TGF-β-mediated silencing of the genomic organizer SATB1 promotes Tfh cell differentiation and formation of intra-tumoral tertiary lymphoid structures

SUMMARY

The immune checkpoint receptor PD-1 on T follicular helper (Tfh) cells promotes Tfh:B cell interactions and appropriate positioning within tissues. Here we examined the impact of regulation of PD-1 expression by the genomic organizer SATB1 on Tfh cell differentiation. Vaccination of CD4^CreSatb1^f/f mice enriched for antigen-specific Tfh cells, and TGF-β-mediated repression of SATB1 enhanced Tfh differentiation of human T cells. Mechanistically, high Icos expression in Satb1^−/− CD4⁺ T cells promoted Tfh cell differentiation by preventing T follicular regulatory cell skewing, and resulted in increased isotype-switched B cell responses in vivo. Ovarian tumors in CD4^CreSatb1^f/f mice accumulated tumor antigen-specific, LIGHT⁺CXCL13⁺IL-21⁺ Tfh cells and tertiary lymphoid structures (TLS). TLS formation decreased tumor growth in a CD4⁺ T cell and CXCL13-dependent manner. Transfer of Tfh cells, but not naïve CD4⁺ T cells, induced TLS at tumor beds and decreased tumor growth. Thus, TGF-β-mediated silencing of Satb1 licenses Tfh cell differentiation, providing insight into the genesis of TLS within tumors.

Graphical Abstract

eTOC

Tertiary lymphoid structures (TLS) are associated with improved outcomes for cancer patients. Chaurio et al. demonstrate that TGF-β-mediated regulation of the genomic organizer Satb1 governs the differentiation of T follicular helper (Tfh) cells and T follicular regulatory cells. Loss of Satb1promotes Icos expression and thereby Tfh cell differentiation, resulting in the assembly of intra-tumoral TLS.

INTRODUCTION

Immuno-oncology today is sharply focused on T cells. However, B cells can present antigens and provide co-stimulation to T cells and coordinated T and B cell activation needs to occur to maximize immune protection. In the tumor microenvironment, T and B cells frequently interact (Biswas et al., 2021), but cooperation between humoral and cellular responses remains incompletely understood (Conejo-Garcia et al., 2020). T follicular helper (Tfh) cell-driven B cell responses play a role in the effectiveness of checkpoint inhibitors (Hollern et al., 2019). However, the drivers of Tfh conversion at tumor beds remain unknown. Some T:B cell intra-tumoral interactions take place in highly organized conglomerates that recapitulate the architecture of secondary lymph nodes, termed tertiary lymphoid structures (TLS) (Sautes-Fridman et al., 2019). Similar to TLS in autoimmune conditions, some intra-tumoral TLS contain germinal centers with interdigitating follicular dendritic cells, plus an adjacent T cell zone composed by CD4⁺ and CD8⁺ lymphocytes and high endothelial venules (Sautes-Fridman et al., 2019). The presence of TLS is associated with positive outcomes in multiple human malignancies, including ovarian (Iglesia et al., 2014; Kroeger et al., 2016; Montfort et al., 2017), pancreatic (Hiraoka et al., 2015), colorectal (Di Caro et al., 2014), bladder (Pfannstiel et al., 2019) and non-small cell lung cancer (Silina et al., 2018).

Because TLS are rarely found in mouse models, the sequence of events leading to the assembly of these structures remains incompletely understood. Nonetheless, a conducive chemokine niche, able to attract diverse hematopoietic components is generally accepted as a seminal event in TLS formation. Accordingly, delivery of the cytokine LIGHT promotes TLS assembly in vivo by inducing the production of CCL21 by tumor endothelial cells (Johansson-Percival et al., 2017). Additional studies also underscore the importance of other chemokines driving the recruitment of the immune cells that partake in the build-up of TLS during dendritic cell: T cell interactions. Mulé and colleagues, for instance, identified a 12-chemokine signature across multiple TLS⁺ human tumors (Messina et al., 2012), while seminal studies by the Engelhard and Storkus laboratories support the importance of lymphotoxin produced by effector lymphocytes, on the one hand, and IL-36 produced by dendritic cells, on the other, as drivers of the recruitment of the specialized cellular types that interact to generate TLS (Peske et al., 2015) (Weinstein and Storkus, 2015). Fibroblast activation is also important for TLS genesis (Buckley et al., 2015).

While the orchestration of the right cytokine and chemokine milieu seems to be crucial for TLS assembly, Tfh cells must be also generated for the formation and maintenance of germinal centers (Crotty, 2014, 2019). BCL6⁺ Tfh cells represent a different lineage of CD4⁺ T cells that secrete CXCL13, a marker of germinal center activity that attracts CXCR5⁺ B cells (Gu-Trantien et al., 2013; Havenar-Daughton et al., 2016), as well as IL-21, which elicits proliferation and maturation of B cells (Walters and Vinuesa, 2016). In turn, B cells act as antigen-presenting cells for CD4⁺ T cells initially primed for Tfh cell differentiation (Crotty, 2019). Proper priming of CD4⁺ T cells to skew their differentiation into Tfh cells could therefore be at the genesis of TLS assembly.

High expression of the immune checkpoint receptor PD-1 in Tfh cells is required for their positioning within tissues and fostering their interaction with B cells, which takes place through PD-L1:PD-1 interactions and CXCR3 repression (Shi et al., 2018). We have shown that the genomic organizer SATB1 governs epigenetic repression of PD-1 in T cells, and that PD-1 expression is induced downstream of TGF-β signaling by SMAD3-mediated repression of Satb1 expression (Stephen et al., 2017). SATB1 creates loops in heterochromatin and serves as a hub for the recruitment of epigenetic modifiers in multiple immune cell types, thus driving their phenotype and, subsequently, inflammatory and immune responses (Cai et al., 2006; Kakugawa et al., 2017; Kitagawa et al., 2017; Naik and Galande, 2019; Stephen et al., 2017; Tesone et al., 2016).

Interestingly, Tfh cells express lower amounts of SATB1 (Georgiev et al., 2018). Given the importance of PD-1 in Tfh:B cell interactions and the role of SATB1 in PD-1 repression, we investigated how SATB1 expression regulates Tfh cell differentiation and, subsequently, TLS formation at tumor beds. Deletion of SATB1 licensed Tfh cell differentiation and the induction of protective isotype-switched antibody responses. Enhanced Tfh differentiation and abrogated regulatory T cell (Treg) conversion of Satb1^−/− CD4⁺ T cells resulted in spontaneous formation of intra-tumoral TLS and decreased tumor growth. Transfer of Tfh cells to tumor-bearing mice was sufficient to elicit TLS formation at tumor beds. Our findings provide insight into TLS assembly in cancer, and suggest that the TGF-b-SATB1axis may be manipulated to enhance anti-tumor immunity.

RESULTS

Repression of Satb1 licenses activated CD4⁺ T cells for effective Tfh cell differentiation

To define the role of the genomic organizer Satb1 in Tfh cell differentiation, we first vaccinated >4 week-old CD4^CreSatb1^f/f mice with ovalbumin (OVA). In this model, OVA-reactive T cells progress through thymic selection and only a slight reduction in the proportions of T cell splenocytes is found in adult mice, with no difference in proliferative responses (Stephen et al., 2017) and recovery of regulatory T cell (Treg) function in adult mice (Kitagawa et al., 2017). Strikingly, >3-fold increases in the proportions of CD3⁺CD4⁺PD-1^hiCxcr5^hiBcl6⁺ Tfh cell subsets were consistently found in mice with Satb1^−/− T cells in both spleens (Figure 1A) and inguinal lymph nodes (Figure 1B) in independent experiments.

Figure 1. Repression of Satb1 licenses activated CD4⁺ T cells for effective Tfh cell differentiation.

(A, B) Quantification of the proportions of Tfh cells in the spleens (A) and inguinal lymph node (B) of OVA-vaccinated CD4^CreSatb1^f/f mice and Satb1^f/f control littermates (n≥20), after 14 days. (C) Corresponding Satb1-dependent increases in OVA antigen-specific Tfh cells obtained from Ly5.1⁺ mice after transfer of purified naïve OT-II Satb1^f/f (n=14) or OT-II CD4^CreSatb1^f/f (n=15) T cells, followed by immunization with NP-OVAL and analysis 5 days later. (D, E) Production of IL-21 (n=9) (D) and CXCL13 (n=13) (E) quantified through intracellular flow cytometry staining after stimulation with PMA (100 ng/mL), ionomycin (0.5 μg/mL) and brefeldin A (10 μg/mL) for 6 hours in CD3⁺CD4⁺Foxp3⁻CD62L⁻CD44⁺PD1^hiCXCR5⁺ Tfh and naïve CD3⁺CD4⁺Foxp3⁻CD62L⁺CD44⁻ cells from the spleens of vaccinated mice as in A&B. (F) Serum levels of CXCL13 in OVA-vaccinated mice as in A&B. (G) Production of LIGHT/TNFSF14 quantified as in D&E (n=11). (H) Differential OVA-specific titers of total IgG responses in the serum of Satb1+/Satb1− (n=12) vaccinated mice as in A&B. Serum from naïve mice were used as control (n=6). (I) Quantification by flow cytometry of surface IgG1 and the GL7 and FAS markers of germinal center activity in splenic CD19⁺B220⁺ B cells in CD4^CreSatb1^f/f mice and littermates from A. Data are presented as median. A-E, G and H are pooled from ≥2 independent experiments. F&I are representative of at least three independent experiments. *p<0.05, **p≤0.01, ***p≤0.001.

These responses were antigen-specific because corresponding Satb1-dependent increases in Tfh cell differentiation were identified in (OVA-specific) OT-II⁺ T cells injected at the time of immunization (Figure 1C). In contrast, although we observed a trend towards basal increases in Tfh cells in CD4^creSatb1^f/f mice, Satb1^−/− antigen-specific (OT-II) cells did not proliferate in the absence of the OVA-antigen (Suppl. Figure 1A&B).

