Tumors that metastasize in the peritoneal cavity typically end up in the omental adipose tissue, a particularly immune-suppressive environment that includes specialized adipose-resident regulatory T cells (Treg). Tregs rapidly accumulate in the omentum after tumor implantation and potently suppress antitumor immunity. However, it is unclear whether these Tregs are recruited from the circulation or derived from preexisting adipose-resident Tregs by clonal expansion. Here we show that Tregs in tumor-bearing omenta predominantly have thymus-derived characteristics. Moreover, naïve tumor antigen-specific CD4+ T cells fail to differentiate into Tregs in tumor-bearing omenta. In fact, Tregs derived from the pretumor repertoire are sufficient to suppress antitumor immunity and promote tumor growth. However, tumor implantation in the omentum does not promote Treg clonal expansion, but instead leads to increased clonal diversity. Parabiosis experiments show that despite tissue-resident (noncirculating) characteristics of omental Tregs in naïve mice, tumor implantation promotes a rapid influx of circulating Tregs, many of which come from the spleen. Finally, we show that newly recruited Tregs rapidly acquire characteristics of adipose-resident Tregs in tumor-bearing omenta. These data demonstrate that most Tregs in omental tumors are recruited from the circulation and adapt to their environment by altering their homing, transcriptional, and metabolic properties.

The omentum is a visceral adipose tissue that connects the spleen, stomach, and pancreas and contains leukocyte-rich, fat-associated lymphoid clusters, also known as milky spots, that filter peritoneal fluid and support immune responses to peritoneal antigens (1–3). Despite its immune activity, the omentum is not protected from tumor colonization (4), and is frequently invaded by peritoneal metastases, particularly advanced-stage ovarian cancer (5, 6). Macrophages and neutrophils in the omentum contribute to tumor colonization and growth (7, 8), as do immune suppressive Tregs, which are linked to poor clinical outcomes (9, 10).

Tissue-resident regulatory T cells (Treg) are found in a variety of nonlymphoid tissues, including muscle, colon, and lung (11–13). As their name implies, tissue-resident Tregs poorly recirculate (14), and are specialized to maintain local tissue homeostasis and repair (15). In the omentum, tissue-resident Tregs suppress adipose inflammation and help to maintain metabolic homeostasis (16). Adipose-resident Tregs rely on the transcription factor PPARγ (17), the alarmin IL33 (18) and interactions with local antigen-presenting cells (14) for their self-renewing and suppressive activities.

Tumor implantation alters Treg activities, including their recirculation patterns. For example, the T-cell receptor (TCR) repertoire of intratumoral Tregs overlaps with that of circulating Tregs, suggesting a role for recruitment in Treg accumulation (19). Moreover, the chemokine receptors CCR4 (20), CCR10 (21), and CXCR4 (22) each help recruit Tregs to various tumor types. In addition, intratumoral Tregs often overexpress Ccr1, Ccr8, and Cxcr6 compared with their splenic counterparts (23), again suggesting a role in recruitment or tissue retention. Intratumoral Tregs may also be derived by the conversion of conventional CD4+ T cells to FoxP3-expressing Tregs in the tumor microenvironment (24, 25). This conversion is best characterized in the intestine, wherein antigen presentation in the context of TGFβ and retinoic acid promotes Treg differentiation (26, 27). Given the activities of TGFβ and retinoid acid in the omentum (28, 29), a similar process might occur during tumor outgrowth in this tissue.

Here we tested whether Treg accumulation in tumor-bearing omenta involved the recruitment of circulating Tregs, the local expansion of adipose-resident Tregs or the de novo generation of Tregs from conventional CD4+ T cells. We found that tumor implantation in the omentum promoted the accumulation of a broadly diverse, rather than clonally expanded, repertoire of thymus-derived Tregs. Parabiosis experiments showed that Treg accumulation occurred via recruitment of preexisting Tregs from systemic sites like the spleen. Newly recruited Tregs rapidly acquired characteristics of adipose-resident Tregs and potently suppressed local antitumor immunity. Collectively, these data suggest that tumor-associated Tregs in the omentum are primarily obtained from the circulation and rapidly adapt to their new environment.

Mice and tumor administration

C57BL/6J (B6), B6.SJL-PtprcaPepcb/BoyJ (CD45.1), B6.Cg-Tg(TcraTcrb)425Cbn/J (OT-II), B6. Tg(CAG-KikGR)75Hadj/J (KikGR), B6.129S4-Il2ratm1Dw/J (CD25KO), and B6.129P2-Tcrβtm1MomTcrδtm1Mom (TCRβδ−/−) mice were originally purchased from Jackson Laboratories. B6.129S6-Foxp3tm1DTR (FoxP3-GFP-DTR) mice were originally obtained from Dr. Alexander Rudensky (Memorial Sloan Kettering Cancer Center, New York, NY). 10BiT-FoxP3-GFP mice were originally obtained from the University of Alabama at Birmingham (UAB, Birmingham, AL) gnotobiotic facility and conventionalized. Male CD45.1 mice were crossed to female FoxP3-GFP-DTR mice to generate heterozygous mice that expressed both CD45.1 and CD45.2. All the mice were bred in the UAB vivarium. At 8 to 12 weeks of age, mice were intraperitoneally injected with 3 × 106 EG7.1.15 or 107 ID8-ovalbumin (OVA) tumor cells. The humane endpoints for the two models were day 20 and week 16, respectively. All animal procedures were approved by the UAB Institutional Animal Care and Use Committee.

Cell culture

EG7.1.15 cells were originally generated by Dr. Steve Schoenberger (La Jolla Institute for Allergy and Immunology) and obtained from Dr. Laura Haynes (Trudeau Institute, Saranac, NY). Cells were maintained in RPMI1640 (Lonza, 12-115F) supplemented with 10% FBS (Peak serum, PS-FB1), 50 μmol/L 2-Mercaptoethanol (Gibco, 21985–023), 0.024 mmol/L NaHCO3 (Cellgro, 23-035-CI), 600 μg/mL G418 (Gibco, 11811-031), 10 μg/mL kanamycin (Gibco, 11815-024), 200 µg/mL penicillin and streptomycin (Corning, 30-002-CI). Stocks of EG7.1.15 cells were cultured seven passages before freezing. Cells were thawed and passaged once before injection. Stocks of EG7.1.15 cells tested negative for Mycoplasma by PCR 9-8-09. ID8 ovarian carcinoma cells were originally obtained from Dr. Yancey Gillespie in 2016 (University of Alabama at Birmingham, Birmingham, AL), and maintained in RPMI1640 supplemented with 10% FBS and 200 µg/mL penicillin and streptomycin. Cells were transfected with 500 ng of pCI-neo-ovalbumin (Addgene, 25098) and 1.5 μL Lipofectamine 2000 (Invitrogen, 11668-019), plated in the aforementioned culture media supplemented with 400 μg/mL G418, and screened by immunofluorescence (Nikon TiElapse) for OVA-expressing clones. Positive clones (ID8-OVA) were then maintained in the media with 200 μg/mL G418. Stocks of ID8-OVA cells were cultured 10 passages before freezing. Cells were thawed and passaged twice before injection. Stocks of ID8-OVA cells tested negative for Mycoplasma (Charles River Research Animal Diagnostic Services, 12-17-14).

Cell isolation

The omentum was removed from mice after cervical dislocation, and weighed using an AL54 analytic balance (Mettler Toledo). Following mechanical disruption using scissors, tissue fragments were incubated in 2 mL of Iscove's Modified Dulbecco's Medium (Corning, 10-016-CV) supplemented with 3.5% BSA fatty acid free (Sigma, A7030), 1 mg/mL collagenase (type XI, Sigma, C7657), and 120 units/mL DNase I (type IV, Sigma, DN25) at 200 rpm (shaking) 37°C for 30 minutes. Cell suspensions were then filtered through 70 µmol/L nylon cell strainers (Corning, 352350), and larger pieces were mechanically dissociated on a wire mesh screen. The colon was opened longitudinally, washed with Hank's Balanced Salt Solution (HBSS; Gibco, 14175-095) containing 15 mmol/L HEPES (Corning, 25-060-CI), and cut into small pieces. The tissue was then incubated in 15 mL of predigestion buffer [HBSS supplemented with 2% FBS, 15 mmol/L HEPES, and 5 mmol/L EDTA (Invitrogen, 15575-038)] at 250 rpm (shaking) 37°C for 15 minutes, vigorously shaken for 30 seconds by hand and passed through a 100 µmol/L nylon mesh. The flow-through was discarded, and the remaining pieces were transferred to new predigestion buffer for incubation. This process was repeated for three times to remove the epithelial fraction. The pieces were then washed with RPMI1640 supplemented with 10% FBS, 10 mmol/L HEPES, 1 mmol/L Sodium Pyruvate (HyClone, SH30239.01), and 2-Mercaptoethanol (1×, Gibco) to remove EDTA, and digested in RPMI1640 supplemented with 5% FBS, 80 µg/mL collagenase (type VIII, Sigma, C2139), 30 μg /mL DNase I (type IV, Sigma), 10 mmol/L HEPES, 1 mmol/L Sodium Pyruvate, and 5 mmol/L CaCl2 at 350 rpm (shaking) 37°C for 35 minutes. The digested samples were filtered through 100 µmol/L (Corning, 352360) and 70 µmol/L nylon cell strainers, pelleted, and resuspended in 40% HBSS-adjusted Percoll (Sigma, P4937). Finally, cell suspensions were overlaid onto 75% HBSS-adjusted Percoll and centrifuged to enrich leukocytes. The spleen as well as mediastinal and inguinal lymph nodes were mechanically dissociated on a wire mesh screen. Red blood cells (RBC) were removed by osmotic lysis in ACK buffer (150 mmol/L NH4Cl, 10 mmol/L KHCO3, and 0.1 mmol/L Na2EDTA). Cells were resuspended in PBS supplemented with 2% bovine calf serum (FCS, HyClone, SH30073.03) and 2 mmol/L EDTA.

Flow cytometry

A list of antibodies, fluorophores, and clones used is provided in Supplementary Table S1. Cell surface staining was performed by incubation of antibodies in PBS supplemented with 10 µg/mL FcBlock (2.4G2, BioXCell, BE0307), 2% FCS, and 2 mmol/L EDTA for 25 minutes on ice. FoxP3 stain was performed using eBioscience Mouse Regulatory T Cell Staining Kit (Thermo Fisher Scientific, 00-5523-00) according to the manufacturer's instructions. For cytokine stain, cells were resuspended in RPMI supplemented with 10% FBS, nonessential amino acid solution (1×, Corning, 25-025-CI), 1 mmol/L sodium pyruvate, 10 mmol/L HEPES, 2-Mercaptoethanol (1×), 200 µg/mL penicillin and streptomycin, and restimulated with 5 ng/mL phorbol 12-myristate 13-acetate (PMA, Sigma, P8139), 65 ng/mL calcium ionophore A23187 (Thermo Fisher Scientific, A1493), and 10 µg/mL brefeldin A (Sigma, B7651) for 4 hours at 37°C. eBioscience Mouse Regulatory T Cell Staining Kit or BD Biosciences Cytofix/Cytoperm Kit (BD Biosciences, 554714) was then used to measure intracellular cytokines according to the manufacturer's instructions. For CellTrace Violet (CTV, Invitrogen, C34557) labeling, cells were resuspended at 7 × 106 cells/mL in PBS and incubated with CTV for 15 minutes at 37°C. Following the labeling, cells were washed with 10% FBS. Flow cytometry was performed on the BD FACSCanto II or FACSymphony system (BD Biosciences). Cell sorting was performed on the BD FACSAria II system (BD Biosciences). Data were analyzed using FlowJo (v9.9.6 and v10.8.1, BD Biosciences).

