 and GR1 immunohistochemistry of kidney sections for the detection of T cells and neutrophils, respectively (S1C and S1E Fig). These cells accumulated in the course of infection in the tubulointerstitial space as well as in the glomeruli (S1D and S1F Fig). Since glomeruli were affected, we looked for the integrity of the glomerular basement membrane that serves as the urine blood barrier. Importantly, we did not find any immune depositions at the glomerular basement membrane (S1G Fig), arguing against the manifestation of membranoproliferative glomerulonephritis in this model. To characterize the T cell immune response during S. aureus infection, we analyzed the cytokine production of CD4 + T cells from the kidney, small intestine, liver and spleen by intracellular cytokine staining (Fig 1E and 1F). While IL-17A-production was detected only at very low levels in healthy mice, S. aureus infection resulted in a continuous increase of the IL-17A-producing population over time. This accumulation of IL-17A-producing T cells was most pronounced in the kidney compared to liver, small intestine and spleen. Interestingly, a population of IL-17A and IFN-γ co-producing CD4 + T cells emerged in the kidney after infection.
We also investigated additional cell types that could contribute as a local source for IL-17A.

PLOS PATHOGENS
T-bet-dependent Th17 cell plasticity controls S. aureus infection

IL-17A-deficient mice show higher bacterial burden
In this model of S. aureus infection, IL-17A and IFN-γ production by CD4 + T cells is very prominent in the kidney. Therefore, we aimed at investigating the influence of IL-17A and IFN-γ on bacterial clearance using gene deficient mice. Interestingly, IL-17A-deficient animals showed increased staining for Gram-positive bacteria (Fig 2A) and had an elevated bacterial burden (Fig 2B) in the kidney compared to wildtype mice. Flow cytometry confirmed the absence of IL-17A + CD4 + T cells in IL-17A-deficient mice (Fig 2C and 2D).

PLOS PATHOGENS
T-bet-dependent Th17 cell plasticity controls S. aureus infection Analysis of renal CD11b + cells exhibited a numerical increase in cell numbers in S. aureus infection (Figs 2E and S2G) but no difference between wildtype or IL-17A-deficient mice. To further characterize CD11b + mononuclear phagocytes, we tested a panel of cell surface markers (Figs 2F and S2H) and identified reduction in MHCII and CD86 expression in IL-17Adeficient animals in the kidney (S2I Fig). CD11b + cells in the liver of infected animals did not display reduction in MHCII and CD86 in IL-17A-deficient mice (S2J-S2N Fig).
In contrast, IFN-γ-deficiency did not influence renal bacteria abundance in histology ( Fig  2G) or culture (Fig 2H). Flow cytometry confirmed the absence of IFN-γ + CD4 + T cells in the corresponding gene-deficient mice (Fig 2I and 2J) and showed a higher percentage of IL-17A production by CD4 + T cells. Based on these data, we concluded that IL-17A contributes to the clearance of bacteria in the kidney while IFN-γ may be dispensable. Furthermore, the Th17 cell amount in IFN-γ-deficient mice is elevated in contrast to WT mice and might be upregulated to combat infection via IL-17A because IFN-γ is missing.

Renal Th17 cells show high plasticity towards a Th1-like phenotype in S. aureus sepsis
In a time course analysis of S. aureus infection, we identified a population of IL-17A and IFNγ co-producing CD4 + T cells that was not detected in a model of experimental glomerulonephritis with a strong Th17 cell response [26]. Next, we aimed at better characterizing these coproducing cells. To answer the question whether IFN-γ -producing cells originate from Th17 cells, we made use of IL-17A fate reporter mice (Il17aCre x R26eYFP) in which T cells that had produced IL-17A are constitutively marked by YFP expression [27] and investigated cytokine production of renal YFP + CD4 + T cells by flow cytometry. We detected an increasing number of YFP + cells in the course of infection and in particular IL-17A-producing and IFN-γ-producing cells expanded over time (Fig 3A-3D), indicating that Th17 cells become more flexible during infection and show plasticity towards a Th1 cell phenotype. Of note, the analysis of cytokine production of YFPcells confirmed high IFN-γ production by these cells (Fig 3E and  3F). A combination of indirect immunofluorescence and confocal microscopy revealed the tubulointerstitial localization of YFP + cells in the kidney (Fig 3G).
To investigate the presence of immunoregulatory T cells in S. aureus sepsis, we used mice with a combined reporter system for Th17 cell fate (Il17aCre x R26eYFP) and acute cytokine expression (IL10 eGFP x Il17a Katushka x Foxp3 mRFP ) (termed fate+) to uncover a possible plasticity of renal Th17 cells into type 1 regulatory T cells (Tr1) defined by IL-10 expression in the absence of Foxp3 [28]. In the kidney, we measured a small increase of Tr1 cells from the Th17 fate (Tr1exTh17 cells) from day 0 to day 6 after infection (S3A Fig). In the small intestine, the numbers of Tr1exTh17 cells were found at lower levels compared to the kidney and their numbers were stable in the course after infection. This discovery highlights the fact that Th17 cells can acquire diverse phenotypes over time.