Differentiated lymphocytes were functional and not merely expressing Tfh cell surface markers, as they produced more IL-21 and CXCL13, compared to naïve and Tfh cells in identically vaccinated Satb1-competent mice; both at the protein (Figure 1D&E) and mRNA abundance (Suppl. Figure 1C). Accordingly, higher levels of serum CXCL13, a marker of germinal center activity (Gu-Trantien et al., 2013; Havenar-Daughton et al., 2016), were also found in the serum of OVA-vaccinated mice carrying Satb1^−/− T cells (Figure 1F). Interestingly, we also found that (Satb1-competent) Satb1^f/f and Satb1^−/− CD4⁺ Tfh cells produced similar levels of the Light/TNFSF14 cytokine (Figure 1G), which is sufficient to trigger TLS formation in cancer through the activation of tumor endothelium (Johansson-Percival et al., 2017). Furthermore, increases in the titers of serum OVA-specific IgG were identified in CD4^CreSatb1^f/f mice, compared to Satb1⁺ littermates (Figure 1H). Hence, upon OVA immunization, CD19⁺CD20⁺ B cells from CD4^CreSatb1^f/f mice showed higher proportions of the cell surface IgG1 isotype compared to control littermates, along with positive expression of FAS and GL7, markers of germinal center activity (Williams et al., 2015), (Figure 1I).

Together, these results indicate that Satb1 silencing in CD4⁺ T cells intrinsically promotes their differentiation into functional antigen-specific Tfh cells, resulting in a chemokine milieu conducive of B-cell recruitment and activation and, accordingly, antigen-specific, isotype-switched humoral responses.

Decreased expression of Satb1 promotes Tfh differentiation through Icos de-repression and decreased T follicular regulatory cell (Tfr) formation

To define possible drivers of SATB1 downregulation during Tfh cell differentiation, we focused on TGF-β, known to decrease Satb1 expression in T cells (Stephen et al., 2017) while also promoting Tfh cell polarization (Schmitt et al., 2014). As shown in Figure 2A&B, under skewing conditions that included known drivers of Tfh vs. Th17 cell differentiation (Schmitt et al., 2014), human CD4⁺ T cells differentiated more effectively into CXCR5⁺PD-1^hiICOS⁺BCL6⁺ Tfh cells in the presence of TGF-β (Crotty, 2014; Marshall et al., 2015; Schmitt et al., 2014), with a consistent reduction of SATB1 (Figure 2C&D). Challenging the notion that TGF-β is constantly immunosuppressive in cancer (Park et al., 2016; Stephen et al., 2017; Stephen et al., 2014), ovarian tumors from the TCGA database expressing higher levels of TGFB exhibited concurrent increases not only in markers of protective humoral responses (Biswas et al., 2021), (Suppl. Figure 2A); but also in markers of denser T cell infiltration (CD3ε, CD8α) or enhanced cytolytic activity (IFNG, PRF1); (Suppl. Figure 2B).

Figure 2. TGF-β-mediated repression of SATB1 expression promotes Tfh cell phenotypes.

Representative dot-plots (left) and histograms (right) of the effect of TGF-β on human Tfh cell differentiation and SATB1 repression, respectively. Increased in vitro expression of human CXCR5⁺PD-1^hi Tfh cells markers from peripheral naïve CD4⁺ T cells in the presence of TGF-β, under different Tfh cell skewing conditions: (A) (lower panel) hIL-23 plus hIL-10; (upper panel) hIL-23; (B) (lower panel) hIL-12 plus hIL-10; (upper panel) hIL-12. (C) TGF-β-dependent Tfh cell differentiation was associated with lower levels of SATB1 (D), as determined by intracellular staining. A-D data (n=3) are representative of two independent experiments with similar results and represent mean ± SEM. *p<0.05, **p≤0.01, ***p≤0.001.

To understand how Satb1 down-regulation licenses Tfh cell differentiation, we first observed that Satb1^−/− CD4⁺ T cells from OVA-vaccinated mice exhibited higher levels of ICOS, a crucial signal for Tfh cell formation (Crotty, 2014) (Figure 3A-C). Supporting a direct role for Satb1 in repressing Icos, chromatin immuno-precipitation (ChIP) experiments identified occupancy of Satb1 around 60 bp upstream of the Icos Transcription Start Sites (Figure 3D), consistent with the silencing activity of Satb1 on other genes (Stephen et al., 2017).

Figure 3. Decreased expression of Satb1 promotes Tfh cell differentiation by derepressing Icos.

(A) Higher expression of Icos in Satb1^−/− CD4⁺ T cells (n=11). (B) Higher proportions of Icos⁺CD4⁺ T cells among lymphocytes in the absence of Satb1 (n≥19). (C) Higher proportions of Icos⁺ Tfh cells after OVA vaccination in the presence of Satb1^−/− CD4⁺ T cells (n=22). (D) (Left) Schematic depiction of Satb1 occupancy at the Icos promoter. Location of primers used for ChIP-PCR is indicated. (Right) Chromatin was immunoprecipitated with anti-Satb1 or control IgG from negatively immunopurified Satb1-competent mouse CD4⁺ T cells. Percent of input before immunoprecipitation (2.5% input values) was quantified by Real-Time Q-PCR (n=7). (E) anti-Icos / IgG (200 μg/mouse) was administered intraperitoneally in congenic Ly5.1⁺ mice reconstituted with OT-II-Satb1^f/f vs. OT-II-CD4^CreSatb1^f/f splenocytes followed by immunization with NP-OVAL in alum, and Tfh cell differentiation was quantified in spleens 5 days later (n≥12). Data are presented as median and pooled from ≥2 independent experiments. **p≤0.01, ***p≤0.001.

As expected from the role of ICOS signaling in Tfh cell differentiation, ICOS blockade in OVA-vaccinated mice reduced, but did not fully eliminate, Satb1-dependent differences in the proportions of OVA-specific Tfh cells (Figure 3E).

To identify additional mechanisms by which Satb1 silencing could promote Tfh cell differentiation, we also investigated the role of Satb1 in the formation of PD-1^hiCXCR5⁺FoxP3⁺ Tfr cells, which inhibit the function of Tfh cells (Sage et al., 2013; Sage et al., 2016) and, through neuritin, B cells (Gonzalez-Figueroa et al., 2021). In agreement with previous reports (Kitagawa et al., 2017), natural Treg function was preserved in adult CD4^CreSatb1^f/f mice (Figure 4A). However, the capacity of CD3⁺CD4⁺Foxp3⁻CD62L⁺CD44⁻ naïve T cells sorted from CD4^CreSatb1^f/fFoxF3^GFP mice to differentiate into FoxP3⁺ Treg cells under skewing conditions was dramatically reduced, while their Satb1-competent counterparts effectively acquired FoxP3 (Figure 4B). Lack of Treg conversion cannot be attributed to a proliferation defect (Stephen et al., 2017), as similar cell counts and Cell-Trace Violet dilution were found under differentiation conditions for both populations (Figure 4C). Mechanistically, ChIP-PCR demonstrated that Satb1 bound to both the CNS1 and CNS2 intronic enhancers in the FoxP3 gene (Figure 4D), originally identified by the Rudensky laboratory (Zheng et al., 2010). Accordingly, the formation of PD-1^hiCXCR5^hiBCL6⁺FoxP3⁺ Tfr cells was reduced in OVA-vaccinated mice carrying Satb1^−/− T cells, with higher ratios of Tfh to Tfr cells, compared to Satb1-competent littermates (Figure 4E-G).

Figure 4. Satb1 is required for the generation of induced regulatory T cells.

(A) Natural CD3⁺CD4⁺FoxP3⁺ Treg from the spleens of adult (week 10) Satb1^−/− mice effectively suppressed CD3/CD28-induced T cell splenocyte activation in a dose-dependent manner. Representative of two independent experiments. (B) CD3⁺CD4⁺Foxp3⁻CD62L⁺CD44⁻ naïve T cells sorted from CD4^CreSatb1^f/fFoxP3^GFP mice or control Satb1^f/fFoxP3^GFP littermates were in vitro differentiated to Treg cells for 3 days. Conversion into FoxP3⁺ Treg cells was compromised in CD4^CreSatb1^f/fFoxP3^GFP cells. (C) (Left) Satb1-dependent differences in Treg conversion (n=15); (Right) Similar proliferation rates were observed under differentiation conditions for Satb1⁺/Satb1⁻ populations. (D) ChIP-PCR for the CNS1 (left panel) and CNS2 (right panel) enhancers in Foxp3. Chromatin was immunoprecipitated with anti-Satb1 or control IgG from FoxP3⁺ Treg cells, in vitro differentiated from naïve Satb1-competent CD4⁺ T cells. Percent of input before immunoprecipitation (2.5% input values) was quantified by Real-Time Q-PCR (n=5). Mean and SEM are shown. (E) Representative dot plots showing the gating strategy for Tfh and Tfr cells in OVA-vaccinated mice carrying Satb1^−/− T cells, compared to Satb1-competent littermates. (F) Reduced generation of PD-1^hiCXCR5⁺Foxp3⁺ Tfr cells with corresponding increases in FoxP3⁻ Tfh cells (n≥20). (G) Higher ratios Tfh to Tfr cells in Satb1^−/− T cells compared to control littermates (n≥20). Data are presented as median and pooled from ≥2 independent experiments. ***p≤0.001.

Taken together, these results demonstrate that Satb1 governs CD4⁺ T cell differentiation into Treg/Tfr and Tfh cells in opposite ways: On the one hand, although Satb1 is dispensable for natural Treg-dependent tolerance in adult mice (Kitagawa et al., 2017), probably due to constitutive de-methylation of CNS2 in thymic Treg, it is required for efficient Treg conversion. On the other hand, Satb1 represses the Icos promoter. Accordingly, Satb1 silencing licensed Tfh cell formation while preventing the generation of Tfr cells.