Adoptive transfer

Splenocytes were obtained from naïve OT-II CD45.1 mice, and CD4+ T cells were magnetically enriched using CD4+ T Cell Isolation Kit (Miltenyi Biotec, 130-117-043) according to the manufacturer's instructions. CD25CD44lo OT-II CD4+ T cells were purified by FACS, and labeled with CTV. A total of 106 cells were intraperitoneally transferred into naïve or tumor-bearing B6 mice. For polyclonal CD4+ T-cell transfer, naïve FoxP3-GFP-DTR mice were sacrificed and the spleens were excised. Following osmotic lysis and magnetic enrichment, CD25FoxP3(GFP)CD44loCD4+ T cells were sorted and labeled with CTV. A total of 2 × 106 cells were intraperitoneally transferred into naïve or tumor-bearing CD45.1 mice.

For adoptive transfer of polyclonal CD4+ T cells with and without Tregs, FoxP3(GFP)CD44loCD4+ T cells were sorted from the spleens of naïve male CD45.1+CD45.2+FoxP3-GFP-DTR mice, and FoxP3(GFP)+CD4+ Tregs were sorted from the spleens of naïve male CD45.1CD45.2+10BiT-FoxP3-GFP mice. A total of 106 CD4+ T cells were transferred intravenously alone, or with 5 × 105 Tregs into TCRβδ−/− mice.

For CD4+ T-cell recruitment assay, total CD4+ T cells were obtained by magnetic enrichment from the spleens of naïve B6 mice. Following CTV labeling, 6 × 106 CD4+ T cells were intraveously transferred into naïve or tumor-bearing CD45.1 mice.

Bone marrow chimeras

Naïve CD45.1 mice were irradiated with two doses of 425 Rads 5 hours apart using an X-ray source (Precision X-Ray). Bone marrow (BM) cells were harvested from the femur and tibia of 5-week-old B6 and CD25KO donor mice, and RBCs were removed by osmotic lysis in ACK buffer. A total of 5 × 106 BM cells of either genotype were intravenously injected into recipient mice, which were then allowed to reconstitute for at least 11 weeks before tumor injection.

Mouse surgeries

For all the mouse surgeries, the flank of mice was shaved 2 to 3 days prior to the operation. Isoflurane was used as anesthesia, and ethanol swabs and Povidone-iodine were used for disinfection. Body temperature of animals was maintained using a heating pad and lamp during surgery. Parabiosis surgery was performed as described previously (30). Briefly, a pair of weight-matched female mice was cohoused for at least 20 days. Longitudinal incisions were made in the skins from the forelimb to the hindlimb, and sutures were run through the olecranon and knee joints to bring the body walls of the partners in close contact. Interrupted sutures and wound clips were then used to stabilize the dorsal and ventral skins.

For photoconversion of splenocytes in KikGR mice, incisions were first made in the skin on the left flank to make the spleen visible, and a smaller incision was then made in the abdominal wall to expose the organ. Sterile cotton swabs rinsed with saline solution were used to gently place the spleen onto sterile aluminum foil, which was used to cover the skin and peritoneal cavity. The spleen was then exposed to violet light (Dymax BlueWave QX4) at an intensity of 95 mw/cm2 for 3 × 200 s, and sterile saline solution was given between the exposure sessions to keep the organ moisturized. Following the photoconversion, the spleen was placed back using cotton swabs, and the abdominal wall was closed with sutures. Interrupted sutures and wound clips were used to close the outer skin.

For splenectomy, incisions were first made in the skin on the left flank to make the spleen visible, and a smaller incision was then made in the abdominal wall to expose the organ. Forceps were used to place the spleen on a medical gauze pad, and the vessels to the splenic hilum were divided with a cautery. Following the removal of the spleen, the abdominal wall was closed with sutures. Interrupted sutures and wound clips were used to close the outer skin. Sham-operated mice underwent the same procedure except for spleen removal.

TCR repertoire library preparation and sequencing

CD25+FoxP3(GFP)+CD4+ Tregs were sorted into RL buffer (Norgen, 51800) from the omenta and spleens of 4 naïve and 4 day 15 EG7.1.15 tumor-bearing FoxP3-GFP-DTR mice. Sorted cells were not pooled, thus each sample represents a Treg population from an organ of an individual mouse (Supplementary Table S2). Total RNA was purified using Single-Cell RNA purification kit (Norgen) according to the manufacturer's instructions. TCR library was prepared using iRepertoire arm-PCR platform (iRepertoire) as described previously (31). Briefly, reverse transcription of RNA was conducted with OneStep RT-PCR kit (Qiagen, 210212) according to the manufacturer's protocol. The PCR product was purified using Ampure XP magnetic beads (Agencourt, A63880), and secondary amplification of the resulting product was performed (GoTaq PCR Kit, Promega, M7660), allowing addition of Illumina adapter sequences. Finally, libraries were purified with Ampure XP magnetic beads and sequenced using Illumina MiSeq 250 nt paired-end read length. The TCR CDR3 sequences were extracted from the raw sequencing data, wherein fastq files were first demultiplexed and reads were then mapped to germline V, D, J, and C reference sequences from the IMGT using iRepertoire bioinformatics tools.

TCR repertoire data analyses

Treg clonotypes with known CDR3 sequences were used as inputs for the R package tcR (version 2.2.1.11; ref. 32). For TCR Vβ gene usage analysis, geneUsage function was used with the parameter “.genes = MOUSE_TRBV.” For diversity analysis, Renyi index was calculated using the R vegan package (version 2.5-2) with the parameter “scales = seq(0, 10, 0.1), hill = FALSE,” where “scale” represents the Renyi scale (α). This generates a series of generalized diversity which covers the natural logarithm of clonotype number (richness, α = 0), Shannon entropy (α = 1), and natural logarithm of inverse Simpson index (α = 2). With α increased, a greater weight is assigned to clonotypes with a larger relative abundance, which reduces the overall diversity of a population but still keeps the index comparable with that of others under the same α. A higher Renyi index corresponds to a more diverse population. Evenness index was calculated by normalizing Renyi index to the natural logarithm of richness. Inverse Simpson index was calculated using repDiversity function with the parameter “.method = ‘inv.simp’,.quant = ‘read.count.’”

Similarity network of Treg clonotypes was constructed on the basis of the approach as described previously (33). Briefly, the top 200 abundant Treg CDR3 nucleotide clonotypes from each of the 4 mice in the same group were combined. Levenshtein distance between CDR3 amino acid sequences of Treg clonotypes was measured using stringdistmatrix function from the R package stringdist (version 0.9.5.5; ref. 34), and two Treg clonotypes with a Levenshtein distance ≤ 1 were deemed similar. The igraph R package (version 1.2.5; ref. 35) was used to construct a network, wherein each node represents a Treg clonotype and edges connect similar clones. The RCy3 R package (version 2.2.9; ref. 36) was used to load the network to Cytoscape (37), which was used for visualization.

The circlize R package (version 0.4.8; ref. 38) was used to generate Circos plots, where two sectors represent Treg TCR repertoires in the omentum and spleen of a representative mouse, and links represent Treg clonotypes that are found in both tissues.

RNA sequencing library preparation

Total RNA was purified using the Quick-RNA MicroPrep kit (Zymo Research, R1050) and used as input for the SMART-seq v4 cDNA synthesis kit (Takara, 634894). A total of 200 pg of cDNA was used as input for the NexteraXT kit (Illumina, FC-131-1096) for tagmentation, PCR amplification and indexing. Final libraries were quantitated by Qubit, pooled at equimolar ratio, and sequenced on a NextSeq500 using 75 bp PE chemistry at the UAB Helfin Genomics Core.

RNA sequencing data analyses

FASTQ reads were first processed by Trim Galore! (version 0.44) to trim adapter and low-quality reads, and then mapped to the mm10 mouse genome using TopHat2 (version 2.1.1; ref. 39) with the default parameters. Samtools (version 1.9; ref. 40) was used to sort the resulting BAM files and generate SAM files, which were then processed by htseq-count of HTSeq (version 0.6.1; ref. 41) with the parameter “-s no” to create gene counts. The count matrix was analyzed using the R package DESeq2 (version 1.20.0; ref. 42), and genes with counts higher than 10 in at least two samples were deemed expressed and kept. For differential expression analysis, the DESeq function was implemented on the raw counts, and the results function was then used with the contrast argument set to extract results from the omentum-over-spleen comparison. Genes with an adjusted P value < 0.1 and an absolute value of log2 fold change ≥ 1 were defined as differentially expressed genes (DEG). For visualization purpose, the regularized-logarithm transformation was conducted using rlog function, and heatmap was generated using the R package pheatmap.

For gene set enrichment analysis (GSEA; ref. 43), all detected genes were ranked by the −log10P value from DESeq2 multiplied by the sign of the fold change. The resulting ranked list was used as input for the GSEA PreRanked analysis, and pathways in the Kyoto Encyclopedia of Genes and Genomes (KEGG) subset of C2, Gene Ontology (GO) biological process of C5 and C7 MSigDB collections were surveyed. To compile a conserved tumor-associated Treg signature, count matrix from GSE116347 (23) was processed using the DESeq2 pipeline described above. Briefly, transcriptomes of Tregs isolated from three types of tumors were compared with those from the matched spleens, and DEGs that were shared among the three comparisons were obtained. This conserved DEG collection was further split into upregulated and downregulated gene sets and used as references in the GSEA PreRanked analysis.

For Ingenuity Pathway Analysis (QIAGEN; ref. 44), the log2 fold change of DEGs (omentum over spleen) was used. Predicted upstream regulators with an activation z-score ≥ 2 or ≤ −2 were considered to be activated or inhibited in the omental immigrant Tregs. The igraph R package was used to construct a network where each node represents a predicted regulator or a target DEG, and edges connect regulators and their targets. The RCy3 R package was used to load the network to Cytoscape, which was used for visualization.

For MiXCR (version 3.0.12;ref. 45), trimmed RNA sequencing (RNA-seq) FASTQ files were used as inputs. The analyze shotgun command was implemented with the parameter “–starting material rna\–receptor-type tcr”. Assembled TCRβ clonotypes were analyzed using the tcR pipeline described above.

Statistical analysis

Statistical tests were performed in GraphPad Prism version 7.0a. Paired and unpaired t tests, and one-way ANOVA followed by the Tukey multiple comparison test were conducted.