Comprehensive gene expression analysis of Th17 cells in the kidney of Il17aCre x R26eYFP mice
To investigate the heterogeneity of renal Th17 cells in more detail, we applied single-cell RNAsequencing (scRNA-seq) to FACS-sorted CD4 + YFP + T cells from S. aureus-infected Il17aCre x R26eYFP mice. Analysis of scRNA-seq data showed eight clusters within the population of Th17 fate cells in the kidney (Fig 4A), which were annotated according to their gene expression profile (Fig 4B). We identified one cluster (cluster 3) with the common profile of Th17 cells characterized by the expression of Il17a, Il17f, Rorc and Rora. Clusters 1 and 4 represented cells in an intermediate cell state, which highly expressed genes linked to bacterial response

PLOS PATHOGENS
T-bet-dependent Th17 cell plasticity controls S. aureus infection and inflammation such as Plac8 (Placenta specific gene 8), Icos (Inducible T-cell co-stimulator) and Tnfaip3 (Tumor necrosis factor alpha-induced protein 3). In cluster 8, we identified genes associated with memory. Interestingly, cells of cluster 5 expressed genes associated with Th1 cells such as Ifng, whereas cluster 2 contained cells with high Ifng expression in combination with Cxcr3. In line with the flow cytometric data of Th17 cells, scRNA-seq revealed IL17-A and IFN-γ co-expressing cells (Fig 4C and 4D). While cluster 3 does not include cells expression only IFN-γ, clusters 2 and 5 show reduced IL-17A positive cells.
To analyze the trajectories of the Th17 cells, we defined cluster 3 with the highest Il17a-expression as starting point of differentiation and obtained four different curves by slingshot analysis (S3B Fig). The trajectory from Th17 to Th1-like cells displayed an increasing Th1 score and a decreasing Th17 score which was among other genes based on the expression of the transcription factors Rorc and Tbx21, respectively (Fig 4E-4G). Looking at the trajectory of Th17 cells towards Th1-like cells, the expression of T-bet increased with pseudotime ( Fig 4H). Together, these data indicate the plasticity of bona fide Th17 cells to Th1 phenotypes in S. aureus infection.

T-bet-deficiency affects Th17 to Th1 transdifferentiation in infected kidneys
Tbx21/T-bet was upregulated in Th17 fate cells that transdifferentiated into Th1 phenotypes. To evaluate the impact of T-bet on Th17 cell transdifferentiation, we made use of S. aureus- Integrating the scRNA-seq data from Tbx21-wildtype and Tbx21-deficient Th17 cells revealed nine different clusters that were annotated according to their gene expression (Fig 5A). One cluster of Th1-like cells was almost absent in T-bet-deficient Th17 cells from Il17aCre x R26eYFP x Tbx21-flox mice (Fig 5B). Gene expression analyses revealed Th17 associated genes in clusters 3 (Th17_c1) and 4 (Th17_c2) and Th1 associated genes in cluster 5 (Th1-like) and 9 (cytotoxic Th1-like) (Fig 5C and 5D). In addition, Th17 fate cells from Il17aCre x R26eYFP mice showed a higher Th1 score and Th17 cells from Il17aCre x R26eYFP x Tbx21-flox infected mice showed a higher Th17 score (Fig 5E).