Satb1-dependent Tfh differentiation drives enhanced B cell responses and TLS formation in orthotopic ovarian tumors

To understand the consequences of enhanced Tfh cell formation in Satb1^−/− T cells in cancer, we challenged CD4^CreSatb1^f/f mice and Satb1-competent littermates with intraperitoneal UPK10 tumors, a mildly immunogenic model of p53/KRas-driven ovarian carcinosarcoma that progresses as solid tumor masses in peritoneal lining (Cubillos-Ruiz et al., 2015; Rutkowski et al., 2015; Scarlett et al., 2012; Tesone et al., 2016). As expected, splenic (Figure 5A) and intratumoral (Figure 5B) CD3⁺CD4⁺PD-1^hiCXCR5^hiBCL6⁺ Tfh cells accumulated in greater proportions when Satb1 was specifically absent in T cells, which resulted in higher Tfh to Tfr cell ratios in the absence of Satb1 expression in T cells (Suppl. Figure 2C&D). As we observed in the OVA-vaccinated mice, Satb1^−/− Tfh cells expressed higher levels of ICOS in both spleen (Suppl. Figure 2E&F) and at tumor beds (Figure 5C&D). Interestingly, intraperitoneal tumors from CD4^CreSatb1^f/f mice were strongly infiltrated by CD45⁺CD19⁺B220⁺ B cells (Suppl. Figure 2G). Moreover, the accumulation of isotype-switched B cells both in the spleen and tumor beds was enhanced in CD4^CreSatb1^f/f mice (Figure 5E&F), indicative of enhanced humoral responses. Accordingly, a higher anti-tumor humoral (IgG1) response was observed in serum from CD4^CreSatb1^f/f tumor-bearing mice, compared to the Satb1-sufficient littermates (Suppl. Figure 3A&B). These responses appear to be maintained at tumor beds, because immunization of naïve mice with irradiated tumor cells did not increase the titer of circulating antibodies (Suppl. Figure 3C).

Figure 5. Satb1-dependent Tfh cell differentiation promotes B cell responses and TLS formation in ovarian cancer.

(A, B) CD4^CreSatb1^f/f mice accumulated more Tfh cells in the spleen (A) and at UPK10 intraperitoneal tumor beds (B), compared to control Satb1^f/f mice; (n≥16). (C) Higher ICOS expression in intra-tumoral Satb1^−/− Tfh cells (n=17) and controls (n=18). (D) Corresponding increases in the accumulation of ICOS⁺CD4⁺ T cells at tumor beds (n=15). (E, F) Increased accumulation of isotype-switched B cells in the spleen (E) and at tumor beds (F) in CD4^CreSatb1^f/f mice, compared to Satb1-competent controls; (n=12). (G) B cell depletion using anti-B220 antibodies IP (200 μg/mouse; days −2, 0, 2, 4, 6 after tumor challenge) accelerates UPK10 tumor growth in Satb1^f/f mice compared to isotype control (n≥9). (H) Spontaneous large PNAd⁺ TLS formation in intraperitoneal tumors in CD4^CreSatb1^f/f mice (top), compared to Satb1^f/f mice (bottom). Representative of ≥3 independent experiments. (I) Quantification of TLS (defined as adjacent conglomerates of T cells and B-cells at tumor beds), based on the area covered by CD19⁺ cells plus the area covered by CD3⁺ cells). (J) Ratio of the area occupied by TLS to the entire tumor section. For I&J: Each point represents a single TLS in the CD4^CreSatb1^f/f (n=9 mice) and Satb1^f/f n=11 mice) groups. (K) TLS assembly was abrogated upon CXCL13 blockade using anti-CXCL13 antibodies IP (150 μg/mouse; days 3, 6, 9 after tumor challenge). (L) Quantification of TLS after CXCL13 blocking as performed in I (2 independent experiments). (M) (Upper panel) Schematic depiction of the experiment. A chunk of IP tumor from a donor mouse was isolated and s.c. transplanted to a naïve recipient. (Lower panel) TLS undergo major changes in the distribution of T and B cells over time. Data are presented as median and pooled from ≥2 independent experiments. *p<0.05, **p≤0.01, ***p≤0.001.

As in ovarian cancer patients (Biswas et al., 2021; Montfort et al., 2017), B cell responses in our model were associated with immune protection, because B cell depletion using anti-B220 antibodies accelerated UPK10 tumor growth (Figure 5G). Interestingly, depletion of CD8⁺ T cells in UPK10 tumor-bearing mice also accelerated malignant progression (Suppl. Figure 3D), which supports that both arms of the adaptive immune system work in coordination to achieve superior immune pressure against the progression of human ovarian cancer (Biswas et al., 2021).

We next assessed tumors histologically to gain deeper insight in the formation of TLS. Strikingly, we found enhanced spontaneous formation of TLS in intraperitoneal tumors in CD4^CreSatb1^f/f mice, in multiple independent experiments. We defined those structures as >0.1 mm² conglomerates of B cells adjacent to accumulations of T cells, clearly identified inside tumor masses and co-stained for peripheral node addressin (PNAd), with a central core of T cells (Figure 5H). These structures contained CD21⁺ follicular dendritic cells, BCL6⁺ B cells (Suppl. Figure 4A), and CD23, a marker representative of fully mature secondary-follicle-like TLS (Suppl. Figure 4B). The presence of TLS correlates with higher densities of B cells and a trend towards increased T cell infiltration outside TLS (Suppl. Figure 3E).

TLS assembly was not restricted to UPK10 tumors, as it was also obvious in intraperitoneal Brpkp110 tumors (Allegrezza et al., 2016) (Suppl. Figure 4C). Supporting previous reports on spontaneous TLS assembly in peritoneal tumors (Engelhard et al., 2018; Rodriguez et al., 2018), we also found scattered conglomerates of T and B cells in tumors growing in Satb1-competent mice, although they were less frequent, poorly organized and smaller (Figure 5H-J). Importantly, TLS assembly was dependent on CXCL13, because CXCL13 blockade abrogated TLS formation in CD4^CreSatb1^f/f intraperitoneal tumor-bearing mice (Figure 5K&L), with higher tumor growth (Suppl. Figure 3F).

Enhanced tumor immunogenicity also promoted TLS formation, because intraperitoneal UPK10 tumors expressing low levels of OVA (UPK10-OVA^lo, quasi-universally orchestrated these structures in WT mice (Figure 5M).

To define possible changes in the distribution of T and B cells in TLS over time, we transplanted fragments of intraperitoneal OVA^lo UPK10 tumors into the flank of naïve WT mice. As shown in Figure 5M, the central distribution of T cells in early-stage TLS changed to the periphery, as B cells occupied the TLS center over time, which is the structure more frequently observed in advanced human tumors (Cillo et al., 2020). Together, these results indicate that increased differentiation of Tfh cells in cancer drives enhanced humoral responses in ovarian cancer, which promotes spontaneous orchestration of TLS where T and B cells dynamically interact at tumor beds.

TLS assembly is driven by antigen-specific Tfh cells and delays ovarian cancer progression

A limitation of the CD4^CreSatb1^f/f model is that deregulation of PD-1 expression abrogates the effector activity of tumor-reactive CD8⁺ T cells (Stephen et al., 2017). To segregate the role of Satb1 specifically in CD4⁺ T cells, and to avoid contamination of secondary lymphoid organs, we reconstituted Rag1^−/− mice with a mix of syngeneic Satb1⁺ wild-type CD8⁺ T cells and B cells, plus either Satb1⁺ or Satb1^−/− CD4⁺ T cells. Both cohorts of mice were subsequently challenged with UPK10 intraperitoneal tumors and followed for malignant progression (Figure 6A), Tfh cell differentiation and potential anti-tumor B cell responses. As expected, increases in CD3⁺CD4⁺CXCR5⁺PD-1^hiBCL6⁺ Tfh cells were found at both spleens (Suppl. Figure 5A) and tumor beds (Figure 6B) in mice reconstituted with Satb1⁻CD4⁺ T cells. Likewise, increased Tfh cell differentiation translated into increased B cell accumulation at both spleens (Suppl. Figure 5B) and in the tumor (Figure 6C), including increases in CD19⁺ cells expressing markers of germinal center activity, such as FAS (Figure 6D & Suppl. Figure 5C) and GL7 (Figure 6E & Suppl. Figure 5D). These responses were associated with corresponding increases in isotype-switched B cells (Figure 6F & Suppl. Figure 5E). Notably, enhanced Tfh cell conversion in the absence of Satb1 was sufficient to induce the orchestration of large TLS, which again at this early stage exhibited a central core of T cells surrounded by B cells (Figure 6G), while small conglomerates of T- and B-cells were found in mice reconstituted with Satb1-competent CD4⁺ T cells (Figure 6G-K).

Figure 6. Tfh cell-driven, TLS-associated humoral responses are associated with delayed malignant progression in ovarian cancer.

(A) Schematic depiction of the experiment. Rag1^−/− mice were reconstituted with a mix of syngeneic wild-type CD8⁺ T cells, wild-type B-cells and either Satb1-competent or Satb1^−/− CD4⁺ T cells. Mice were subsequently challenged with UPK10 intraperitoneal tumors and followed for malignant progression and humoral responses. (B-F) Flow cytometry quantification of (B) Tfh cells, (C) CD19⁺ B cells (n=12); (D, E) Germinal center markers FAS (n=11) and GL7 (n=13), and (F) B cell isotype switching at tumor beds in reconstituted Rag1^−/− mice (n=7). (G) Spontaneous orchestration of large PNAd⁺ TLS in intraperitoneal tumors in Rag1^−/− mice reconstituted with CD4^CreSatb1^f/f CD4⁺ T cells (Upper panel), but not upon reconstitution with Satb1^f/f control CD4⁺ T cells (Lower panel). (H, I) Intra-tumoral TLS, defined as adjacent conglomerates (>0.1mm²) of CD3⁺ T cells and CD19⁺ B-cells at tumor beds, were quantified based on the (H) B cell content (CD19⁺ area) and (I) T cells (CD3⁺ area) inside the structure. (J) At least 60% of the Satb1^−/− CD4⁺ T cells reconstituted mice, developed large TLS (>2 mm²) quantified as CD19⁺ plus CD3⁺ area. (K) Ratio of the area occupied by TLS to the entire tumor section. (L) Differences in IP tumor burden for the mice depicted in A. Each point represents the sum of total tumor content in a single mouse (n≥20). (M) Multiplex immunofluorescence for CD3⁺CD4⁺BCL6⁺CXCL13⁺ Tfh cells inside intra-tumoral TLS. G&M are representative of ≥3 independent experiments. From graphics H to K (n=10 mice) for both CD4^CreSatb1^f/f and Satb1^f/f experimental groups. For J&K each point represents a single TLS. *p<0.05, **p≤0.01, ***p≤0.001.