Data availability

RNA-seq and TCR repertoire data from this study are available from the NCBI Gene Expression Omnibus GSE181124 and BioProject PRJNA750800, respectively.

Tregs accumulate in the omentum during tumor progression

To understand how tumor implantation affects the abundance of Tregs in the omentum, we intraperitoneally injected EG7.1.15 tumor cells into syngeneic B6 mice. Tumor colonization increased omental weight starting from day 10, and large tumors were observed at day 20 (Fig. 1A). The frequency and number of CD4+FoxP3+ Tregs increased dramatically over this same time (Fig. 1BD). We also observed that the expression of Ki67, a marker of cell proliferation, decreased as the number of Tregs increased (Fig. 1E and F). Omental Tregs abundantly express CD25, ST2 (IL33 receptor), KLRG1, PD-1 and CTLA4 under steady-state conditions (Fig. 1G). The expression of CD25 and ST2 remained stable over time whereas that of KLRG1, CD69, PD-1, and CTLA4 was further elevated as tumors progressed (Fig. 1G). Moreover, tumor-associated Tregs in the omentum poorly produced Th1-, Th2-, and Th17-associated cytokines (Supplementary Fig. S1). The increase in Tregs in the omentum coincided with a decrease in CD8+ T-cell function, as revealed by reduced IFNγ and TNFα production (Fig. 1H and I). Similar results were observed using ID8-OVA tumor cells (Supplementary Fig. S2).

Figure 1.

Tregs accumulate in the omentum during tumor growth. A, EG7.1.15 tumor cells were intraperitoneally injected into C57BL/6 mice and omental mass measured. Data shown are mean ± SEM for 4–7 mice/timepoint. B, Flow cytometry analysis of CD4+ FoxP3+ Tregs in omentum. Plots are gated on CD4+ T cells. The frequency (C), and absolute and normalized numbers (D) of CD4+ T-cell subsets were determined by flow cytometry. Data shown are mean ± SEM for 4–7 mice/timepoint. E, Flow cytometry analysis of Ki67 expression by omental Tregs. Plots are gated on CD4+ FoxP3+ Tregs. F, Frequency of Ki67+ Tregs. Data shown are mean ± SEM for 4–7 mice/timepoint. G, Expression of indicated proteins in omental Tregs during tumor progression. Data shown are mean ± SEM for 4–12 mice/timepoint. H, Flow cytometry analysis of cytokine expression by omental cells after restimulation with PMA and calcium ionophore. Plots are gated on CD8+ T cells. I, Frequency of cytokine-producing CD8+ T cells. Data shown are mean ± SEM for 4–7 mice. Significance was tested using one-way ANOVA (G) or unpaired t test (I). All data are representative of at least two independent experiments.

Figure 1.

Tregs accumulate in the omentum during tumor growth. A, EG7.1.15 tumor cells were intraperitoneally injected into C57BL/6 mice and omental mass measured. Data shown are mean ± SEM for 4–7 mice/timepoint. B, Flow cytometry analysis of CD4+ FoxP3+ Tregs in omentum. Plots are gated on CD4+ T cells. The frequency (C), and absolute and normalized numbers (D) of CD4+ T-cell subsets were determined by flow cytometry. Data shown are mean ± SEM for 4–7 mice/timepoint. E, Flow cytometry analysis of Ki67 expression by omental Tregs. Plots are gated on CD4+ FoxP3+ Tregs. F, Frequency of Ki67+ Tregs. Data shown are mean ± SEM for 4–7 mice/timepoint. G, Expression of indicated proteins in omental Tregs during tumor progression. Data shown are mean ± SEM for 4–12 mice/timepoint. H, Flow cytometry analysis of cytokine expression by omental cells after restimulation with PMA and calcium ionophore. Plots are gated on CD8+ T cells. I, Frequency of cytokine-producing CD8+ T cells. Data shown are mean ± SEM for 4–7 mice. Significance was tested using one-way ANOVA (G) or unpaired t test (I). All data are representative of at least two independent experiments.

Close modal

Conventional CD4+ T cells can convert to Tregs in the tumor microenvironment (24, 25). To determine the origins of Tregs that accumulate in tumor-bearing omenta, we first examined the expression of Helios and Neuropilin-1 (Nrp-1), markers of thymus-derived Tregs (46–48). We found that over 65% of omental Tregs expressed Helios highly prior to tumor implantation, and the proportion of Helios+ Tregs increased further as tumors progressed (Fig. 2A; Supplementary Fig. S3A). Similarly, the percentage of Helios+Nrp-1+ Tregs was elevated (Fig. 2A), whereas RORγt+Helios Tregs, which are reported to be of extrathymic origin (49, 50), comprised a smaller fraction in the omentum than in the colon and reduced further during tumor progression (Fig. 2B). To test whether tumor antigen-specific CD4+ T cells differentiated into Tregs in tumor-bearing omenta, we adoptively transferred CTV-labeled, naïve OT-II cells (CD45.1) into naïve and tumor-bearing congenic CD45.2 mice. We found that although OT-II T cells diluted CTV in the tumor-bearing omentum, they did not turn on FoxP3 expression (Fig. 2C; Supplementary Fig. S3B). To determine whether this result extended to polyclonal CD4+ T cells, we transferred naïve conventional CD4+ T cells from FoxP3-GFP mice into naïve and tumor-bearing congenic CD45.1 mice. Again, very few transferred cells turned on FoxP3 expression (Fig. 2D). To test whether competition from endogenous Tregs prevented naïve CD4+ T cells from differentiating into Tregs, we transferred naïve conventional CD4+ T cells, either alone or with Tregs, into TCRβδ−/− animals and inoculated them with EG7.1.15 tumors. We found that very few naïve CD4+ T cells converted to Tregs, regardless of whether they were transferred with or without Tregs (Fig. 2E and G). In contrast, most of the cotransferred Tregs maintained FoxP3 expression (Fig. 2F and G). Importantly, the cotransfer of Tregs promoted tumor growth (Fig. 2H). Taken together, these data suggest that naïve CD4+ T cells do not differentiate into Tregs in tumor-bearing omentum, but that preexisting Tregs accumulate in the omentum following tumor colonization.

Figure 2.

Naïve CD4+ T cells do not become Tregs in omental tumors. EG7.1.15 tumor cells were intraperitoneally injected into C57BL/6 mice, and the expression of Helios, Nrp-1 (A), and RORγt (B) in omental and colonic lamina propria (LP) Tregs was analyzed by flow cytometry. Plots are gated on CD4+FoxP3+ Tregs, and the numbers indicate the percentage of cells for the gate shown. Data are plotted as mean ± SEM for 5 mice/time. Significance tested by one-way ANOVA. C, Purified, CTV-labeled CD45.1+CD44loCD25 OT-II were transferred into naïve or day-9 tumor-bearing mice and cells in the omenta were analyzed by flow cytometry 7 days later. Plots are gated on CD45.1+CD4+ cells, and the numbers indicate the percentage of cells for that quadrant. Data are plotted as mean ± SEM. N = 5–10. Significance tested using unpaired t test. Data are combined from two independent experiments. D, Purified CD45.2+CD44loCD25FoxP3(GFP) CD4+ T cells were transferred into naïve or day-9 tumor-bearing mice and cells in the omenta were analyzed by flow cytometry 7 days later. Plots are gated on transferred CD45.2+CD4+ cells, and the numbers indicate the percentage of cells for the gate shown. Data are plotted as mean ± SEM. N = 6–8. Significance tested using unpaired t test. Data are combined from two independent experiments. E–H, Purified CD45.1+CD44loFoxP3(GFP) CD4+ T cells were transferred with or without CD45.1FoxP3(GFP)+ CD4+ Tregs into T cell–deficient mice and tumor cells were injected 7 days later. Cells in the omenta were analyzed 15 days later by flow cytometry. Plots are gated on CD4+CD45.1+ cells (E) or CD4+CD45.1 cells (F), and the numbers indicate the percentage of cells for the gate shown. G, The percentage of indicated cells that express FoxP3. H, Omental mass on day 15. Data are plotted as mean ± SEM. N = 7–8. Significance tested using one-way ANOVA (G) or unpaired t test (H). Data are representative of two independent experiments.

Figure 2.

Naïve CD4+ T cells do not become Tregs in omental tumors. EG7.1.15 tumor cells were intraperitoneally injected into C57BL/6 mice, and the expression of Helios, Nrp-1 (A), and RORγt (B) in omental and colonic lamina propria (LP) Tregs was analyzed by flow cytometry. Plots are gated on CD4+FoxP3+ Tregs, and the numbers indicate the percentage of cells for the gate shown. Data are plotted as mean ± SEM for 5 mice/time. Significance tested by one-way ANOVA. C, Purified, CTV-labeled CD45.1+CD44loCD25 OT-II were transferred into naïve or day-9 tumor-bearing mice and cells in the omenta were analyzed by flow cytometry 7 days later. Plots are gated on CD45.1+CD4+ cells, and the numbers indicate the percentage of cells for that quadrant. Data are plotted as mean ± SEM. N = 5–10. Significance tested using unpaired t test. Data are combined from two independent experiments. D, Purified CD45.2+CD44loCD25FoxP3(GFP) CD4+ T cells were transferred into naïve or day-9 tumor-bearing mice and cells in the omenta were analyzed by flow cytometry 7 days later. Plots are gated on transferred CD45.2+CD4+ cells, and the numbers indicate the percentage of cells for the gate shown. Data are plotted as mean ± SEM. N = 6–8. Significance tested using unpaired t test. Data are combined from two independent experiments. E–H, Purified CD45.1+CD44loFoxP3(GFP) CD4+ T cells were transferred with or without CD45.1FoxP3(GFP)+ CD4+ Tregs into T cell–deficient mice and tumor cells were injected 7 days later. Cells in the omenta were analyzed 15 days later by flow cytometry. Plots are gated on CD4+CD45.1+ cells (E) or CD4+CD45.1 cells (F), and the numbers indicate the percentage of cells for the gate shown. G, The percentage of indicated cells that express FoxP3. H, Omental mass on day 15. Data are plotted as mean ± SEM. N = 7–8. Significance tested using one-way ANOVA (G) or unpaired t test (H). Data are representative of two independent experiments.

Close modal

To confirm that preexisting Tregs colonize tumor-bearing omenta, we reconstituted lethally irradiated CD45.1 mice with BM cells from wildtype (WT→WT) or CD25KO (CD25KO→WT) mice. After reconstitution and tumor implantation, we found that the vast majority of the total leukocytes in the spleen were of donor origin (Supplementary Fig. S4A), and the percentage of major lymphocyte subsets was similar in both groups (Supplementary Fig. S4B). Similarly, about 90% of the leukocytes in the omentum were donor derived (Fig. 3A and B), and most of omental CD8+ T cells were donor derived in both groups (Fig. 3,C and D). In contrast, although most of the Tregs of chimeric mice receiving WT cells were of donor origin, virtually all the Tregs of chimeric animals receiving CD25KO BM cells were of host origin (Fig. 3E and F), consistent with previous studies on the essential role of CD25 in Treg development and the radioresistant properties of Tregs (51, 52). Despite their different origins, Tregs accumulated to the same frequency and total number in tumor-bearing omenta in both groups (Fig. 3E and F) and suppressed CD8+ T-cell activity to the same extent (Fig. 3G), leading to similar tumor outgrowth (Fig. 3H). These results show that Tregs generated prior to tumor implantation are sufficient to impair CD8+ T-cell responses and promote tumor growth.