PLOS PATHOGENS
T-bet-dependent Th17 cell plasticity controls S. aureus infection Th17 cells with T-bet-deficiency might show an arrest in their transdifferentiation to Th1-like cells. Trajectory analyses using the slingshot algorithm exhibited a trajectory line from cluster 3 (high Th17 gene profile) to cluster 5 (high Th1 gene profile), which is absent in Tbx21-deficient cells (S3C Fig). IFNg but also Nkg7 and Ctsw are among the top genes most exclusively expressed by these cells (Fig 5F). Taken together, this data supports the notion that Tbx21/T-bet is essential for transdifferentiation of Th17 cells in S. aureus infection to Th1 like cells.

Th1-exTh17 cells drive bacterial clearance in the kidney
After identifying a population of T-bet-dependent Th17 fate cells by scRNA-seq, we aimed at understanding the role of these cells in S. aureus infection and in particular their role in the control of the pathogen. After infection with S. aureus, an extended level of bacteria and abscess formation in the kidney was found in mice with T-bet-deficiency in Th17 fate cells at day 10 after infection compared to T-bet/Tbx21-wildtype mice (Fig 6A-6C).
To further evaluate the interleukin expression of Th17 fate cells, we performed flow cytometry of T cells from the kidney after infection. This analysis verified similar IL-17A production of Th17 fate cells from Il17aCre x R26eYFP mice and Il17aCre x R26eYFP x Tbx21-flox mice. In contrast, levels of IFN-γ producing cells in Il17aCre x R26eYFP x Tbx21-flox mice were reduced compared to wildtype mice. This affected both IL-17A + /IFN-γ + and IL-17A neg /IFNγ + cells (Fig 6D and 6E). Of note, IFN-γ production by YFP neg CD4 + T cells was not altered (S4A and S4B Fig).
To investigate if Th1exTh17 cells exert their antibacterial properties by IFN-γ, we neutralized this cytokine using a monoclonal antibody. Anti-IFN-γ treatment at day 4 and 7 during S. aureus infection (Fig 6F) of Il17aCre x R26eYFP mice and Il17aCre x R26eYFP x Tbx21-flox mice did not reverse the phenotype of higher bacterial load in mice with T-bet-deficient Th17 fate cells (Fig 6G). Flow cytometry of YFP pos CD4 + T cells displays reduction of IFN-γ production in T-bet deficient mice (S4C and S4D Fig) and support our observations in IFN-γ deficient mice (see Fig 2G and 2H). While this data indicates that IFN-γ is dispensable for the elimination of renal S. aureus, it underscores the impact of a T-bet-dependent anti-bacterial Th17 cell subset.
In addition to CD4 + T cells, we had identified also γδ T cells to contribute to renal IL-17A in S. aureus infection (S2A-S2C and S4E and S4F Figs). Since we could not exclude that T-bet positive γδ T cells are responsible for bacterial clearance, we performed adoptive CD4 + T cell transfer from either Il17aCre x R26eYFP mice or Il17aCre x R26eYFP x Tbx21-flox mice into RAG -/mice ( Fig 6H). Our analysis shows higher bacterial load in the kidney of mice transferred with Tbx21-flox cells. In contrast, there was no difference between the two groups in bacteria isolated from liver and spleen (Fig 6I). Note that the kidney yielded the highest bacterial numbers also in the T cell transfer model.
While the expression of IL-17A was higher in T-bet/Tbx21-flox CD4 + cells and IFN-γ was increased in T-bet-competent CD4 + cells, the co-expression of IL-17A and IFN-γ was not different between these groups (Fig 6J and 6K). Characterization of renal CD11b + cells did not show major differences (S4G and S4H Fig). Taken together, our data support the notion that T-bet expressing cells of the Th17 fate are highly efficient in supporting the elimination of S. aureus in the kidney.

Discussion
The adaptive immune response to bacterial challenges remains incompletely understood. In this study, we identify a key role of Th17 cells during bacterial clearance in the kidney. In