TLS in mice reconstituted with Satb1-defective CD4⁺ T cells also contained CD21⁺ dendritic cells, BCL6⁺ B-cells (Suppl. Figure 6A&B) and CD23⁺ cells (Suppl. Figure 6C) that characterize bona fide lymphoid organs, and were infiltrated by CD3⁺CD4⁺BCL6⁺ Tfh cells producing CXCL13 (Figure 6M), which chemoattracts CXCR5⁺ B cells.

CD4⁺ T cell-driven B cell responses were again associated with a reduction in tumor growth, further supporting the protective role of humoral activity in ovarian cancer (Figure 6L & Suppl. Figure 5F). Because the only difference in this experiment is the expression of Satb1 in CD4⁺ T cells, these results indicate that Tfh cells are crucial for TLS formation and pressure against malignant progression.

To confirm that Tfh cell conversion is sufficient to drive TLS assembly in cancer, we reconstituted Rag1^−/− mice with WT T- and B-cells, and challenged them with UPK10 tumors, admixed with either CD3⁺CD4⁺PD-1^hiICOS⁺Bcl6⁺ Tfh cells (Suppl. Figure 6D) or naïve CD3⁺CD4⁺CD62L⁺CD44⁻ T lymphocytes from the UPK10 tumor-draining inguinal lymph nodes from a different cohort of mice (Figure 7A). The addition of Tfh cells resulted not only in more TLS formation (Figure 7B&C), but also promoted the orchestration of large TLS (>2 mm²) in 50% of mice, while no TLS of this size was found in tumors developed in the presence of naïve control CD4⁺ T cells from the same donors (Figure 7D). Tfh cell-induced TLS exhibited CD31⁺PNAd⁺ HEVs, primarily located in the B-cell area of the germinal center (Figure 7D). Finally, tumors in mice with TLS induced by Tfh cell transfer grew less, compared to those developed in mice receiving control CD4⁺ T cells (Figure 7E). Together, these results further support that Satb1-dependent conversion of Tfh cells is the genesis of TLS formation in cancer, and open new avenues for eliciting TLS formation in unresectable tumors.

Figure 7. Satb1-dependent Tfh cells trigger the formation of large TLS at tumor beds.

(A) Scheme of the experiment: Rag1^−/− mice were reconstituted with a mix of syngeneic wild-type CD4⁺ and CD8⁺ T cells, and wild-type B-cells. Fourteen days later, mice were subsequently challenged with UPK10 intraperitoneal tumors, admixed with either CD3⁺CD4⁺PD-1^hiICOS⁺(BCL-6⁺) Tfh cells or CD4⁺CD3⁺CD62L⁺CD44⁻ naïve T cells, sorted from the tumor-draining inguinal lymph nodes of syngeneic tumor-bearing donor mice. TLS assembly was assessed after 14 days. BCL-6 expression was verified in sorted populations independently. (B) Quantification of total TLS area (TLS defined as adjacent conglomerates of CD3⁺ T cells and CD19⁺ B-cells) at tumor beds. (C) Ratio of the area occupied by TLS to the entire tumor section. (D) (Upper panel) Representative IHC of adjacent tumor sections showing large TLSs in Rag1^−/− mice reconstituted with Tfh cells, consisting of defined areas of CD19⁺ and CD3⁺ aggregations with corresponding PNAd and CD31 positive signal, and its controls (Lower panel). (E) Differences in IP tumor growth for the mice depicted in A. Each point represents the sum of total tumor content in a single mouse (n≥8). Pooled from two independent experiments. For B&C each point represents a single TLS for both Tfh cell (n=4 mice) and naïve T cell (n=4 mice) groups. Data are presented as median and representative of two independent experiments with similar results. *p<0.05, **p≤0.01.

DISCUSSSION

We found that decreased expression of Satb1 in CD4⁺ T cells licensed antigen-specific Tfh cell differentiation through Icos de-repression and by impairing Tfr cell formation, driving isotype-switched antibody responses. Enhanced generation of Tfh cells created a chemokine niche that promoted spontaneous assembly of TLS at tumor beds, while Tfh cells were sufficient to elicit TLS formation in the tumor microenvironment.

SATB1 is a genomic organizer with multiple activities in different immune cell types (Cai et al., 2006; Kakugawa et al., 2017; Kitagawa et al., 2017; Naik and Galande, 2019; Stephen et al., 2017; Tesone et al., 2016). By creating loops in heterochromatin and recruiting epigenetic modifiers to enhancers and promoters, SATB1 governs complex phenotypes that go beyond the narrower range of genes typically regulated by individual transcription factors. Our results unveil Icos as a direct target of Satb1-dependent repression. In addition, Treg cell conversion was abrogated when CD4⁺ T cells lack Satb1. While Sakaguchi and colleagues identified an initial defect in thymic Treg in the absence of Satb1 specifically in T cells, differences in immunosuppressive activity are corrected in adulthood (Kitagawa et al., 2017). In contrast, our results show that adult CD4⁺ T cells remained unable to differentiate into FoxP3⁺ cells without Satb1 under TGF-β-dependent skewing conditions, suggesting a different role in thymic versus peripheral Treg cells. It is likely that the binding of Satb1 to the intronic enhancers of FoxP3 (Kitagawa et al., 2017) becomes irrelevant once DNA de-methylation has been established in natural Treg cells, but remains necessary for FoxP3 expression in converted Treg lymphocytes. Nevertheless, our study unveils that the combined activities of SATB1 include a negative role in Tfh cell differentiation, along with other previously identified genetic repressors (Wang et al., 2014), and show that SATB1 is a necessary player in Treg conversion.

Another key finding of our study is that Tfh cells produced the cytokine milieu that is at the genesis of intratumoral TLS. Tfh cells were consistently located at the central core of TLS, which we propose represented the initial stage of TLS formation, before germinal centers changed this architecture. Other studies point to fibroblasts as pivotal drivers of TLS formation at the earliest phases of TLS establishment, preceding lymphocyte infiltration in non-tumor tissues (Buckley et al., 2015; Nayar et al., 2019). The importance of fibroblasts in cancer TLS is supported by studies from the Mulé group, where a lymph node-derived stromal cell line induced the formation of TLS in immunogenic tumor models (Zhu et al., 2018). Fibroblast reticular cells in secondary lymphoid organs produce CCL21, which mediates homing of T cells, but not B lymphocytes. Consistently, CCL21 has been included in a 12-chemokine signature that reliably identifies TLS in multiple human tumors (Messina et al., 2012). However, CCL21 production by different cell types (i.e., tumor endothelial cells) can be induced through signaling by the TNF-α family member cytokine LIGHT (Johansson-Percival et al., 2017). We found that Tfh cells were significant producers of LIGHT, so that Tfh cells at tumor beds could indeed provide the primordial signal for TLS assembly in cancer, besides their necessary role in the formation of germinal centers.

Another element found to be critical for TLS formation at tumor beds by other authors is lymphotoxin secreted by CD8⁺ T cells and NK cells, which acts on tumor blood vessels to transform them into “lymph node-like” vasculature (Peske et al., 2015). This recapitulates the importance of lymphotoxin derived from lymphoid tissue inducer cells (LTi) during the development of secondary lymphoid organs (Germain et al., 2015). However, we found that Tfh cells were substantial producers of the lymphotoxin homologous TNFSF14/LIGTH cytokine, while activated CD4⁺ T cells can substitute LTi cells in TLS generated in autoimmune conditions (Marinkovic et al., 2006), further supporting that Tfh cells are a crucial player in the assembly of primordial intratumoral TLS.

Overall, our results point to a model whereby activated CD4⁺ T cells that decrease expression of SATB1 in the tumor microenvironment (i.e., in response to TGF-β signaling) secrete LIGHT, which activates CCL21 production by endothelial cells and likely fibroblasts; CXCL13, which drives the recruitment of CXCR5⁺ B cells (Gu-Trantien et al., 2017); and IL-21, which activates B cells in situ. A crosstalk between T and B cells is established through ICOS:ICOS-ligand and CD40:CD40L interactions, while B cells sustain T cell activation through antigen presentation, eventually leading to the recruitment, differentiation and distribution of dendritic cells, high endothelial venules, fibroblasts, plasma cells, neutrophils and macrophages that compose canonical TLS in cancer. Intra-tumoral administration of autologous antigen-specific Tfh cells in metastatic cancers or tumors adhered to unresectable locations could therefore promote the generation of TLS where anti-tumor T cells could find a protective niche to exert immune pressure against the progression of advanced malignancies, including supporting existing immunotherapies.

Limitations of the study

Our study unveiled Tfh cell differentiation as a seminal event during the assembly of TLS in cancer. We cannot rule out that there are multiple ways of orchestrating TLS, but intratumoral Tfh administration was sufficient to elicit TLS formation. Since Tfh cells came from the lymph nodes of tumor-bearing mice, future studies should determine whether these lymphocytes need to be tumor antigen-specific or not. It will be also important to confirm these findings in other tumor context, as well as in human cancer patients.