Figure 3.

Preexisting Tregs are sufficient for tumor growth in the omentum. Lethally irradiated CD45.1 mice were reconstituted with BM from WT or CD25KO mice and implanted intraperitoneally with tumor cells. A, Host and donor cells in the omentum were identified by flow cytometry. Numbers indicate the mean ± SD for the gate shown. B, The numbers of host and donor cells in the omentum are plotted as mean ± SEM. C, Host and donor CD8+ T cells in the omentum were identified by flow cytometry. Numbers indicate the mean ± SD for the gate shown. D, The numbers of host and donor CD8+ T cells in the omentum are plotted as mean ± SEM. E, Host and donor Tregs in the omentum were identified by flow cytometry. Numbers indicate the mean ± SD for the gate shown. F, The numbers of host and donor Tregs in the omentum are plotted as mean ± SEM. G, Cytokine expression of omental CD8+ T cells after restimulation. The frequency of cytokine expressing CD8+ T cells is plotted as mean ± SEM. H, Omental mass is plotted as mean and SEM. N = 5–6. Significance tested using unpaired t test. Data are representative of three independent experiments.

Figure 3.

Preexisting Tregs are sufficient for tumor growth in the omentum. Lethally irradiated CD45.1 mice were reconstituted with BM from WT or CD25KO mice and implanted intraperitoneally with tumor cells. A, Host and donor cells in the omentum were identified by flow cytometry. Numbers indicate the mean ± SD for the gate shown. B, The numbers of host and donor cells in the omentum are plotted as mean ± SEM. C, Host and donor CD8+ T cells in the omentum were identified by flow cytometry. Numbers indicate the mean ± SD for the gate shown. D, The numbers of host and donor CD8+ T cells in the omentum are plotted as mean ± SEM. E, Host and donor Tregs in the omentum were identified by flow cytometry. Numbers indicate the mean ± SD for the gate shown. F, The numbers of host and donor Tregs in the omentum are plotted as mean ± SEM. G, Cytokine expression of omental CD8+ T cells after restimulation. The frequency of cytokine expressing CD8+ T cells is plotted as mean ± SEM. H, Omental mass is plotted as mean and SEM. N = 5–6. Significance tested using unpaired t test. Data are representative of three independent experiments.

Close modal

Omental tumors accumulate Tregs via recruitment rather than clonal expansion

To test whether Treg accumulation in tumors was primarily driven by clonal expansion, we sorted CD25+ FoxP3(GFP)+ Tregs from the omenta of naïve or tumor-bearing animals and sequenced the CDR3 region of their TCRβ chains. Of note, the sorted Treg population represents the entire repertoire of naïve omenta and over 70% of cells from tumor-bearing omenta. Therefore, difference in the number of input cells between the two groups (Supplementary Table S2) is not due to sampling bias but reflects enlargement of the omental Treg compartment, and very few clones would have escaped this analysis. We also adjusted the number of sequence reads accordingly to ensure similar coverage among samples (Supplementary Table S2). As a result, we found that CDR3 length distribution was essentially identical between the two groups (Fig. 4A) and observed more and smaller clones of Tregs in tumor-bearing omenta compared with naïve omenta (Fig. 4B and C). Moreover, the elevated number of clonotypes (richness) translated to increased repertoire diversity for Tregs from tumor-bearing omenta compared with those from naïve omenta, as revealed by Renyi diversity curve (Fig. 4D) and the inverse Simpson index (Fig. 4E). To evaluate the evenness of the Treg repertoire, we normalized the Renyi diversity to richness, and found that tumors decreased the repertoire evenness (Fig. 4F). These results suggest that tumor progression in the omentum increases the Treg repertoire diversity, which is likely due to the recruitment of additional Treg clones from other sites.

Figure 4.

Clonal diversity of Tregs increases after tumor implantation in the omentum. The CDR3 sequence of TCRβ was sequenced from Tregs in the omenta of naïve and day-15 tumor-bearing mice. A, Distribution of CDR3 length is plotted as the mean ± SD. B, Representative tree maps of TCRβ sequence diversity, in which each colored segment represents a clonotype and the size reflects the relative abundance. C, The number of Treg nucleotide clonotypes. D, Renyi diversity plot of TCRβ sequences from Tregs in naïve and tumor-bearing omenta. The blue shaded band represents a pointwise 95% confidence interval for the model. E, Inverse Simpson index of TCRβ sequences from Tregs in naïve and tumor-bearing omenta. Significance was determined using unpaired t test. F, Evenness plot of TCRβ sequences from Tregs in naïve and tumor-bearing omenta. The blue shaded band represents a pointwise 95% confidence interval for the model. G, Treg clonotypes found in at least 3 mice were identified and their relative abundance plotted as a heatmap. Clones found only in tumor-bearing mice are indicated in purple. H, The 200 most abundant nucleotide clonotypes from each mouse were translated to amino acid sequences and the Levenshtein distance between them was used to generate networks in which each node represents a clonotype; the size of the dot indicates relative abundance and clones with a Levenshtein distance ≤ 1 are connected by an edge.

Figure 4.

Clonal diversity of Tregs increases after tumor implantation in the omentum. The CDR3 sequence of TCRβ was sequenced from Tregs in the omenta of naïve and day-15 tumor-bearing mice. A, Distribution of CDR3 length is plotted as the mean ± SD. B, Representative tree maps of TCRβ sequence diversity, in which each colored segment represents a clonotype and the size reflects the relative abundance. C, The number of Treg nucleotide clonotypes. D, Renyi diversity plot of TCRβ sequences from Tregs in naïve and tumor-bearing omenta. The blue shaded band represents a pointwise 95% confidence interval for the model. E, Inverse Simpson index of TCRβ sequences from Tregs in naïve and tumor-bearing omenta. Significance was determined using unpaired t test. F, Evenness plot of TCRβ sequences from Tregs in naïve and tumor-bearing omenta. The blue shaded band represents a pointwise 95% confidence interval for the model. G, Treg clonotypes found in at least 3 mice were identified and their relative abundance plotted as a heatmap. Clones found only in tumor-bearing mice are indicated in purple. H, The 200 most abundant nucleotide clonotypes from each mouse were translated to amino acid sequences and the Levenshtein distance between them was used to generate networks in which each node represents a clonotype; the size of the dot indicates relative abundance and clones with a Levenshtein distance ≤ 1 are connected by an edge.

Close modal

Although we did not observe dramatic clonal expansion in tumor-associated Tregs, we considered the possibility that tumor antigens might drive the limited expansion of related clones in multiple animals and reduce the repertoire evenness. To address this possibility, we enumerated the CDR3 clonotypes (at the nucleotide level) that appeared in at least 3 of 8 animals. We found 35 of these public clonotypes (Fig. 4G), of which only 16 were observed exclusively in tumor-bearing omenta (Fig. 4G, purple), suggesting that they may be responding to tumor antigens. These 16 clonotypes accounted for only 1.4% (±1.1%) of total CDR3 reads from omental tumors, implying that tumor antigen-driven clonal expansion of Tregs is very limited.

TCRs with similar CDR3 amino acid sequences may respond to the same antigens (33). To define clusters of structurally similar TCRs, we combined the 200 most abundant nucleotide clones from each of the 4 mice in the same group, calculated the Levenshtein distance between their CDR3 amino acid sequences, and connected those with a distance ≤1 (33). Using this threshold, Treg clones whose CDR3 amino acid sequences differed by at most one were included in the same connected component. We found that the complexity of TCR networks was similar between naïve and tumor-bearing groups, and a great portion of Treg clones still remained isolated (Fig. 4H). We obtained a similar result by analyzing the TCR network complexity of splenic Tregs from naïve and tumor-bearing mice (Supplementary Fig. S5). These data suggest that Treg accumulation in tumor-bearing omenta is not due to antigen-driven clonal expansion.

Tregs in perigonadal adipose tissue are considered tissue resident (14), raising the possibility that Tregs in omental adipose tissue are also tissue resident. To test this possibility, we performed parabiosis surgery in which we joined the circulation of two naïve mice for 28 days and found that circulating Tregs in the blood had reached equilibrium (50% host-derived and 50% partner-derived), whereas those in the omentum did not (Supplementary Fig. S6), suggesting that many omental Tregs are tissue resident. To test whether Treg accumulation in tumor-bearing omenta is driven by recruitment of circulating Tregs, we surgically joined tumor-bearing mice with congenic tumor-free mice (Fig. 5A). After 14 days, we found that circulating Tregs in the blood achieved equilibrium (Fig. 5B), whereas those in the omentum had not reached equilibrium in either partner, although the frequency of partner-derived Tregs was significantly higher in the omentum of tumor-bearing parabionts (Fig. 5C). Moreover, because of the increased number of cells in tumor-bearing omenta, the total number of partner-derived Tregs was more than 90-fold higher in tumor-bearing parabionts than in naïve parabionts (Fig. 5D).

Figure 5.

Tumor-bearing omenta recruit Tregs from circulation. A–D, Day-2 tumor-bearing CD45.1 mice were surgically joined with naïve C57BL/6 mice and omenta obtained from both partners on day 14 after surgery. B, Frequency of host- and donor-derived Tregs in the blood. Error bars represent mean ± SEM of four partners. C, Representative plots of host- and donor-derived Tregs in the omenta of paired mice. Plots are gated on CD4+FoxP3+ cells. Number indicates percentage of partner-derived Tregs. Frequency of partner-derived Tregs in the omenta is shown with pairs connected by a line. D, Number of partner-derived Tregs in the omenta with pairs connected by a line. N = 4 pairs. Significance tested using paired t test. Data are combined from two independent experiments. E–H, Naïve CD45.1 mice were surgically joined with naïve C57BL/6 mice, tumors implanted intraperitoneally in the CD45.1 partner and omenta obtained from both partners on day 14 after tumor injection. F, Frequency of host- and donor-derived Tregs in the blood. Error bars represent mean ± SEM of five partners. G, Representative plots of host- and donor-derived Tregs in the omenta of paired mice. Plots are gated on CD4+ FoxP3+ cells. Number indicates percentage of partner-derived Tregs. Frequency of partner-derived Tregs in the omenta is shown with pairs connected by a line. H, Number of partner-derived Tregs in the omenta with pairs connected by a line. N = 5 pairs. Significance tested using paired t test. Data are combined from three independent experiments. I and J, CTV-labeled CD4+ T cells from C57BL/6 mice were intravenously injected into naïve and day-14 tumor-bearing CD45.1 mice and donor-derived cells in the omenta (OM) or inguinal lymph node (LN) were enumerated 19 hours later using flow cytometry. I, Representative plots were gated on CD4+ cells and the number indicates the percentage ± SD of donor cells. The number of donor-derived CD4+ cells is plotted as mean ± SEM. J, The number of donor-derived Tregs is plotted as mean ± SEM. N = 8–9. Significance was tested using unpaired t test. Data are representative of two independent experiments.