PLOS PATHOGENS
T-bet-dependent Th17 cell plasticity controls S. aureus infection

PLOS PATHOGENS
T-bet-dependent Th17 cell plasticity controls S. aureus infection particular, we found that T-bet/Tbx21-expression in Th17 cells gives rise to a highly effective anti-bacterial T cell subset.
Comparative analysis of different tissues revealed a robust T cell response particularly in the kidney of mice after S. aureus infection. This is in line with a recent publication, in which we identified the highest bacterial titers after systemic S. aureus infection in the kidney [19]. In the present study, we investigated the early adaptive immune response to S. aureus infection and found IL-17A-producing CD4 + T cells to be highly abundant in the kidney. It is well established that in vitro co-culture of S. aureus antigens with immune cells drives a robust Th17 cell response [11,29]. In this study, we provide evidence that Th17 cells indeed play a major role in controlling S. aureus infection in living animals. This is in line with a report indicating that T cells contribute to bacterial clearance in the kidney [30]. Furthermore, neutralizing IL-17A in local S. aureus infection resulted in larger abscesses in the skin [31]. Our data extents these results and indicates that Th17 cells contribute to bacterial clearance in the kidney.
The concept of terminally differentiated T cell subsets has been challenged in the past decade by studies showing transdifferentiation of T cells, including the plasticity of Th17 cells into Th1 [32,33] or Tr1 [28] phenotypes. A high degree of Th17 cell plasticity has been demonstrated in several mouse models of autoimmune diseases such as experimental autoimmune encephalomyelitis, a model of multiple sclerosis [27], and in RB high colitis, a model of inflammatory bowel disease [34] using adoptive cell transfer of purified T cell subsets or fluorescent fate reporter mice. In contrast, in the murine kidney, we have seen very limited Th17 cell plasticity in mouse models of immune-mediated renal disease [26]. The findings of varying degrees of Th17 cells plasticity in different tissues and under different inflammatory triggers suggest that T cell plasticity strongly depends on the local micro-environment. In this study, we observed a high level of Th17 cell plasticity in the kidney after S. aureus infection based on intracellular cytokine staining. By employing scRNA-seq of Th17 cells, we obtained the full picture of Th17 heterogeneity in the kidney in bacterial infection. Using RNA-slingshot analysis, we identified a trajectory from IL-17A expressing cells to Th1-like cells with expression of Ifng and Cxcr3. The potential of Th17 cells to develop into Th1 phenotypes is in line with previous publications from animal models of autoimmune diseases. Specifically, Brucklacher-Waldert et al. have investigated Tbx21-deletion in Th17 cells [35]. In line with this publication, T-bet-deficiency in ex-Th17 cells hinders the development of IFN-γ expressing cells, but IL-17A and IFN-γ co-producing cells still emerge. However, while Helicobacter hepaticus infection in the intestine is not altered by T-bet-deficiency in ex-Th17, our data in S. aureus infection shows T-bet-dependent anti-bacterial function of Th1 phenotypes that derive from Th17 cells.
The function of T-bet in T cells might be depending on the local micro-environment in the tissue. In models for immune-mediated diseases, Th17 cells that produce T-bet-dependent IFN-γ have been shown to be proinflammatory in the central nervous system [36] but T-bet expression in Th17 cells can also modify intestinal inflammation by regulating IL-23 receptor

PLOS PATHOGENS
T-bet-dependent Th17 cell plasticity controls S. aureus infection [37]. Our results add to this knowledge by showing that proinflammatory cells can have protective properties in bacterial infection in the kidney, while clearance of extrarenal bacteria does not seem to be dependent on T-bet expression by Th17 cells. Our scRNA data indicate that cells from cluster 5 in Fig 5A have a specific pathogen related function to clear infections with S. aureus. Although IFN-γ is upregulated during infection and particularly expressed by these transdifferentiated Th1 cells, this cytokine alone seems not to be the key factor of bacterial clearance, since neither INFg -/mice have higher bacterial burden after infection nor does anti-IFN-γ treatment reverse the phenotype in conditional T-bet-deficiency in Th17 cells. It is reasonable to speculate that not one single cytokine of transdifferentiated Th1 cells is responsible for the function but that a certain anti-bacterial profile cumulatively confers these effects. Future studies are required to uncover the precise effector functions of these protective antibacterial transdifferentiated Th1 cells.
Plasticity is supported by a publication showing that adoptive transfer of Th17 polarized cells into TCR αβ-deficient mice resulted in increasing numbers of IFN-γ expressing cells [38]. However, based on the design of this study, it was not possible to distinguish Th17 cell plasticity from expansion of Th1 cells that were co-transferred. Our approach using Th17 fate reporter mice shows plasticity of Th17 cells by cytokine profile analysis and single cell RNAsequencing. Moreover, we show the functional relevance of this transdifferentiation by using Tbx21 flox animals in combination with the Il17a-Cre mice in S. aureus elimination and highlight Tbx21-expressing ex-Th17 cells as an important T cell subset that comprises highly efficient anti-bacterial properties.
In conclusion, our study provides experimental data that supports the role of Th17 cells in the immune response to S. aureus infection in mice. We have uncovered a population of effector T cells in the kidney with expression of Tbx21 derived from Th17 cells that is highly effective in bacterial clearance. These data indicate, that therapeutic depletion of Th17 cells or neutralizing their cytokines in settings of autoimmune diseases [39] might have detrimental effects in systemic infections. Further studies are needed for a better understanding of the tissue-specific immune response to advocate bacterial clearance by supporting the T cell mediated immune response in the local micro-environment in order to reduce associated high mortality as well as the resulting tissue damage.