We also show that CXCL13 blockade abrogated intratumoral TLS assembly. A caveat is that CXCL13 is also produced by stromal cells such as follicular dendritic cells and marginal reticular cells, so these effects cannot be exclusively attributed to Tfh cells.

Finally, we unveiled changes in the distribution of T and B cells during the progression of TLS in vivo. A limitation is that TLS evolved at a different anatomical location (the flank vs. the peritoneal cavity), which could drive specific architectural patterns.

STAR METHODS

RESOURCE AVAILABILITY

Lead Contact

Further information and requests for supporting data and resources should be directed to and will be fulfilled upon reasonable request by the Lead Contact, Jose R. Conejo-Garcia, MD, PhD (jose.conejo-garcia@moffitt.org).

Materials availability

This study did not generate new unique reagents. The authors declare that all the results supporting the findings of this study are available within the paper and its figures.

Data and code availability

This paper does not report original code. Any additional information required to reanalyze the data reported, as well as raw image for TLS IHC studies, are available from the lead contact upon request.

EXPERIMENTAL MODELS

Mice and human cells

Healthy human buffy coats were obtained from LifeSouth (Gainesville, FL, USA). Studies were exempted from Institutional Review Board (IRB) approval.

Satb1^−/− mice were generated in the Wistar Institute’s transgenic facility on a C57BL/6 background (Tesone et al., 2016). Satb1^f/f mice were crossed with either CD4^Cre mice (Taconic, 4196M) or FoxP3^GFP (Jackson Labs.) to generate CD4^CreSatb1^f/f or Satb1^f/fFoxP3^GFP respectively. The CD4^CreSatb1^f/f mice were crossed with Satb1^f/fFoxP3^GFP to generate CD4^CreSatb1^f/fFoxP3^GFP and control littermates. The OT-II CD4^CreSatb1^f/f mice were generated by crossing Satb1^f/f mice with the OT-II (C57BL/6-Tg (TcraTcrb)425Cbn/Crl, strain code 643 (Charles River) to generate OT-II Satb1^f/f in a C57BL/6 background. After three generations, OT-II Satb1^f/f were crossed with CD4^CreSatb1^f/f to generate triple-transgenic OT-II CD4^CreSatb1^f/f and control littermates. 6–8-week-old female Rag1-deficient (Rag1^−/−) mice of the same age group were procured from Jackson Laboratory. Mice were maintained by the animal facility of the Moffitt Cancer Center, with a 12 light/12 dark cycle, at 18–23 °C with 40–60% humidity and maintained in pathogen-free barrier facilities. All animal studies were reviewed and approved by the Institutional Animal Care and Use Committee at the University of South Florida Research Integrity and Compliance department.

METHOD DETAILS

In vivo tumor challenge

In vitro culture of UPK10 cells, a p53/KRas-driven ovarian carcinosarcoma cell line, was done in complete R10 medium defined as RPMI-1640 (Sigma), supplemented with L-glutamine (2 mM, Gen Clone), penicillin (100 U/mL, Lonza), streptomycin (100 μg/mL Lonza) and 10% fetal bovine serum (Sigma) at 37°C, 5% CO₂. Unless otherwise indicated, UPK10-induced-intraperitoneal (IP) tumors were initiated by injecting 6 × 10⁶ cells/mouse in PBS into the peritoneal cavity. Animals were monitored for health concerns and euthanized after 14 days. IP tumor contents were extracted for either tumor dissociation or prepared for immunohistochemistry analysis. For flow cytometry analysis, tumor tissues were dissected mechanically into single-cell suspensions in complete medium. UPK10 flank tumors were generated by subcutaneous injections of 10⁷ cells in the posterior inferior flank of mouse. Animals were monitored and euthanized after 14 days. When necessary, inguinal lymph nodes were extracted for either flow cytometry analysis or cell sorting. Gross tumor mass was determined by weighing excised tissues in an analytical balance (Mettler Toledo) with necessary taring.

Mice reconstitution and immunizations

Rag1^−/− mice were retro-orbitally reconstituted with a mix of syngeneic wild-type CD8⁺ T cells (1 × 10⁶), wild-type B-cells (6 × 10⁶) and either Satb1⁺ or Satb1^−/− CD4⁺ T cells (2 × 10⁶). Immunopurified mouse cells were isolated by using the CD8⁺ (19853), B (19854) and CD4⁺ (19852) negative selection kits (StemCell Technologies), respectively. Unless mentioned otherwise, immunization experiments were performed by a subcutaneous (s.c.) injection of Ovalbumin (albumin from chicken egg white; Sigma, A5503), at 50 μg into each flank resuspended in Complete Freund’s Adjuvant (Sigma), resulting in 100 μg per mouse.

In studies investigating antigen cellular-specific responses, 1 × 10⁶ of either naïve OT-II CD4^CreSatb1^f/f or OT-II Satb1^f/f cells, were isolated with the Mouse Naïve CD4⁺ T Cell negative selection Kit (StemCell Technologies: 9765), and retro-orbitally injected in the recipient B6 CD45.1 congenic mice (The Jackson Labs.). Immediately after, mice were IP immunized with 4-Hydroxy-3-nitrophenylacetyl Ovalbumin (NP-OVAL, Biosearch Technologies) (50 μg/mouse) in Imject™ Alum Adjuvant (ThemoFisher Scientific). Five days after immunization, mice were euthanized for spleen isolation and flow cytometry analysis. In the case of immunizations with ovarian tumor cells, (20 × 10⁶) UPK10 cells were X-ray irradiated in PBS with a total dose of 40 Gy in a XRAD 160. Immediately after, cells were centrifuged, and IP injected in 0.3 ml PBS. Serums were isolated from whole blood after 14 days.

In vivo mouse antibodies for cells and cytokine depletion

B cells were depleted in vivo in Satb1^f/f mice by IP injections (200 μg/mouse) of anti-mouse B220 (RA3.3A1/6.1 (TIB-146), BioXCell) vs. rat IgG isotype control (200 μg/mouse) 2 days before tumor challenge and every other day, at days 0, 2, 4 and 6. CD8⁺ T cells were in vivo depleted in Satb1^f/f mice by IP injections (150 μg/mouse) of anti-mouse CD8α (YTS 169.4, BioXCell) vs. rat IgG2b (LTF-2) isotype control (150 μg/mouse) 2 days before tumor challenge and every two days, at days 0, 2, 4, 6 and 8. CXCL13 was blocked by IP injections (150 μg/mouse) of anti-mouse CXCL13 antibody (143614, R&D) vs. control rat IgG2A, every 3 days at days 3, 6 and 9 from the tumor challenge. All injections were done in 300 μl phosphate buffered saline (PBS).

Subcutaneous tumor implantation

Animals were prepared for surgery and anesthetized in according to the IACUC approved protocols. A 0.5 cm skin incision was made on the flank, 0.2-0.5 cm to the proximal femur of the animal. Freshly IP isolated tumor chunks (0.5x0.5 cm²) from a donor tumor bearing mouse were kept in PBS. Afterwards, a tumor piece per mouse flank was implanted subcutaneously, and the incision closed with stainless steel wound clips. A piece of tumor from the donor mouse was preserved as a control, for IHC studies, just before the s.c flank tumor implantation. Mice were regularly monitored and euthanized after 14 days of tumor implantation or when tumor size reached 2 cm².

Flow cytometry assays

Samples were prepared as single cell suspensions and resuspended in 1% Fetal Bovine Serum (FBS) in PBS prior to staining. Cells were stained with fluorescent conjugated antibodies in 1% FBS. The conjugated antibodies and probes used for flow cytometry are listed in the Key Resources Table. Studies involving intracellular staining were performed with the Foxp3 Transcription Factor Staining Buffer Set (eBioscience, ThermoFisher Scientific) following manufacturer’s instructions. In case of the mouse cytokines IL-21, CXCL13 and Light/Tnfsf14, total splenocytes from s.c OVA-vaccinated mice were stimulated with PMA (100 ng/mL), ionomycin (0.5 μg/mL) and brefeldin A (10 μg/mL) for 6 hours in complete medium prior to the staining.