Figure 5.

Tumor-bearing omenta recruit Tregs from circulation. A–D, Day-2 tumor-bearing CD45.1 mice were surgically joined with naïve C57BL/6 mice and omenta obtained from both partners on day 14 after surgery. B, Frequency of host- and donor-derived Tregs in the blood. Error bars represent mean ± SEM of four partners. C, Representative plots of host- and donor-derived Tregs in the omenta of paired mice. Plots are gated on CD4+FoxP3+ cells. Number indicates percentage of partner-derived Tregs. Frequency of partner-derived Tregs in the omenta is shown with pairs connected by a line. D, Number of partner-derived Tregs in the omenta with pairs connected by a line. N = 4 pairs. Significance tested using paired t test. Data are combined from two independent experiments. E–H, Naïve CD45.1 mice were surgically joined with naïve C57BL/6 mice, tumors implanted intraperitoneally in the CD45.1 partner and omenta obtained from both partners on day 14 after tumor injection. F, Frequency of host- and donor-derived Tregs in the blood. Error bars represent mean ± SEM of five partners. G, Representative plots of host- and donor-derived Tregs in the omenta of paired mice. Plots are gated on CD4+ FoxP3+ cells. Number indicates percentage of partner-derived Tregs. Frequency of partner-derived Tregs in the omenta is shown with pairs connected by a line. H, Number of partner-derived Tregs in the omenta with pairs connected by a line. N = 5 pairs. Significance tested using paired t test. Data are combined from three independent experiments. I and J, CTV-labeled CD4+ T cells from C57BL/6 mice were intravenously injected into naïve and day-14 tumor-bearing CD45.1 mice and donor-derived cells in the omenta (OM) or inguinal lymph node (LN) were enumerated 19 hours later using flow cytometry. I, Representative plots were gated on CD4+ cells and the number indicates the percentage ± SD of donor cells. The number of donor-derived CD4+ cells is plotted as mean ± SEM. J, The number of donor-derived Tregs is plotted as mean ± SEM. N = 8–9. Significance was tested using unpaired t test. Data are representative of two independent experiments.

Close modal

Because it takes some time to establish circulatory equilibrium in parabiotic mice, we performed a similar experiment in which we joined two naïve congenic mice for 14 days, and then injected tumors into the CD45.1+ partner (Fig. 5E). After 14 days (28 days after joining), we found that circulating Tregs in the blood had reached equilibrium (Fig. 5F). Interestingly, although the frequency of partner-derived Tregs in the naïve parabiont was only 28%, the frequency of partner-derived Tregs in the tumor-bearing parabiont had reached equilibrium (47%; Fig. 5G). Again, we observed a dramatic increase in the total number of partner-derived Tregs in the tumor-bearing parabionts (Fig. 5H).

As a complementary method, we intravenously transferred CTV-labeled CD4+ T cells into naïve and day 14 tumor-bearing congenic animals and enumerated donor cells in the omentum and lymph node 19 hours later. Although very few transferred T cells homed to naïve omentum, many more homed to tumor-bearing omentum (Fig. 5I). We observed a similar increase in the homing of transferred Tregs to tumor-bearing omentum compared with naïve omentum (Fig. 5J). In contrast, similar numbers of T cells homed to the inguinal lymph nodes of naïve and tumor-bearing mice (Fig. 5I and J). Taken together, these data suggest that, despite the tissue-resident properties of omental Tregs under steady-state conditions, tumor implantation leads to a dramatic increase in Treg recruitment from the circulation.

Splenic Tregs contribute to the omental compartment during tumor development

We next compared the TCRβ repertoire of omental and splenic Tregs before and after tumor implantation. We found that Tregs in the spleen and omenta of naïve mice had minimal overlapping clonotypes, but that following tumor implantation, the number of overlapping clonotypes was strikingly augmented (Fig. 6A and B; Supplementary Fig. S7). Moreover, overlapping clonotypes also comprised a larger fraction of total CDR3 reads in the omentum after tumor implantation (Fig. 6C), suggesting that the increased similarity of the splenic and omental Treg TCR repertoires is not solely due to the different numbers of input cells. Interestingly, the Vβ gene usage of splenic Tregs did not change after tumor implantation, but the Vβ gene usage of omental Tregs changed dramatically after tumor implantation and occupied a distinct space on the PCA plot (Fig. 6D). We observed a similar trend when we examined TCR Vβ genes identified in conventional RNA-seq data from splenic and omental Tregs in tumor-bearing mice (Supplementary Fig. S8). These data suggest that Tregs are recruited from circulating sources like those in the spleen, wherein cells with particular antigen specificities may be preferentially selected.

Figure 6.

Splenic Tregs accumulate in tumor-bearing omenta. A–D, Splenic and omental Tregs were obtained from naïve and day-15 tumor-bearing FoxP3-GFP-DTR mice and the CDR3 region of TCRβ chains were sequenced. A, Circos plots identify shared clonotypes (connected by a line) in the omentum and spleen of representative animals. Line width indicates relative clonal abundance. B, The number of shared clonotypes in the omentum and spleen. C, The percentage of Treg CDR3 reads in the omentum that are from shared clonotypes. D, Principal component analysis of Vβ gene usage by splenic and omental Tregs from naïve and tumor-bearing mice. N = 4. Significance tested using unpaired t test. E–J, The spleens of day-12 tumor-bearing KikGR mice were exposed to violet light and photoconverted (red) cells that had trafficked to the omentum were identified 3 days later. F, Flow cytometry plots are gated on CD45+ cells in the omentum and show the frequency (mean ± SD) of red cells. G, The number of CD45.2+red+ cells in the omentum is plotted as mean ± SEM. H, The percentage of red+CD4+CD25+ cells is plotted as mean ± SEM. I, Plots of red+CD4+ T cells concatenated from 3 animals showing the percentage of CD25hi Tregs. J, The number of red+CD25+CD4+ Tregs is shown as mean ± SEM of N = 3 mice per group. Significance tested using unpaired t test. Data are representative of two independent experiments. K–M, Splenectomized (Splx) or sham-operated C57BL/6 mice were intraperitoneally injected with tumor cells and the omental mass (K), frequency and number of Tregs in the omenta (L), and number of Tregs in the inguinal and mediastinal lymph nodes (M) were enumerated on day 15. Treg frequency is plotted as mean ± SEM. Omental mass is plotted as mean ± SEM. N = 6. Significance was tested using unpaired t test. Data are representative of two independent experiments.

Figure 6.

Splenic Tregs accumulate in tumor-bearing omenta. A–D, Splenic and omental Tregs were obtained from naïve and day-15 tumor-bearing FoxP3-GFP-DTR mice and the CDR3 region of TCRβ chains were sequenced. A, Circos plots identify shared clonotypes (connected by a line) in the omentum and spleen of representative animals. Line width indicates relative clonal abundance. B, The number of shared clonotypes in the omentum and spleen. C, The percentage of Treg CDR3 reads in the omentum that are from shared clonotypes. D, Principal component analysis of Vβ gene usage by splenic and omental Tregs from naïve and tumor-bearing mice. N = 4. Significance tested using unpaired t test. E–J, The spleens of day-12 tumor-bearing KikGR mice were exposed to violet light and photoconverted (red) cells that had trafficked to the omentum were identified 3 days later. F, Flow cytometry plots are gated on CD45+ cells in the omentum and show the frequency (mean ± SD) of red cells. G, The number of CD45.2+red+ cells in the omentum is plotted as mean ± SEM. H, The percentage of red+CD4+CD25+ cells is plotted as mean ± SEM. I, Plots of red+CD4+ T cells concatenated from 3 animals showing the percentage of CD25hi Tregs. J, The number of red+CD25+CD4+ Tregs is shown as mean ± SEM of N = 3 mice per group. Significance tested using unpaired t test. Data are representative of two independent experiments. K–M, Splenectomized (Splx) or sham-operated C57BL/6 mice were intraperitoneally injected with tumor cells and the omental mass (K), frequency and number of Tregs in the omenta (L), and number of Tregs in the inguinal and mediastinal lymph nodes (M) were enumerated on day 15. Treg frequency is plotted as mean ± SEM. Omental mass is plotted as mean ± SEM. N = 6. Significance was tested using unpaired t test. Data are representative of two independent experiments.

Close modal

To directly test whether Tregs trafficked from the spleen to omentum in tumor-bearing mice, we exposed the spleens of naïve and tumor-bearing KikGR mice to violet light, which photoconverts the Kikume Green-Red protein from green fluorescence to red fluorescence, and quantified photoconverted cells in their omenta 3 days later (Fig. 6E). We found very few red fluorescent cells in the omenta of mice not exposed to violet light, but identified substantial populations of red cells in the omenta of both naïve and tumor-bearing mice (Fig. 6F). The presence of tumor increased total cell recruitment from the spleen about 10-fold (Fig. 6G). The frequency of Tregs among total photoconverted leukocytes (Fig. 6H) or CD4+ T cells (Fig. 6I) was higher in the omenta of tumor-bearing mice, leading to a more than 30-fold difference in the number of Tregs recruited to naïve and tumor-bearing omenta (Fig. 6J), suggesting that Tregs are preferentially recruited from the spleen to the omentum in response to tumors. To test whether the spleen was the exclusive source of Tregs for tumor-bearing omenta, we performed splenectomy prior to tumor injection. However, the loss of the spleen had no impact on tumor size (Fig. 6K) and led to a modest and statistically insignificant reduction in the frequency (P = 0.06) and number (P = 0.24) of Tregs recruited to tumor-bearing omenta (Fig. 6L). Interestingly, splenectomy increased the number of Tregs in tumor-draining and nondraining lymph nodes (Fig. 6M), suggesting compensatory mechanisms are engaged to maintain overall Treg numbers. Taken together, our data show that circulating Tregs, including those in the spleen, are an important source of Tregs that are recruited to the omentum during tumor development.

Tregs that home to tumor-bearing omentum acquire tissue- and tumor-resident features

The noncirculating, tissue-resident Tregs in naïve omentum likely have features of adipose-resident Tregs (17). To test whether Tregs that were newly recruited to tumor-bearing omentum had similar properties, we surgically joined tumor-bearing mice with congenic naïve mice, sorted partner-derived (newly recruited) Tregs from the omenta and spleens of tumor-bearing (CD45.1+) parabionts on day 14 and performed RNA-seq analysis (Fig. 7A). We found that Tregs newly recruited to the omentum or spleen expressed divergent transcriptomes, with 593 differentially expressed genes (Fig. 7B). The immigrant Tregs in the omentum exhibited a more activated phenotype compared with their splenic counterparts, with higher expression of Gzmb, Id2, and Il10, and less Sell, Lef1, and Id3 (Fig. 7C, left). Chemotactic receptors were also differentially expressed, with higher expression of Ccr1, Ccr8, and Cxcr6 and lower expression of S1pr1, Ccr7, and Cxcr5 in omental immigrants (Fig. 7C, middle). Moreover, genes linked to visceral adipose tissue–associated Treg signatures, such as Areg, Coq10b, and Dgat1, were also increased in omental immigrants (Fig. 7C, right).