Ethics statement
All animal experiments were conducted according to the National Institutes of Health Guide for the Care and Use of Laboratory Animals as well as the German law for the welfare of animals. All animal experiments were approved by local authorities (BGV Hamburg, G35/16 and G82/19).

PLOS PATHOGENS
T-bet-dependent Th17 cell plasticity controls S. aureus infection

S. aureus sepsis
Bacteria (S. aureus strain SH1000) were homogenized by ultrasound and injected in tail vein of the mice in 100 μl PBS (Gibco by Life Technologies, Karlsbad, California). Bacteria were injected at indicated concentration (cfu/μL). For quantification of renal bacteria, kidneys were dissected and incubated for 40 min at 37˚C in RPMI1640 medium (including 0.01 M HEPES; 0.5 mL filtered FBS and 0.4 μg/ mL DNase and 8 μg/ mL collagenase D). GentleMACS (Miltenyi Biotec, Bergisch Gladbach, Germany) and sonication for 60 seconds using a Bandelin Sonorex Super RK103 were used for homogenization.
For quantification of hepatic or splenic bacteria these organs were passed through 70 μm nylon mesh (Corning, Corning, New York) and resuspended in PBS before sonification. The samples were incubated for 24 h at 37˚C on LB agar plates and colonies were counted. S. aureus was verified by MALDI-TOF.

Anti-IFN-γ treatment and CD4 + isolation and transfer
For anti-IFN-γ, mice were treated on day 4 and 7 after S. aureus infection with 500 μg/mouse i. p. α-Interferon-γ antibody XMG1.2. For T cell transfer, Leucocytes were isolated as described above from spleen of donor mice. CD4 + isolation was performed with the CD4 + T Cell Isolation Kit, mouse (Miltenyi Biotec) with manual cell separation by LS columns and MACS separator. 5x10 5 CD4 + cells were injected intravenously per mouse. After 3 days, mice were infected with S. aureus as described above.

PLOS PATHOGENS
T-bet-dependent Th17 cell plasticity controls S. aureus infection PBS (Live Technologies, Karlsbad, California) was used for labelling of dead cells for 2 min at room temperature before erythrocyte lysis (17 mM Tris-HCl (pH 7.6) and 144 mM ammonium chloride). Liver tissue was passed through a 100 μm and 40 μm nylon mesh (Corning, New York) in PBS and further processed like kidney tissue.

Flow cytometry
For stimulation, leucocytes were incubated in X-Vivo 20 Serum-free Hematopoietic Cell Medium (Lonza, Basel, Switzerland) supplemented with β-mercaptoethanol, brefeldin A, PMA and ionomycin for 3-4 h at 37˚C at 5% CO 2 (Heraeus instruments, Hanau, Germany). To prevent unspecific binding, leucocytes were incubated in blocking solution (MACS buffer containing 10% mouse serum) for 5 min at 4˚C. Next, samples were stained with the following antibodies: CD45 PercP For assessment of leucocytes counts in the kidney or liver by flow cytometry, cells were mixed with CountBright absolute counting beads (Life Technologies, Karlsbad, Carlifornia) and CD45 PerCP mAb (1:200). The individual cell frequencies were adjusted to the CD45 cell count. Flow cytometry was performed with a BD FACS LSRII or a LSR II Fortessa (BD Biosciences, Franklin Lakes, New Jersey) and data was analyzed with FlowJo (Tree Star).