Key Resources Table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Mouse Antibodies
anti-mouse B220 (RA3.3A1/6.1 (TIB-146))	BioXCell	RRID: AB_1107651 Cat#BE0067
anti-mouse CD8α (YTS 169.4)	BioXCell	RRID: AB_10950145 Cat#BP0117
BUV395 Hamster anti-mouse CD3e (145-2C11)	BD Biosciences	RRID: AB_2738278 Cat#563565 / 565992
BV711 rat anti-mouse CD45 (30-F11)	BD Biosciences	RRID: AB_2687455 Cat# 563709
Alexa Fluor® 700 rat anti-mouse CD44, (IM7)	BD Biosciences	RRID: AB_1727480 Cat# 560567
BUV737 anti-mouse CD4 (RM4-5)	BD Biosciences	RRID: AB_2870165 Cat# 612843 / 612844
PE Mouse anti-Bcl-6 (K112-91)	BD Biosciences	RRID: AB_2869985 Cat# 566979
APC rat anti-mouse IgG1 (A85-1)	BD Biosciences	RRID: AB_1645625 Cat# 560089
BV711 mouse anti-mouse CD45.1 (A20)	BD Biosciences	RRID: AB_2872214 Cat# 747742
PerCP/Cyanine5.5 anti-mouse/human GL7 Antigen (GL7)	Biolegend	RRID: AB_2562978 Cat# 144609 / 144610
Brilliant Violet 605 anti-mouse CD45 Antibody (30-F11)	Biolegend	RRID: AB_2562341 Cat# 103139 / 103155 / 103140
Brilliant Violet 510 anti-mouse CD62L (MEL-14)	Biolegend	RRID: AB_2561537 Cat# 104441
PE/Dazzle 594 anti-mouse CD279 (PD-1) (29F.1A12)	Biolegend	RRID: AB_2566005 Cat# 135127 / 135228
Brilliant Violet 605 anti-mouse CD45.2 Antibody (104)	Biolegend	RRID: AB_2563485 Cat# 109841
Brilliant Violet 421 anti-mouse CD185 (CXCR5) (L138D7)	Biolegend	RRID: AB_2562127 Cat# 145511 / 145512
Brilliant Violet 785 anti-human/mouse/rat CD278 (ICOS) (C398.4A)	Biolegend	RRID: AB_2629729 Cat# 313533 / 313534
Alexa Fluor 488 anti-mouse/rat/human FOXP3 Antibody (150D)	Biolegend	RRID: AB_439747 Cat# 320011 / 320012
Brilliant Violet 510 anti-mouse/human CD45R/B220 Antibody (RA3-6B2)	Biolegend	RRID: AB_2561394 Cat# 103247 / 103248
Brilliant Violet 605 anti-mouse CD95 (Fas) Antibody (SA367H8)	Biolegend	RRID: AB_2728202 Cat# 152612
APC anti-mouse CD25 Antibody (PC61)	Biolegend	RRID: AB_312860 Cat# 102011 / 102012
Brilliant Violet 785 anti-mouse I-A/I-E Antibody (M5/114.15.2)	Biolegend	RRID: AB_2565977 Cat# 107645
Brilliant Violet 605 anti-mouse CD11c Antibody (N418)	Biolegend	RRID: AB_11204262 Cat# 117333 / 117334
PE-Cyanine7 anti-mouse CD19 (1D3)	Tonbo Biosciences	RRID: AB_2621840 Cat# 60-0193
Anti-mouse IFN-gamma (XMG1.2)	Tonbo Biosciences	RRID: AB_2621810 Cat# 50-7311
Anti-mouse IL-4 (11B11)	Tonbo Biosciences	RRID: AB_2621465 Cat# 40-7041
Anti-Mouse CD3e (145-2C11)	Tonbo Biosciences	RRID: AB_2621659 Cat# 35-0031
Anti-mouse CD28 (37.51)	Tonbo Biosciences	RRID: AB_2621492 Cat# 70-0281
Rat anti-mouse CXCL13/BLC/BCA-1 MAb antibody (143614)	R&D Systems	RRID: AB_2086062 Cat#MAB470
rat IgG2A (Isotype) Alexa Fluor 647 (54447)	R&D Systems	Cat# IC006R
rat IgG2B (Isotype) Alexa Fluor 647 (141945)	R&D Systems	Cat# IC013R
Alexa Fluor 647 anti-mouse LIGHT/TNFSF14 (885310)	R&D Systems	Cat# FAB17942R
Alexa Fluor 647 Rat anti mouse CXCL13 (RM0205-19N1)	Novus Biologicals	Cat#NBP2-12223AF647
IL-21 Monoclonal Antibody (Clone FFA21), PE, eBioscience.	ThermoFisher Scientific	RRID: AB_1834466 Cat# 12-7211-82
anti-mouse CD23 (Invitrogen, SP23)	ThermoFisher Scientific	RRID: AB_10986348 Cat# MA5-14572
rabbit anti-mouse CD3 (SP7)	Abcam	RRID: AB_443425 Cat# ab16669
anti-mouse CD31	Abcam	RRID: AB_726362 Cat# ab28364
anti-mouse PNAd (MECA-79)	Millipore	Cat# MABF2050
rabbit anti-mouse CD19 (D4V4B)	Cell Signaling	RRID: AB_2800152 Cat# 90176
PE anti-mouse H-2Kb bound to SIINFEKL	Biolegend	RRID: AB_10895905 Cat#141604
Human Antibodies
BV421 mouse anti-Bcl-6 (K112-91)	BD Biosciences	RRID: AB_2738159 Cat# 563363
BV786 mouse anti-human CD45 (HI30)	BD Biosciences	RRID: AB_2716864 Cat# 563716
BB515 rat anti-human CXCR5 (CD185)(RF8B2)	BD Biosciences	RRID: AB_2738871 Cat# 564624 / 564625
BV421 Mouse IgG1 k Isotype Control (X40)	BD Biosciences	RRID: AB_11207319 Cat# 562438
BV650 Mouse anti-human CD279 (PD-1), (EH12.1)	BD Biosciences	RRID: AB_2738595 Cat# 564104
Alexa Fluor 647 mouse anti-SATB1 (14/SATB1)	BD Biosciences	RRID: AB_11153310 Cat# 562378
BUV395 mouse anti-human CD3 (SK7)	BD Biosciences	RRID: AB_2744382 Cat# 564000 / 564001
Brilliant Violet 785 anti-human CD4 (RPA-T4)	Biolegend	RRID: AB_2564381 Cat# 300553 / 300554
PE-Dazzle 594 anti-human/mouse/rat CD278 (ICOS), (C398.4A)	Biolegend	RRID: AB_2566129 Cat# 300553 / 300554
Brilliant Violet 711 anti-human CD25 (BC96)	Biolegend	RRID: AB_11219793 Cat# 300553 / 300554
PE Anti-Human CD25 (BC96)	Tonbo Biosciences	RRID: AB_2621758 Cat# 50-0259
Other Antibodies
Normal Rabbit IgG control isotype	Cell Signaling	RRID: AB_1031062 Cat# 2729
InVivoMAb polyclonal rat IgG	BioXCell	RRID: AB_1107795 Cat# BE0094
InVivoPlus rat IgG2b isotype control (LTF-2)	BioXCell	RRID: AB_2891360 Cat# BP0090
Rat IgG2A Isotype Control (54447)	R&D Systems	RRID: AB_357349 Cat# MAB006
ε-CD3 monoclonal antibody (SP7)	ThermoFisher Scientific	RRID: AB_10982026 Cat# MA5-14524
Recombinant anti-CD20 antibody [SP32]	Abcam	RRID: AB_1139386 Cat# ab64088
Recombinant anti-CD19 antibody [SP291] - C-terminal	Abcam	Cat# ab245235
Anti-Bcl6 antibody	Abcam	Cat# ab272859
Recombinant anti-CD21 antibody [EP3093]	Abcam	RRID: AB_1523292 Cat# ab75985
Recombinant anti-SATB1 antibody [EPR3951]	Abcam	RRID: AB_10862207 Cat# ab109122
Anti-Satb1 - C-terminal	Abcam	Cat# ab228772
Recombinant DNA
Plasmid: pLV-EF1a-IRES-Blast	Addgene	Cat# 85133
pcDNA3-deltaOVA	Addgene	Cat# 64595
VSV.G	Addgene	Cat# 14888
pCMV delta R8.2	Addgene	Cat# 12263
Bacterial and virus strains
One Shot TOP10 Chemically Competent E. coli	ThyermoFisher SCIENTIFIC	Cat# C404003
Chemicals, peptides, and recombinant proteins
Rainbow Fluorescent Particles, 1 peak (3.0-3.4 μm) Mid Range Intensity	BioLegend	Cat#422905
Zombie NIR™ Fixable Viability Kit	BioLegend	Cat#423106
4-Hydroxy-3-nitrophenylacetyl Ovalbumin (NP-OVAL)	Biosearch Technologies	N-5051-100
Dako antibody diluent (Agilent)	Carpenteria, CA	Cat#SO809
Ionomycin (calcium salt)	Cayman Chemical Co.	Cat# 11932
Phorbol 12-myristate 13-acetate (PMA)	Cayman Chemical Co.	Cat# 10008014
Brefeldin A	Cell Signaling	Cat# 9972S
Ficoll-Paque gradient	GE Healthcare Systems	Cat# 17-1449-02
EDTA pH 9.0 buffer	Invitrogen	Cat# 15575-038
Recombinant Human IL-2	PeproTech	Cat# 200-02
rhIL-10	PeproTech	Cat# 200-10
rhIL-12 p70	PeproTech	Cat# 200-12
rhIL-23	PeproTech	Cat# 200-23
Recombinant Human TGF-β1	R&D Systems	7754-BH-100/CF
Ovalbumin (albumin from chicken egg white)	Sigma	Cat# A5503
anti-CD3/CD28 mouse T-activator beads	ThermoFisher Scientific	Cat# 11456D
Cell Trace Violet	ThermoFisher Scientific	Cat# C34571
Dynabeads™ Human T-Activator CD3/CD28	ThermoFisher Scientific	Cat# 11161D
Foxp3 Transcription Factor Staining Buffer Set	ThermoFisher Scientific	Cat# 00-5523-00
Imject™ Alum Adjuvant	ThermoFisher Scientific	Cat# 77161
SYBR Green PCR Master Mix	ThermoFisher Scientific	Cat# 4344463
DEPC-treated Water	Thermo Scientific	Cat# R0601
Gel RED Nucleic Acid Stain	Biotium	Cat# 41002
Agarose LE, Quick Dissolve	Genesee Scientific	Cat# 20-102QD
Pharmco Products Ethyl Alcohol (200 Proof)	Fisher Scientific	Cat# 111000200
β-Mercaptoethanol, BME	Sigma	Cat#M6250-10ML
Bacto AGAR	BD	Cat# 240230
Difco LB Broth, Lennox	BD	Cat# 240110
S.O.C. Medium	ThermoFisher (Invitrogen™)	Cat#15544034
Cell Conditioning 1 (CC1)	Ventana products	Cat #950-124
Discovery EZ Prep wash solution	Ventana products	Cat #950-510
Buffered Formalde-Fresh 10% formalin	Fisher Chemical	Cat# SF93-20
Hematoxylin	Sigma Aldrich	Cat#H9627
Blasticidine S hydrochloride	Sigma Aldrich	Cat#15205
EcoRI	New England Biolabs	Cat# R0101S
HpaI	New England Biolabs	Cat# R0105S
JetPRIME	Life science research. PolyPlus.	Ref#101000027
Critical commercial assays
OPAL TM 7-Color Automation IHC kit	AKOYA Biosciences	Cat# NEL821001KT
Mouse anti-Ovalbumin Quantitative ELISA Kit	Alpha Diagnostic International	Cat# 600-105-OGG
BLC/CXCL13 Mouse ELISA Kit	ThermoFisher Invitrogen	Cat# EMCXCL13
RNeasy Plus Micro Kit	QIAGEN	Cat#74034
EasySep™ Mouse CD8+ T Cell Isolation Kit	StemCell Technologies	Cat# 19853
EasySep™ Mouse B Cell Isolation Kit	StemCell Technologies	at# 19854
EasySep™ Mouse CD4+ T Cell Isolation Kit	StemCell Technologies	Cat# 19852
Mouse Naïve CD4+ T Cell negative selection Kit	StemCell Technologies	Cat# 9765
Mouse T Cell negative selection kit	StemCell Technologies	Cat# 19851
Human Naïve CD4+ T Cell isolation kit	StemCell Technologies	Cat# 17555
High-Capacity Reverse Transcription kit, Applied Biosystems	ThermoFisher Scientific	Cat# 4368814
Ventana ChromoMap kit	Ventana products	Cat# 760-159
Experimental models: Cell lines
UPK10 ovarian carcinosarcoma	PMID: 22351930 PMID: 25533336	N/A
UPK10-OVAlo ovarian carcinosarcoma	This paper	N/A
Brpkp110 breast-p53-KRas-p110alpha	PMID: 27803104	N/A
HEK293T	ATCC	Cat# CRL-3216
Biological Samples
Blood buffy coat units	LifeSouth (Gainesville, FL, USA)	N/A
Experimental models: Organisms/strains
Mouse: C57BL/6	The Jackson Laboratory/Charles River	Strain: 000664/027
Mouse: B6 Cd45.1 (B6.SJL-Ptprca)	The Jackson Laboratory	Strain: 002014
Mouse: B6 Rag1 KO (B6.129S7-Rag1tm1Mom/J)	The Jackson Laboratory	Strain: 002216
Mouse: C57BL/6 Tg(TcraTcrb)425Cbn/Crl OT-II	Charles River	Strain: 643
Mouse: B6 CD4CreSatb1f/f	PMID: 26876172 PMID: 28099864	N/A
Mouse: B6 OT-II+- CD4CreSatb1f/f	This paper	N/A
Mouse: B6 Satb1f/fFoxp3GFP	This paper	N/A
Mouse: B6 CD4CreSatb1f/fFoxp3GFP	This paper	N/A
Oligonucleotides
Primer sequence (5′→3′) mouse Il-21 F: AATCAAGCCCCCAAGGGCCA, R: AGCTGGGCCACGAGGTCAAT	Integrated DNA Technologies, IDT	N/A
Primer sequence (5′→3′) mouse Cxcl13 F: AAGTTACGCCCCCTGGGAATGGCT, R: ACTCACTGGAGCTTGGGGAGTTG	Integrated DNA Technologies, IDT	N/A
Primer sequence (5′→3′) mouse Light/Tnfsf14 F: ATCTCACCAGGCCAACCCAGCA, R: ACGCCGCTCAGCTGCACTTT	Integrated DNA Technologies, IDT	N/A
Primer sequence (5′→3′) Gapdh (mRNA normalization) F: CCTGCACCACCAACTGCTTA, R: AGTGATGGCATG GACTGTGGT	Integrated DNA Technologies, IDT	N/A
Primer sequence (5′→3′) Icos promoter F: TAC ATC ACC GGG TAC TTG CCA T; R: GCT CAA AAG TGT CAG TGA AGT CGT	Integrated DNA Technologies, IDT	N/A
Primer sequence (5′→3′) CSN1 region F: ACA AGG CCA AAC AAC AAT GAT AGA CA, R: TCA ACA TGA GGA ACA GTG CAG GA.	Integrated DNA Technologies, IDT	N/A
Primer sequence (5′→3′) CSN2 region F:GAC AAC AGG GCC CAG ATG TAG AC; R: GGC GTT CCT GTT TGA CTG TTT CT.	Integrated DNA Technologies, IDT	N/A
Software and algorithms
Prism, v8	GraphPad	N/A
FlowJo, v10.8	BD Biosciences	N/A
Aperio Image Scope, v12.4	Leica Biosystems	N/A
Definiens Tissue Studio, v4.4.2	Definiens Inc.	N/A