Figure 7.

Tregs recruited to the omentum and spleen express distinct transcriptomes. A, Day-2 tumor-bearing CD45.1+ FoxP3-GFP-DTR mice were surgically joined with naïve CD45.1FoxP3-GFP-DTR mice and RNA-seq was performed on partner-derived Tregs from the omenta (OM) and spleens (SPL) of tumor-bearing partners on day 14. B, Genes expressed more highly in omental (green) and splenic (red) immigrant Tregs are visualized by volcano plot. C, DEGs associated with cellular activation (left), migration (middle), or adipose-resident Treg signatures (right) are plotted as a heat map. D, The ranked gene list was compared with gene sets from the KEGG and GO biological process collections of GSEA MSigDB. E, The ranked gene list was compared with published gene sets defining adipose-associated and tumor-associated Treg signatures. F, Upstream regulator analysis shows the −log10 overlap P value of regulators activated (green) or repressed (red) in omental immigrant Tregs. G, A network plot shows predicted regulators (diamonds) or target DEGs (circles) and size indicates statistical significance.

Figure 7.

Tregs recruited to the omentum and spleen express distinct transcriptomes. A, Day-2 tumor-bearing CD45.1+ FoxP3-GFP-DTR mice were surgically joined with naïve CD45.1FoxP3-GFP-DTR mice and RNA-seq was performed on partner-derived Tregs from the omenta (OM) and spleens (SPL) of tumor-bearing partners on day 14. B, Genes expressed more highly in omental (green) and splenic (red) immigrant Tregs are visualized by volcano plot. C, DEGs associated with cellular activation (left), migration (middle), or adipose-resident Treg signatures (right) are plotted as a heat map. D, The ranked gene list was compared with gene sets from the KEGG and GO biological process collections of GSEA MSigDB. E, The ranked gene list was compared with published gene sets defining adipose-associated and tumor-associated Treg signatures. F, Upstream regulator analysis shows the −log10 overlap P value of regulators activated (green) or repressed (red) in omental immigrant Tregs. G, A network plot shows predicted regulators (diamonds) or target DEGs (circles) and size indicates statistical significance.

Close modal

To better understand the pathways that were affected, we performed GSEA against gene sets in the KEGG and GO biological process collections. We found that newly recruited omental Tregs overexpressed genes associated with cell migration and chemotaxis, as well as immune-suppressive cytokines and metabolic pathways linked to lipids and some amino acids (Fig. 7D). In contrast, pathways linked to mTOR signaling, the citric acid cycle and protein sumoylation were decreased in these cells relative to their splenic counterparts (Fig. 7D). We next compared our ranked gene list with gene lists associated with adipose-resident Tregs (53) and tumor-associated Tregs (23), and found that Tregs recruited to tumor-bearing omentum were enriched for both gene signatures (Fig. 7E). To validate this observation, we examined the expression of the hallmark tissue-resident Treg marker ST2 in both the photolabeling and parabiosis models and found it elevated in omental immigrant Tregs (Supplementary Fig. S9). These data suggest that Tregs newly recruited to tumor-bearing omentum rapidly acquire an activated, adipose-resident phenotype.

To identify the molecules that control the transcriptional program of omental Tregs, we performed upstream regulator analysis. This analysis predicted a variety of cytokines, T-cell signaling molecules and metabolic regulators that should positively (Fig. 7F, green) or negatively (Fig. 7F, red) regulate the transcriptional program of omental Tregs. Not surprisingly, several of these molecules are linked to genes expressed by adipose-resident Tregs and form a network of gene targets with interconnecting regulatory hubs (Fig. 7G). For example, CD36, FABP4, and FASN are positively regulated hubs involved in lipid metabolism and are predicted to control additional lipid metabolism genes like Scd1 and Dgat1 (Fig. 7F and G). These gene expression signatures are consistent with the programming of adipose-resident (17) and tumor-resident Tregs (54). In contrast, BCL6, BACH2, FOXO1, and KLF2 were predicted as negatively regulated hubs that promote gene expression signatures related to resting or memory cells in lymphoid tissues (Fig. 7F and G). These data suggest that circulating Tregs rapidly adapt to the microenvironment in the omentum/tumor and acquire adipose-resident and tumor-resident characteristics.

Here we show that tumor growth in the omentum promotes the rapid accumulation of Tregs, which does not involve the de novo generation of Tregs from naïve CD4+ T cells. In fact, Tregs derived from the pretumor repertoire are sufficient to suppress antitumor immunity and promote tumor growth. Although many Tregs in the omentum are tissue resident, those cells are not clonally expanded in tumor-bearing omentum. Instead, tumor growth promotes the recruitment of circulating Tregs and increases Treg clonal diversity in the omentum. TCR repertoire analysis also suggests that many newly recruited Tregs come from the spleen. Importantly, newly recruited Tregs rapidly acquire the characteristics of adipose-resident Tregs. These data define the dynamics of Treg accumulation in tumor-bearing omenta and demonstrate their capacity to adapt their gene expression programs to the local environment.

Although the omentum expresses factors like TGFβ (29) and retinoic acid (28), which are important for Treg differentiation (27), and tumor-associated Tregs can locally differentiate from naïve CD4+ T cells (24, 25), we find that very few naïve CD4+ T cells differentiate into Tregs in tumor-bearing omenta. Consistent with studies of other adipose tissues (14), we find that most omental Tregs express Helios and Nrp-1, markers of thymus-derived Tregs (46–48). We also find that the number of Helios-expressing and Nrp-1–expressing Tregs increases over time after tumor implantation. In fact, Tregs that exist before tumor implantation are sufficient to populate tumor-bearing omenta, suppress local CD8+ T-cell responses and promote tumor growth. Thus, we conclude that thymus-derived Tregs, rather than peripherally induced Tregs, are the primary source of tumor-associated Tregs in the omentum.

Many Tregs in epididymal adipose tissue are tissue resident (14), consistent with our parabiosis data showing poor recirculation of omental Tregs prior to tumor implantation. Moreover, Tregs in epididymal adipose tissue have a TCR repertoire with lower diversity than that of their lymphoid counterparts (11, 13, 14), suggesting local antigen-driven selection and expansion. Thus, one might conclude that omental Tregs clonally expand in response to tumor implantation. However, despite the nearly 40-fold increase in Tregs within a few weeks after tumor implantation in the omentum, we find no evidence of clonal expansion. Instead, we observe reduced Treg proliferation and increased TCR repertoire diversity over time after tumor implantation, suggesting that Treg accumulation is primarily driven by recruitment. Consistent with this idea, our parabiosis experiments also show that partner-derived (circulating) Tregs accumulate in tumor-bearing omentum more rapidly than in omentum without tumor. In situ photolabeling directly shows that splenic Tregs migrate to omental tumors. Taken together, our data support a model in which Tregs in tumor-bearing omenta originate from circulating precursors, including those in the spleen (13, 55).

What may be the mechanism for the recruitment of circulating Tregs to omental tumors? A variety of cells in the tumor microenvironment, including tumor cells (21, 22) and tumor-associated macrophages (56) secrete chemoattractants, which may drive the influx of circulating Tregs. In line with this idea, tumor-associated Tregs upregulate chemokine receptors such as CCR4 (20) and CCR10 (21). In fact, the top three pathways enriched in Tregs that recently immigrated to the omentum are associated with chemotaxis and cell migration, and we identify several chemokine receptors, including CCR1, CCR8, and CXCR6, that may facilitate Treg trafficking to tumor-bearing omentum or retain Tregs in that tissue (12, 15). Conversely, receptors like CCR7 and CXCR5 that retain T cells in lymphoid tissues (57), as well as S1PR1, which promotes cellular egress into the blood (58) are reduced in Tregs that have recently migrated to the omentum. These results provide important implications for the potential targets that could be used to inhibit the migration of circulating Tregs to omental tumors as therapies. In other preclinical tumor models, chemokine receptor antagonists (59, 60) have been effective in disrupting Treg recruitment and potentiating antitumor immunity. Therefore, it would be of interest to investigate which specific chemokine receptor(s) mediate(s) circulating Treg trafficking in the context of omental tumors and develop therapeutic strategies using the corresponding inhibitors and/or antibodies.

Our upstream regulator analysis reveals strengthened TCR signaling in omental immigrant Tregs compared with splenic immigrant Tregs, suggesting that Tregs in tumor-bearing omenta may be responding to local antigens. Consistent with this idea, Treg maintenance in epididymal fat depends on MHC II expression by local antigen-presenting cells (14). Moreover, TCR-transgenic Tregs specific for an epididymal fat–associated antigen more efficiently home there upon adoptive transfer compared with their polyclonal counterparts (55). We also find that omental Tregs exhibited distinct Vβ gene usage to their splenic counterparts, regardless of whether tumors develop. In fact, circulating Tregs that are recruited to omental tumors and spleen of the same mouse have different Vβ gene usage, suggesting that Tregs with varied antigen specificities may be skewed to migrating to different tissues, and their interaction with local antigens is an important step in their recruitment or retention.

Tumor-associated Tregs also exhibit distinct activation status and metabolic activities (54) compared with their lymphoid counterparts. For example, they express immunosuppressive cytokines like IL10 and IL35 more highly (61), and display elevated oxidative metabolism (54). Some of these gene expression signatures overlap with those of tissue Tregs (12, 23), thereby raising the question of whether Tregs recruited to tumors acquire these phenotypes, or alternatively, whether the features of preexisting tissue Tregs are mixed with those of recruited cells in bulk RNA-seq experiments. Previous work using trajectory analysis of single-cell RNA-seq data to examine Treg relationships in normal skin, lymph node, and melanoma lesions, suggests that tumor-associated Tregs are more closely connected with those in the lymph node than those in normal skin (12), favoring the model of cell recruitment and adaptation. Our experiments use parabiosis to definitively identify newly recruited Tregs in tumors and spleen and show that those in omental tumors rapidly acquire gene expression signatures of adipose-resident and tumor-resident Tregs, including gene sets associated with lipid and retinol metabolism, reinforcing the idea that recruited Tregs undergo adaptation to local conditions. This metabolic adaptation may help both the persistence and suppressive activity of newly recruited Tregs in omental tumors. In support of this idea, Tregs from murine melanoma depend on CD36 expression to maintain mitochondrial fitness, and the ablation of this molecule reduces the accumulation and function of intratumoral Tregs (62). CD36, FABP4, and FASN are predicted as activated regulators and linked to metabolic genes such as Scd1 and Dgat1 that are upmodulated in omental immigrant Tregs, which suggests the important roles of lipid metabolism in supporting these newly recruited cells in omental tumors. Meanwhile, omental milky spots contain hypoxic regions that support tumor cell self-renewal (63), and HIF1A, an important regulator of glycolysis (64), is predicted as an activated regulator in omental immigrant Tregs and may be used by them to adapt to the hypoxic microenvironment. This agrees with previous work showing that intratumoral Tregs use both glycolysis and fatty acid oxidation as energy supply (54). Therefore, it would be of interest to study how the perturbation of these above-mentioned regulators and metabolic pathways affects Treg-mediated immunosuppression in the context of omental tumors.