Single cell RNA-isolation and library construction
Renal CD45 + CD3 + CD4 + YFP + cells from Il17aCre x R26eYFP mice were sorted using an Aria Fusion cytometer (BD Biosciences, Franklin Lakes, New Jersey) and collected in MACS buffer containing 2% FCS. These FACS-sorted cells underwent droplet-based single cell analysis and transcriptome library preparation using the Chromium Single Cell 3´Reagent Kits v2, Chromium Single Cell 3´Library & Gel Bead Kit v2, Chromium Single Cell A Chip Kit and Chromium i7 Multiplex Kit according to the manufacture´s protocols (10x Genomics, Pleasanton, California).

Pre-processing, dimensional reduction and cluster of single-cell RNA sequencing data
The processing of the single-cell data was done using the R software version 4.0.0 (2020-04-24). For the sake of reproducibility, we set the global seed to 0. If not mentioned otherwise, we

PLOS PATHOGENS
T-bet-dependent Th17 cell plasticity controls S. aureus infection ran the methods with default parameters. The R-Package Seurat (version 4.0.1) was used for pre-processing, dimensional reduction and cluster identification.
For the analysis of cells from IL-17A fate reporter mice, we removed cells according to the number of different genes expressed per cell (nGenes) (<673 (10th quantile) and >3924 (95th quantile)) and their percentage of mitochondrial genes (>8% mitochondrial genes). Next, the data was normalized and mitochondrial confounding effects diminished by applying Seurat's SCTransform [44] and adjusting for the percentage of mitochondrial genes. For dimensional reduction, principal components (PC) were calculated using the method Run PCA (features = Variable Features). We selected the PCs 1-40 to construct the Shared Nearest Neighbor Graph (Method: 'Find Neighbours') and the UMAP-Space (Method: 'Run UMAP'). The clusters were calculated by the Louvain algorithm (Find Clusters (resolution = 0.6)). While identifying the resulting clusters, we removed two clusters with high scores for cycle phases S, G2, M and another cluster with low alpha/beta TCR expression.
For comparative analysis of wildtype and Tbx21-deficient cells, we removed cells according to the number of different genes expressed per cell for Il17a Cre xTbx21 wt (<1000 and >3924 (95th quantile)) and Il17a Cre xTbx21 flox (<1000 and >2467 (95th quantile)) and their percentage of mitochondrial genes (>8%) independently. Additionally, cells with high cell cycle scores (Method: 'Cell Cycle Scoring'), were removed (Il17a Cre xTbx21 wt : S-score >0.2 and G2/Mscore > 0.2 / Il17a Cre xTbx21 flox : S-score >0.1 and G2/M-score > 0). Next, Il17a Cre xTbx21 wt and Il17a Cre xTbx21 flox were integrated by log-normalization each sample independently (Method: 'NormalizeData'), selection of the top 500 variable genes common in both samples (Method: 'Select Integration Features') and identification of common anchors (Method: 'Find Integration Anchors'). The samples were integrated using Seurats IntegrateData and the data was scaled by removal of mitochondrial confounding effects (Method: 'Scale Data'). The PCs 1-20 were used to construct the Shared Nearest Neighbor Graph (Method: 'Find Neighbours') and the UMAP-Space (Method: 'Run UMAP'). The clusters were calculated by the Louvain algorithm (Find Clusters (resolution = 0.5)). Clusters with low alpha/beta TCR expression were removed. Finally, clustering and dimensional reduction steps were performed.

Th17 trajectory construction and analysis
To construct the trajectories, we ran the R-Package slingshot (version 1.8.0) and its function slingshot (omega = T, stretch = FALSE). Seurat's UMAP coordinates and clusters were provided as data input. Cluster 3 annotated with Th17 was set as the start cluster. Next, we wanted to determine which genes alter along the slingshot trajectories. We used the R-package trade Seq (version 1.4.0) and fitted a generalized additive model on the genes by applying the method fit GAM (nknots = 6). As count matrix, we took the raw counts of the Seurat object and kept only variable genes. The cell weights and pseudotime was transferred by giving the slingshot object. Subsequently, we tested for genes associated with pseudotime using slingshots association test (lineages = T). To create Fig 4D,

PLOS PATHOGENS
T-bet-dependent Th17 cell plasticity controls S. aureus infection (nPoints = 50) to predict the expression of the top 50 genes ordered by Wald statistic for 50 timestamps along the pseudotime trajectory of lineage 1. Subsequently, we visualized the predicted gene expression of these genes using the function pheatmap (cluster_col = FALSE, border_color = NA, scale = "row") from the R-Package pheatmap (version 1.0.12).