Cell proliferation assays were performed by resuspending cells in 1 μM Cell Trace Violet in PBS for 15 min at 37 °C (ThermoFisher) and washed twice in complete medium. Cell viability was assessed through staining with the Zombie NIR dye (Biolegend) for 15 min in PBS. The reaction was stopped by washing with R10.

Samples were run either on a BD LSRII for flow analysis or sorted using a BD FACS Aria. All sorted populations were confirmed to be >98% pure through post-sort analysis. For assays evaluating protein expression levels using MFI, we used the Rainbow Fluorescent Particles, (Biolegend) in order to normalize voltage variance between measurements. Data were analyzed using FlowJo software v10.7.

Immunosuppression assay

Splenic CD3⁺ T cells were immunopurified from Satb1^f/f mice (Mouse T Cell negative selection kit, 19851; StemCell Technologies) and labeled with the proliferation tracker Cell Trace Violet as indicated. T cell proliferation was stimulated by adding mouse specific anti-CD3/CD28 dynabeads (ThermoFisher) at a 1:1 T cell to bead ratio according to the manufacturer’s protocol. CD3⁺CD4⁺FoxP3^GFP natural Treg cells (nTreg cells) were sorted from both CD4^CreSatb1^f/fFoxP3^GFP and Satb1^f/fFoxP3^GFP control littermates. Immunopurified CD3⁺ T cells (2×10⁵) were subsequently co-cultured at 1:2, 1:4, 1:8, and 1:16 T cell to nTreg cells ratios and incubated for 3 days prior to flow cytometry analysis.

Mouse inducible Treg cell differentiation

CD3⁺CD4⁺Foxp3⁻CD62L⁺CD44⁻ naïve sorted splenocytes from CD4^CreSatb1^f/fFoxP3^GFP mice and Satb1^f/fFoxP3^GFP control littermates were plated with anti-CD3/CD28 mouse T-activator beads (ThermoFisher) for 3 days in R10 under the following skewing conditions: IL-2, 100 U/mL (PeproTech), recombinant mouse TGF- β1, 10 ng/mL (R&D Systems), anti-mouse IFN-gamma (XMG1.2), and anti-mouse IL-4 (11B11), both (5 μg/mL) from Tonbo Biosciences.

Cytokine stimulation of naïve human T cells

Naïve human cells were isolated by using Human Naïve CD4⁺ T Cell isolation kit (Stemcell Technologies, 17555) from peripheral blood mononuclear cells (PBMCs) after Ficoll-Paque gradient (GE Healthcare Systems). Naïve human CD4⁺ T cells were incubated with Dynabeads™ Human T-Activator CD3/CD28 (ThermoFisher, 11161D) for 3 days in R10 under the following skewing conditions: a) rhIL-23 (25 ng/ml, PeproTech); b) rhIL-23 (25 ng/ml, PeproTech) plus rhIL-10 (10 ng/ml PeproTech); c) rhIL-12 p70 (1 ng/ml, PeproTech) and d) rhIL-12 p70 (1 ng/ml, PeproTech) plus rhIL-10 (10 ng/ml PeproTech). The former conditions were cultured in the presence/absence of hTGF-β (5 ng/ml) (R&D Systems).

Multiplex immunofluorescence

Formalin-fixed and paraffin-embedded (FFPE) tissue samples were immunostained using the AKOYA Biosciences OPAL TM 7-Color Automation IHC kit (Waltham, MA) on the BOND RX autostainer (Leica Biosystems, Vista, CA). The OPAL 7-color kit uses tyramide signal amplification (TSA)-conjugated to individual fluorophores to detect various targets within the multiplex assay. Sections were baked at 65°C for one hour then transferred to the BOND RX (Leica Biosystems). All subsequent steps (ex., deparaffinization, antigen retrieval) were performed using an automated OPAL IHC procedure (AKOYA). OPAL staining of each antigen occurred as follows: heat induced epitope retrieval (HIER) was achieved with EDTA pH 9.0 buffer for 20 min at 95°C before the slides were blocked with AKOYA blocking buffer for 10 min. Then slides were incubated with primary antibody, CD20 (AbCAM, SP291, 1:50, dye 520) at RT for 60 min followed by OPAL RbHRP polymer and one of the OPAL fluorophores during the final TSA step. Individual antibody complexes are stripped after each round of antigen detection. This was repeated four more times using the following antibodies, ε-CD3 (ThermoFisher, SP7, HIER- EDTA pH 9.0, 1:100, dye540), CD19 (AbCAM, SP291, HIER- EDTA pH 9.0, 1:200, dye620), BCL6 (AbCAM, Lot# ab272859, HIER- EDTA pH 9.0, 1:600, dye550), CD21 (AbCAM, EP3093, HIER- EDTA pH 9.0, 1:300, dye690); CD4 (Cell Signaling Technology, D7D2Z, 1:100), BCA1 (AbCAM , ab199043, EPR19259-147, 1:50) and CD34 (AbCAM , ab81289 EP373Y, 1:300).

After the final stripping step, DAPI counterstain was applied to the multiplexed slide and later removed from BOND RX for coverslipping with ProLong Diamond Antifade Mountant (ThermoFisher Scientific). Autofluorescence slides (negative control) were included, which use primary and secondary antibodies omitting the OPAL fluors and DAPI. All slides were scanned with the Vectra®3 Automated Quantitative Pathology Imaging System and the data from the multispectral camera were accessed by the imaging InForm software. Respective positive-control tissue adjacent sections were stained with isotype control antibodies to rule out false-positive staining. Positivity thresholds were determined per marker based on published nuclear or cytoplasmic staining patterns.