Although a variety of tumor types can metastasize to the omentum (6), our results are particularly relevant to patients with advanced ovarian cancer, which frequently metastasizes throughout the peritoneal cavity, including the omentum. Treg accumulation is associated with poor outcomes in ovarian cancer (10), and the frequency of Tregs in the omentum dramatically increases with disease severity (9). Our data suggest that these Tregs are recruited from the circulation and rapidly acquire an adipose-resident phenotype, including changes to their cellular metabolism. Given the linkage of Treg immunosuppressive activity with their metabolic programming (54), these data will help us understand the drivers of immune suppression in the omentum and to develop ways to counteract that suppression and reinvigorate antitumor immunity.

D. Starenki reports personal fees from iRepertoire, Inc outside the submitted work. P.A. Molina reports grants from NIH during the conduct of the study. R.C. Arend reports other support from AstraZeneca (advisory board), Clovis (advisory board), Leap Therapeutics (advisory board), Merck (advisory board), Tesaro/GSK (advisory board), and VBL Therapeutics—Steering Commitee outside the submitted work. No disclosures were reported by the other authors.

M. Liu: Conceptualization, formal analysis, investigation, writing–original draft, writing–review and editing. D. Starenki: Formal analysis, investigation, methodology, writing–review and editing. C.D. Scharer: Formal analysis, investigation, visualization, methodology, writing–review and editing. A. Silva-Sanchez: Formal analysis, investigation, writing–review and editing. P.A. Molina: Resources, investigation, writing–review and editing. J.S. Pollock: Resources, supervision, writing–review and editing. S.J. Cooper: Resources, formal analysis, supervision, writing–review and editing. R.C. Arend: Resources, supervision, funding acquisition, investigation, writing–review and editing. A.F. Rosenberg: Conceptualization, formal analysis, investigation, writing–review and editing. T.D. Randall: Conceptualization, resources, supervision, funding acquisition, writing–review and editing. S. Meza-Perez: Conceptualization, formal analysis, supervision, investigation, writing–review and editing.