Histology
Kidney was fixed overnight in 4% paraformaldehyde, washed with PBS and embedded in paraffin with a tissue processor (Leica, Wetzlar, Germany). For GR1 staining, paraffin blocks were cut with a microtome (Leica) to 1. For CD3 staining, paraffin blocks were cut with a microtome to 1.5 μm sections. Sections were transferred to a kryo frost microscope slide and incubated over night at 40˚C. Paraffin was removed by a descending alcohol series ended in dH 2 O. Slides were unmasked by cooking on 90˚C in Dako-buffer pH 9 for 15 min and cooled down on room temperature for 25 min. Sections were framed with Dakopen (DAKO, Hamburg, Germany) to keep solutions on the slide. Blocking was performed with blocking buffer from polapkit. The primary antibody rabbit polyclonal anti-human CD3 DAKO (1:1000) was incubated overnight at 4˚C. For AP-reaction incubation with anti-rabbit ap complex from polapkit 30 min at room temperature and for detection 12 min in new fuchsin solution in the dark. Before nucleus staining with haematoxylin slides were washed with distillated water. Slides were covered with gummi arabicum and cover glass. Slides were analyzed by counting CD3 + cells in 15 visual fields and 15 glomeruli in 400x optical magnification.
For PAS staining, paraffin blocks were cut with a microtome to 1.0 μm sections. Sections were transferred to a microscope slide and incubated over night at 40˚C. Paraffin was removed by a descending alcohol series ended in dH 2 O. Sections were incubated for 15 min at room temperature in 1% periodic acid, washed with dH 2 O, incubated in Schiff reagent for 40 min at room temperature and washed with dH 2 O. Nucleus staining was performed by staining with haematoxylin for 1 min and bleached with HCL alcohol. Slides were drained by an ascending alcohol series and were covered in Eukitt and cover glass.

PLOS PATHOGENS
T-bet-dependent Th17 cell plasticity controls S. aureus infection For Gram staining, paraffin blocks were cut with a microtome to 2.0 μm sections. Sections were transferred to a microscope slide and incubated over night at 40˚C. Paraffin was removed by a descending alcohol series ended in dH2O. Sections were first incubated 10 min at room temperature in nuclear fast red solution, second in gram´s crystal violet solution for 5 min and third in Lugol's solution for 1 min at room temperature. Slides were bleaches with Anilin (Emsure, Merck Millipore, Darmstadt, Germany) for 2 min, drained with 100% ethanol and xylol and covered in Eukitt and cover glass.

Electron microscopy
For electron microscopy the selected part of the mouse kidney was transferred from 4% formaldehyde into a cacodylate buffer with sucrose for 10 min at 80˚C. Afterwards, osmium tetroxyde was applied for 2 h. The specimen was washed in cacodylate buffer plus sucrose twice for 5 min. Subsequently, the sample was contrasted with uranyl acetate for 1 h. The specimen was put into ethanol baths with rising ethanol concentrations for 5 min in each bath, followed by bathing in Methyl tert-butyl ether (MTBE) plus epoxide mixture (in a 1:3 dilution) twice for 5 min each. Afterwards, the specimens were embedded in an epoxide mixture at 60˚C for 48 h and then at 100˚C for 11½ h. Semithin and ultrathin sections were cut on a Leica Microsystems microtome. Grids were purchased from Polyscience. The grids were then analyzed using electron microscopes (EM 109 and EM 902, Zeiss, Oberkochen, Germany) equipped with digital electron microscope cameras (Tröndle). 3 glomeruli from each mouse were analyzed.

Statistics
Statistical analyses were performed using Graph Pad Prism (La Jolla, CA). Data represent mean ± SEM. The following tests were used: Mann-Whitney test (two-tailed) and Dunnett's multiple comparison one-way ANOVA analysis. The results are shown as indicated at

PLOS PATHOGENS
T-bet-dependent Th17 cell plasticity controls S. aureus infection