Analysis of TCGA data

Molecular and clinical data from TCGA for ovarian serous cystadenocarcinoma (designated OV) were downloaded from the cBio Cancer Genomics Portal (http://www.cbioportal.org/), Broad Firehose website (https://gdac.broadinstitute.org/) and Genomic Data Commons Data Portal (https://portal.gdc.cancer.gov/). A total of 428 patients with matched clinical information and tumor RNA-seq data was used in this study. Raw RNA-seq reads were aligned to the GRCh37 human transcriptome using STAR24 (v.2.5.3a). Uniquely aligned reads were counted against Gencode v.19 using htseq-count31 (v.0.6.1) and then normalized using DESeq2 taking into account batches and RNA composition bias. All statistical analysis and visualization were performed on normalized count.

Immunohistochemistry, TLS definition and quantification.

In all cases, the whole mouse IP tumor contents were processed for TLS analysis. Tumor tissues were fixed in 10% formalin, embedded in paraffin, and serially sectioned for H&E and immunohistochemistry staining. All tumor sections slides were systematically stained using a Ventana Discovery XT automated system (Ventana Medical Systems, Tucson, AZ) as indicated by the manufacturer. Briefly, slides were deparaffinized with EZ Prep solution (Ventana). Heat-induced antigen retrieval method was used. Either primary rabbit anti-mouse CD3 (ab16669, Abcam, Cambridge, MA) was used at 1:200 dilution in Dako antibody diluent (Carpenteria, CA) or rabbit anti-mouse CD19 (90176, Cell Signaling, Danvers, MA) at 1:1000 dilution, and incubated for 30 min. For the case of the anti-mouse PNAd (MABF2050, Millipore, Billerica, MA) we used at 1:50 dilution in Dako antibody diluent and was incubated for 60 min. For the anti-mouse CD23 (Invitrogen, SP23) was used at 1:25 dilution. Retrieval was Cell Conditioning 1 (Ventana product) and incubation for 3 hours. The anti-mouse CD31 (Abcam ab28364) was used at 1:100. Retrieval was cell conditioning 1 and incubation was 32 minutes. The Ventana OmniMap anti-rabbit secondary antibody was used for 15 min. The detection system used was the Ventana ChromoMap kit and slides were counterstained with Hematoxylin. Finally, stained slides were dehydrated and coverslipped.

Slides were scanned using the Aperio™ ScanScope AT2 (Leica Biosystems, Vista, CA) with a 20x/0.75NA objective lens and an optical doubler to bring the final magnification to 40X. These images and meta-data were then imported into the Definiens Tissue Studio (Munich, Germany) software suite for segmentation (Payne et al., 2020). In Tissue Studio a machine learning algorithm was used to segment each tissue section image into positive CD19, CD3 areas and negative areas. The results of the segmentation were further refined using proximity classifiers, meaning positive clusters within 20 μm were merged. This refinement step was necessary to help to get the total positive CD19, CD3 area per TLS. We defined TLS as >0.1 mm² conglomerates of CD19⁺B cells adjacent to accumulations of CD3⁺ T cells, clearly identified inside tumor masses and co-stained for peripheral node addressin (PNAd), with a central core of T cells. Likewise, large TLS were defined as regions >2 mm². TLS areas were calculated by summing up the CD3 and CD19 positively stained zones in proximity (CD19⁺ plus CD3⁺ contents).

In addition to the TLS quantification, an overall estimate of the tumors’ density infiltration by both CD3⁺ T cells and CD19⁺ B cells outside the TLS, in tumors developed in mice with Satb1-competent vs. Satb1^−/− T cells, was performed using an Aperio Positive Pixel Count® v9.0 algorithm with the following thresholds: [Hue Value =.1; Hue Width =.5; Color Saturation Threshold =0.04; IWP(High) = 220; Iwp(Low)=Ip(High) = 175; Ip(low) =Isp(High) =100 Isp(Low) =0] to segment positive staining of various intensities. The algorithm was applied to the entire digital image to determine the percentage of positive biomarker staining by applicable area. The TLS area was then subtracted from total biomarker area in order to determine the amount of positive area outside of the TLS regions.

Quantitative real time PCR

RNA was isolated (RNeasy Plus Micro Kit, QIAGEN) from CD3⁺CD4⁺CD62L⁻CD44⁺PD-1^hiICOS⁺CXCR5⁺ Tfh and CD3⁺CD4⁺CD62L⁺CD44⁻ naive cells immediately after sorting from total splenocytes and inguinal lymph nodes from s.c OVA-vaccinated CD4^CreSatb1^f/f mice and Satb1^f/f control littermates at day 14. Reverse transcription was carried out using High-Capacity Reverse Transcription kit (Applied Biosystems). SYBR Green PCR Master Mix (Applied Biosystems) was used in a 7900 HT Fast Real-Time PCR System Software analysis (2.4), (Applied Biosystems). The following primer sequences were used (5′→3′): mouse Il-21 F: AATCAAGCCCCCAAGGGCCA, R: AGCTGGGCCACGAGGTCAAT; mouse Cxcl13 F: AAGTTACGCCCCCTGGGAATGGCT, R: ACTCACTGGAGCTTGGGGAGTTG; mouse Light/Tnfsf14 F: ATCTCACCAGGCCAACCCAGCA, R: ACGCCGCTCAGCTGCACTTT; and Gapdh (mRNA normalization) F: CCTGCACCACCAACTGCTTA, R: AGTGATGGCATGGACTGTGGT. The average of 3 independent replicas was calculated using the ΔΔ threshold cycle (Ct) method and was normalized to the endogenous reference control gene GAPDH.

Chromatin immunoprecipitation and Chip-PCR

Immunopurified CD4⁺ T cells from spleens (19852, EasySep) of naïve Satb1^f/f mice were stimulated in vitro with immobilized anti-mouse CD3e (5 μg/mL) (145-2C11) (Tonbo Biosciences) and soluble anti-mouse CD28 (1 μg/mL) (37.51) (Tonbo Biosciences) for 30 hours. ChIP-PCR assays were performed as we previously reported (Stephen et al., 2017; Tesone et al., 2016), following standard methods with a chromatin shearing protocol of three cycles of 10 minutes each one, at high intensity (320W), consisting each minute as follows: (ON 30 seconds, OFF 30 seconds), followed by 5 minutes OFF between cycles, set up in a Diagenode Bioruptor 300. Shearing chromatin fragments of 100 – 300 bp were immunoprecipitated overnight with the anti-Satb1 - C-terminal (Abcam, ab228772) and the rabbit IgG as a control isotype (Cell Signaling; 2729). The percent input method, in this case 2.5 % of starting chromatin, was used. The following primers were used for the quantification of the Icos promoter: (5′→3′): F: TAC ATC ACC GGG TAC TTG CCA T; R: GCT CAA AAG TGT CAG TGA AGT CGT. For the case of the CNS1 and CNS2 regions, we used in vivo differentiated iTreg cells (kept for 6 days under skewing conditions) from immunopurified naïve CD4⁺ sorted cells (StemCell Technologies: 9765). Likewise, samples were immunoprecipitated overnight with the anti-Satb1 - C-terminal (Abcam, ab109122). The following primers were used for the quantification of the CSN1 region: (5′→3′): F: ACA AGG CCA AAC AAC AAT GAT AGA CA, R: TCA ACA TGA GGA ACA GTG CAG GA. Likewise, the following primers were used for the quantification of the CSN2 region: (5′→3′): F:GAC AAC AGG GCC CAG ATG TAG AC; R: GGC GTT CCT GTT TGA CTG TTT CT. Input and immunoprecipitated DNA were analyzed with a SYBR Green PCR Master Mix (Applied Biosystems) in a 7900 HT Fast Real-Time PCR System Software analysis (2.4), (Applied Biosystems).

Generation of the UPK10-OVA^low cell line

The lentiviral constitutive expression vector pLV-EF1a- OVAdeltaN-2A-IRES-blast was created by cloning chicken ovalbumin lacking the sequence coding for amino acids 1-49 from the vector pcDNA3-deltaOVA (Addgene 64595) into the pLV-EF1a IRES-blast vector backbone (Addgene #85133) using 5′ EcoRI and 3′ HpaI restriction sites by using standard techniques. UPK10 cells were transduced to express OVA using the packing cell line HEK293T and the packaging plasmid, VSVG (Addgene #14888) and Delta 8.2 (Addgene # 12263), (following the JetPRIME reagent protocol). Two days after transduction, UPK10 cells expressing OVA were selected under Blasticidin (6ug/mL). UPK10-OVA^low expressing cells were sorted after staining with the PE anti-mouse H-2Kb bound to SIINFEKL. Sorted fractions were further expanded in R10 supplemented with Blasticidin (6ug/mL).

Enzyme-linked immunosorbent assays

Quantification of total IgG and CXCL13 from serum of subcutaneous OVA-vaccinated CD4^CreSatb1^f/f mice and Satb1^f/f control littermates, harvested at day 14, was done according to manufacturer protocol of the following kits: a) mouse anti-Ovalbumin Quantitative ELISA Kit (Alpha Diagnostic International, 600-105-OGG); b) BLC/CXCL13 Mouse ELISA Kit (ThermoFisher Scientist EMCXCL13).

Statistics analysis

Unless mentioned otherwise, all data are presented as median. Unpaired two-tailed t-test was used for calculating differences between means between two experimental groups. One-way ANOVA with Tukey’s post hoc test or two-way ANOVA (mixed model) followed by Sidak’s test, were used for calculating differences between means using multiple comparisons between groups. Analyses were carried out in Graph Pad Prism (v.8.0). The number of replicates per experiment is indicated in the legends. A significance threshold (0.05) for p values was used. Unless noted otherwise, all experiments were repeated at least twice and with similar results.

HIGHLIGHTS.

• Satb1^−/− CD4⁺ T cells show heightened antigen-specific Tfh cell differentiation.

• SATB1-mediated suppression of Icos expression is required for Tfr cell differentiation.

• CD4^CreSatb1^f/f mice spontaneously develop intra-tumoral tertiary lymphoid structures.

• Tfh cell transfer elicits intra-tumoral TLS assembly and reduces tumor growth.