The authors would like to thank Uma Mudunuru, Thomas Simpler and Rebecca Burnham for animal husbandry and genotyping, Sagar Hanumanthu in the Flow Cytometry and Single Cell (FCSC) Services Core for cell sorting, and Michael Crowley and David Crossman in the Heflin Center for Genomic Sciences for sequencing. The FCSC Core is supported by the Center for AIDS Research, AI027767, and the O'Neal Comprehensive Cancer Center, CA013148. The Heflin Center for Genomic Sciences is supported by the O'Neal Comprehensive Cancer Center, CA013148. This work was supported by R01 CA216234 to T.D. Randall and Department of Defense Early Career Investigator Award, W81XWH-18-1-0231, to R.C. Arend.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Meza-Perez
S
,
Randall
TD
.
Immunological functions of the omentum
.
Trends Immunol
2017
;
38
:
526
36
.
2.
Rangel-Moreno
J
,
Moyron-Quiroz
JE
,
Carragher
DM
,
Kusser
K
,
Hartson
L
,
Moquin
A
, et al
.
Omental milky spots develop in the absence of lymphoid tissue-inducer cells and support B and T cell responses to peritoneal antigens
.
Immunity
2009
;
30
:
731
43
.
3.
Benezech
C
,
Luu
NT
,
Walker
JA
,
Kruglov
AA
,
Loo
Y
,
Nakamura
K
, et al
.
Inflammation-induced formation of fat-associated lymphoid clusters
.
Nat Immunol
2015
;
16
:
819
28
.
4.
Oosterling
SJ
,
van der Bij
GJ
,
Bogels
M
,
van der Sijp
JR
,
Beelen
RH
,
Meijer
S
, et al
.
Insufficient ability of omental milky spots to prevent peritoneal tumor outgrowth supports omentectomy in minimal residual disease
.
Cancer Immunol Immunother
2006
;
55
:
1043
51
.
5.
Clark
R
,
Krishnan
V
,
Schoof
M
,
Rodriguez
I
,
Theriault
B
,
Chekmareva
M
, et al
.
Milky spots promote ovarian cancer metastatic colonization of peritoneal adipose in experimental models
.
Am J Pathol
2013
;
183
:
576
91
.
6.
Coccolini
F
,
Gheza
F
,
Lotti
M
,
Virzi
S
,
Iusco
D
,
Ghermandi
C
, et al
.
Peritoneal carcinomatosis
.
World J Gastroenterol
2013
;
19
:
6979
94
.
7.
Etzerodt
A
,
Moulin
M
,
Doktor
TK
,
Delfini
M
,
Mossadegh-Keller
N
,
Bajenoff
M
, et al
.
Tissue-resident macrophages in omentum promote metastatic spread of ovarian cancer
.
J Exp Med
2020
;
217
:
e20191869
.
8.
Lee
W
,
Ko
SY
,
Mohamed
MS
,
Kenny
HA
,
Lengyel
E
,
Naora
H
.
Neutrophils facilitate ovarian cancer premetastatic niche formation in the omentum
.
J Exp Med
2019
;
216
:
176
94
.
9.
Pearce
OMT
,
Delaine-Smith
RM
,
Maniati
E
,
Nichols
S
,
Wang
J
,
Bohm
S
, et al
.
Deconstruction of a metastatic tumor microenvironment reveals a common matrix response in human cancers
.
Cancer Discov
2018
;
8
:
304
19
.
10.
Curiel
TJ
,
Coukos
G
,
Zou
L
,
Alvarez
X
,
Cheng
P
,
Mottram
P
, et al
.
Specific recruitment of regulatory T cells in ovarian carcinoma fosters immune privilege and predicts reduced survival
.
Nat Med
2004
;
10
:
942
9
.
11.
Burzyn
D
,
Kuswanto
W
,
Kolodin
D
,
Shadrach
JL
,
Cerletti
M
,
Jang
Y
, et al
.
A special population of regulatory T cells potentiates muscle repair
.
Cell
2013
;
155
:
1282
95
.
12.
Miragaia
RJ
,
Gomes
T
,
Chomka
A
,
Jardine
L
,
Riedel
A
,
Hegazy
AN
, et al
.
Single-cell transcriptomics of regulatory T cells reveals trajectories of tissue adaptation
.
Immunity
2019
;
50
:
493
504
.
13.
Delacher
M
,
Imbusch
CD
,
Hotz-Wagenblatt
A
,
Mallm
JP
,
Bauer
K
,
Simon
M
, et al
.
Precursors for nonlymphoid-tissue treg cells reside in secondary lymphoid organs and are programmed by the transcription factor BATF
.
Immunity
2020
;
52
:
295
312
.
14.
Kolodin
D
,
van Panhuys
N
,
Li
C
,
Magnuson
AM
,
Cipolletta
D
,
Miller
CM
, et al
.
Antigen- and cytokine-driven accumulation of regulatory T cells in visceral adipose tissue of lean mice
.
Cell Metab
2015
;
21
:
543
57
.
15.
Panduro
M
,
Benoist
C
,
Diane
M
.
Tissue Tregs
.
Annu Rev Immunol
2016
;
34
:
609
33
.
16.
Gyllenhammer
LE
,
Lam
J
,
Alderete
TL
,
Allayee
H
,
Akbari
O
,
Katkhouda
N
, et al
.
Lower omental t-regulatory cell count is associated with higher fasting glucose and lower β-cell function in adults with obesity
.
Obesity
2016
;
24
:
1274
82
.
17.
Cipolletta
D
,
Feuerer
M
,
Li
A
,
Kamei
N
,
Lee
J
,
Shoelson
SE
, et al
.
PPAR-gamma is a major driver of the accumulation and phenotype of adipose tissue Treg cells
.
Nature
2012
;
486
:
549
53
.
18.
Vasanthakumar
A
,
Moro
K
,
Xin
A
,
Liao
Y
,
Gloury
R
,
Kawamoto
S
, et al
.
The transcriptional regulators IRF4, BATF and IL-33 orchestrate development and maintenance of adipose tissue-resident regulatory T cells
.
Nat Immunol
2015
;
16
:
276
85
.
19.
Ahmadzadeh
M
,
Pasetto
A
,
Jia
L
,
Deniger
DC
,
Stevanovic
S
,
Robbins
PF
, et al
.
Tumor-infiltrating human CD4(+) regulatory T cells display a distinct TCR repertoire and exhibit tumor and neoantigen reactivity
.
Sci Immunol
2019
;
4
:
eaao4310
.
20.
Gobert
M
,
Treilleux
I
,
Bendriss-Vermare
N
,
Bachelot
T
,
Goddard-Leon
S
,
Arfi
V
, et al
.
Regulatory T cells recruited through CCL22/CCR4 are selectively activated in lymphoid infiltrates surrounding primary breast tumors and lead to an adverse clinical outcome
.
Cancer Res
2009
;
69
:
2000
9
.
21.
Facciabene
A
,
Peng
X
,
Hagemann
IS
,
Balint
K
,
Barchetti
A
,
Wang
LP
, et al
.
Tumour hypoxia promotes tolerance and angiogenesis via CCL28 and T(reg) cells
.
Nature
2011
;
475
:
226
30
.
22.
Righi
E
,
Kashiwagi
S
,
Yuan
J
,
Santosuosso
M
,
Leblanc
P
,
Ingraham
R
, et al
.
CXCL12/CXCR4 blockade induces multimodal antitumor effects that prolong survival in an immunocompetent mouse model of ovarian cancer
.
Cancer Res
2011
;
71
:
5522
34
.
23.
Magnuson
AM
,
Kiner
E
,
Ergun
A
,
Park
JS
,
Asinovski
N
,
Ortiz-Lopez
A
, et al
.
Identification and validation of a tumor-infiltrating Treg transcriptional signature conserved across species and tumor types
.
Proc Natl Acad Sci U S A
2018
;
115
:
E10672
E81
.
24.
Alonso
R
,
Flament
H
,
Lemoine
S
,
Sedlik
C
,
Bottasso
E
,
Peguillet
I
, et al
.
Induction of anergic or regulatory tumor-specific CD4(+) T cells in the tumor-draining lymph node
.
Nat Commun
2018
;
9
:
2113
.
25.
Wang
L
,
Pino-Lagos
K
,
de Vries
VC
,
Guleria
I
,
Sayegh
MH
,
Noelle
RJ
.
Programmed death 1 ligand signaling regulates the generation of adaptive Foxp3+CD4+ regulatory T cells
.
Proc Natl Acad Sci U S A
2008
;
105
:
9331
6
.
26.
Worbs
T
,
Bode
U
,
Yan
S
,
Hoffmann
MW
,
Hintzen
G
,
Bernhardt
G
, et al
.
Oral tolerance originates in the intestinal immune system and relies on antigen carriage by dendritic cells
.
J Exp Med
2006
;
203
:
519
27
.
27.
Coombes
JL
,
Siddiqui
KR
,
Arancibia-Carcamo
CV
,
Hall
J
,
Sun
CM
,
Belkaid
Y
, et al
.
A functionally specialized population of mucosal CD103+ DCs induces Foxp3+ regulatory T cells via a TGF-beta and retinoic acid-dependent mechanism
.
J Exp Med
2007
;
204
:
1757
64
.
28.
Okabe
Y
,
Medzhitov
R
.
Tissue-specific signals control reversible program of localization and functional polarization of macrophages
.
Cell
2014
;
157
:
832
44
.
29.
Roane
BM
,
Meza-Perez
S
,
Katre
AA
,
Goldsberry
WN
,
Randall
TD
,
Norian
LA
, et al
.
Neutralization of TGFbeta improves tumor immunity and reduces tumor progression in ovarian carcinoma
.
Mol Cancer Ther
2021
;
20
:
602
11
.
30.
Allie
SR
,
Bradley
JE
,
Mudunuru
U
,
Schultz
MD
,
Graf
BA
,
Lund
FE
, et al
.
The establishment of resident memory B cells in the lung requires local antigen encounter
.
Nat Immunol
2019
;
20
:
97
108
.
31.
McCaw
TR
,
Li
M
,
Starenki
D
,
Cooper
SJ
,
Liu
M
,
Meza-Perez
S
, et al
.
The expression of MHC class II molecules on murine breast tumors delays T-cell exhaustion, expands the T-cell repertoire, and slows tumor growth
.
Cancer Immunol Immunother
2019
;
68
:
175
88
.
32.
Nazarov
VI
,
Pogorelyy
MV
,
Komech
EA
,
Zvyagin
IV
,
Bolotin
DA
,
Shugay
M
, et al
.
tcR: an R package for T cell receptor repertoire advanced data analysis
.
BMC Bioinformatics
2015
;
16
:
175
.
33.
Madi
A
,
Poran
A
,
Shifrut
E
,
Reich-Zeliger
S
,
Greenstein
E
,
Zaretsky
I
, et al
.
T cell receptor repertoires of mice and humans are clustered in similarity networks around conserved public CDR3 sequences
.
Elife
2017
;
6
:
e22057
.
34.
Van der Loo
MP
.
The stringdist package for approximate string matching
.
R J
2014
;
6
:
111
22
.
35.
Csardi
G
,
Nepusz
T
.
The igraph software package for complex network research
.
InterJournal, complex systems
2006
;
1695
:
1
9
.
36.
Gustavsen
JA
,
Pai
S
,
Isserlin
R
,
Demchak
B
,
Pico
AR
.
RCy3: Network biology using Cytoscape from within R
.
F1000Res
2019
;
8
:
1774
.
37.
Shannon
P
,
Markiel
A
,
Ozier
O
,
Baliga
NS
,
Wang
JT
,
Ramage
D
, et al
.
Cytoscape: a software environment for integrated models of biomolecular interaction networks
.
Genome Res
2003
;
13
:
2498
504
.
38.
Gu
Z
,
Gu
L
,
Eils
R
,
Schlesner
M
,
Brors
B
.
circlize Implements and enhances circular visualization in R
.
Bioinformatics
2014
;
30
:
2811
2
.
39.
Kim
D
,
Pertea
G
,
Trapnell
C
,
Pimentel
H
,
Kelley
R
,
Salzberg
SL
.
TopHat2: accurate alignment of transcriptomes in the presence of insertions, deletions and gene fusions
.
Genome Biol
2013
;
14
:
R36
.
40.
Li
H
,
Handsaker
B
,
Wysoker
A
,
Fennell
T
,
Ruan
J
,
Homer
N
, et al
.
The sequence alignment/map format and SAMtools
.
Bioinformatics
2009
;
25
:
2078
9
.
41.
Anders
S
,
Pyl
PT
,
Huber
W
.
HTSeq–a Python framework to work with high-throughput sequencing data
.
Bioinformatics
2015
;
31
:
166
9
.
42.
Love
MI
,
Huber
W
,
Anders
S
.
Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2
.
Genome Biol
2014
;
15
:
550
.
43.
Subramanian
A
,
Tamayo
P
,
Mootha
VK
,
Mukherjee
S
,
Ebert
BL
,
Gillette
MA
, et al
.
Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles
.
Proc Natl Acad Sci U S A
2005
;
102
:
15545
50
.
44.
Kramer
A
,
Green
J
,
Pollard
J
Jr
,
Tugendreich
S
.
Causal analysis approaches in Ingenuity Pathway Analysis
.
Bioinformatics
2014
;
30
:
523
30
.
45.
Bolotin
DA
,
Poslavsky
S
,
Davydov
AN
,
Frenkel
FE
,
Fanchi
L
,
Zolotareva
OI
, et al
.
Antigen receptor repertoire profiling from RNA-seq data
.
Nat Biotechnol
2017
;
35
:
908
11
.
46.
Thornton
AM
,
Korty
PE
,
Tran
DQ
,
Wohlfert
EA
,
Murray
PE
,
Belkaid
Y
, et al
.
Expression of Helios, an Ikaros transcription factor family member, differentiates thymic-derived from peripherally induced Foxp3+ T regulatory cells
.
J Immunol
2010
;
184
:
3433
41
.
47.
Weiss
JM
,
Bilate
AM
,
Gobert
M
,
Ding
Y
,
Curotto de Lafaille
MA
,
Parkhurst
CN
, et al
.
Neuropilin 1 is expressed on thymus-derived natural regulatory T cells, but not mucosa-generated induced Foxp3+ T reg cells
.
J Exp Med
2012
;
209
:
1723
42
, S1.
48.
Yadav
M
,
Louvet
C
,
Davini
D
,
Gardner
JM
,
Martinez-Llordella
M
,
Bailey-Bucktrout
S
, et al
.
Neuropilin-1 distinguishes natural and inducible regulatory T cells among regulatory T cell subsets in vivo
.
J Exp Med
2012
;
209
:
1713
22
.
49.
Sefik
E
,
Geva-Zatorsky
N
,
Oh
S
,
Konnikova
L
,
Zemmour
D
,
McGuire
AM
, et al
.
MUCOSAL IMMUNOLOGY. Individual intestinal symbionts induce a distinct population of RORgamma(+) regulatory T cells
.
Science
2015
;
349
:
993
7
.
50.
Ohnmacht
C
,
Park
JH
,
Cording
S
,
Wing
JB
,
Atarashi
K
,
Obata
Y
, et al
.
MUCOSAL IMMUNOLOGY. The microbiota regulates type 2 immunity through RORgammat(+) T cells
.
Science
2015
;
349
:
989
93
.
51.
Toomer
KH
,
Lui
JB
,
Altman
NH
,
Ban
Y
,
Chen
X
,
Malek
TR
.
Essential and non-overlapping IL-2Ralpha-dependent processes for thymic development and peripheral homeostasis of regulatory T cells
.
Nat Commun
2019
;
10
:
1037
.
52.
Komatsu
N
,
Hori
S
.
Full restoration of peripheral Foxp3+ regulatory T cell pool by radioresistant host cells in scurfy bone marrow chimeras
.
Proc Natl Acad Sci U S A
2007
;
104
:
8959
64
.
53.
Feuerer
M
,
Herrero
L
,
Cipolletta
D
,
Naaz
A
,
Wong
J
,
Nayer
A
, et al
.
Lean, but not obese, fat is enriched for a unique population of regulatory T cells that affect metabolic parameters
.
Nat Med
2009
;
15
:
930
9
.
54.
Pacella
I
,
Procaccini
C
,
Focaccetti
C
,
Miacci
S
,
Timperi
E
,
Faicchia
D
, et al
.
Fatty acid metabolism complements glycolysis in the selective regulatory T cell expansion during tumor growth
.
Proc Natl Acad Sci U S A
2018
;
115
:
E6546
E55
.
55.
Li
C
,
DiSpirito
JR
,
Zemmour
D
,
Spallanzani
RG
,
Kuswanto
W
,
Benoist
C
, et al
.
TCR transgenic mice reveal stepwise, multi-site acquisition of the distinctive fat-Treg phenotype
.
Cell
2018
;
174
:
285
99
.
56.
Krishnan
V
,
Tallapragada
S
,
Schaar
B
,
Kamat
K
,
Chanana
AM
,
Zhang
Y
, et al
.
Omental macrophages secrete chemokine ligands that promote ovarian cancer colonization of the omentum via CCR1
.
Commun Biol
2020
;
3
:
524
.
57.
Muller
G
,
Hopken
UE
,
Lipp
M
.
The impact of CCR7 and CXCR5 on lymphoid organ development and systemic immunity
.
Immunol Rev
2003
;
195
:
117
35
.
58.
Cyster
JG
,
Schwab
SR
.
Sphingosine-1-phosphate and lymphocyte egress from lymphoid organs
.
Annu Rev Immunol
2012
;
30
:
69
94
.
59.
Tan
MC
,
Goedegebuure
PS
,
Belt
BA
,
Flaherty
B
,
Sankpal
N
,
Gillanders
WE
, et al
.
Disruption of CCR5-dependent homing of regulatory T cells inhibits tumor growth in a murine model of pancreatic cancer
.
J Immunol
2009
;
182
:
1746
55
.
60.
Jackson
JJ
,
Ketcham
JM
,
Younai
A
,
Abraham
B
,
Biannic
B
,
Beck
HP
, et al
.
Discovery of a potent and selective CCR4 antagonist that inhibits treg trafficking into the tumor microenvironment
.
J Med Chem
2019
;
62
:
6190
213
.
61.
Sawant
DV
,
Yano
H
,
Chikina
M
,
Zhang
Q
,
Liao
M
,
Liu
C
, et al
.
Adaptive plasticity of IL-10(+) and IL-35(+) Treg cells cooperatively promotes tumor T cell exhaustion
.
Nat Immunol
2019
;
20
:
724
35
.
62.
Wang
H
,
Franco
F
,
Tsui
YC
,
Xie
X
,
Trefny
MP
,
Zappasodi
R
, et al
.
CD36-mediated metabolic adaptation supports regulatory T cell survival and function in tumors
.
Nat Immunol
2020
;
21
:
298
308
.
63.
Miao
ZF
,
Wang
ZN
,
Zhao
TT
,
Xu
YY
,
Gao
J
,
Miao
F
, et al
.
Peritoneal milky spots serve as a hypoxic niche and favor gastric cancer stem/progenitor cell peritoneal dissemination through hypoxia-inducible factor 1alpha
.
Stem Cells
2014
;
32
:
3062
74
.
64.
Nagao
A
,
Kobayashi
M
,
Koyasu
S
,
Chow
CCT
,
Harada
H
.
HIF-1-dependent reprogramming of glucose metabolic pathway of cancer cells and its therapeutic significance
.
Int J Mol Sci
2019
;
20
:
238
.

Supplementary data