Canonical Secretomes, Innate Immune Caspase-1-, 4/11-Gasdermin D Non-Canonical Secretomes and Exosomes May Contribute to Maintain Treg-Ness for Treg Immunosuppression, Tissue Repair and Modulate Anti-Tumor Immunity via ROS Pathways

We performed a transcriptomic analyses using the strategies we pioneered and made the following findings: 1) Normal lymphoid Tregs, diseased kidney Tregs, splenic Tregs from mice with injured muscle have 3, 17 and 3 specific (S-) pathways, respectively; 2) Tumor splenic Tregs share 12 pathways with tumor Tregs; tumor splenic Tregs and tumor Tregs have 11 and 8 S-pathways, respectively; 3) Normal and non-tumor disease Tregs upregulate some of novel 2641 canonical secretomic genes (SGs) with 24 pathways, and tumor Tregs upregulate canonical secretomes with 17 pathways; 4) Normal and non-tumor disease tissue Tregs upregulate some of novel 6560 exosome SGs with 56 exosome SG pathways (ESP), tumor Treg ESP are more focused than other Tregs; 5) Normal, non-tumor diseased Treg and tumor Tregs upregulate some of novel 961 innate immune caspase-1 SGs and 1223 innate immune caspase-4 SGs to fulfill their tissue/SG-specific and shared functions; 6) Most tissue Treg transcriptomes are controlled by Foxp3; and Tumor Tregs had increased Foxp3 non-collaboration genes with ROS and 17 other pathways; 7) Immune checkpoint receptor PD-1 does, but CTLA-4 does not, play significant roles in promoting Treg upregulated genes in normal and non-tumor disease tissue Tregs; and tumor splenic and tumor Tregs have certain CTLA-4-, and PD-1-, non-collaboration transcriptomic changes with innate immune dominant pathways; 8) Tumor Tregs downregulate more immunometabolic and innate immune memory (trained immunity) genes than Tregs from other groups; and 11) ROS significantly regulate Treg transcriptomes; and ROS-suppressed genes are downregulated more in tumor Tregs than Tregs from other groups. Our results have provided novel insights on the roles of Tregs in normal, injuries, regeneration, tumor conditions and some of canonical and innate immune non-canonical secretomes via ROS-regulatory mechanisms and new therapeutic targets for immunosuppression, tissue repair, cardiovascular diseases, chronic kidney disease, autoimmune diseases, transplantation, and cancers.

We performed a transcriptomic analyses using the strategies we pioneered and made the following findings: 1) Normal lymphoid Tregs, diseased kidney Tregs, splenic Tregs from mice with injured muscle have 3, 17 and 3 specific (S-) pathways, respectively; 2) Tumor splenic Tregs share 12 pathways with tumor Tregs; tumor splenic Tregs and tumor Tregs have 11 and 8 S-pathways, respectively; 3) Normal and non-tumor disease Tregs upregulate some of novel 2641 canonical secretomic genes (SGs) with 24 pathways, and tumor Tregs upregulate canonical secretomes with 17 pathways; 4) Normal and nontumor disease tissue Tregs upregulate some of novel 6560 exosome SGs with 56 exosome SG pathways (ESP), tumor Treg ESP are more focused than other Tregs; 5) Normal, non-tumor diseased Treg and tumor Tregs upregulate some of novel 961 innate immune caspase-1 SGs and 1223 innate immune caspase-4 SGs to fulfill their tissue/SGspecific and shared functions; 6) Most tissue Treg transcriptomes are controlled by Foxp3; and Tumor Tregs had increased Foxp3 non-collaboration genes with ROS and 17 other pathways; 7) Immune checkpoint receptor PD-1 does, but CTLA-4 does not, play significant roles in promoting Treg upregulated genes in normal and non-tumor disease tissue Tregs; and tumor splenic and tumor Tregs have certain CTLA-4-, and PD-1-, non-
Foxp3 is the major transcription factor (TF), and coexpression of lineage-specifying transcription factors alters the potential function and flexibility of subsets of CD4 + T cell; this, in turn, favors the autoimmune pathology (86). Tregs are specialized in the suppression of immuno-pathological reactions in the host immune system against antigens and dangers (36). In addition to inhibition of adaptive immune response, Tregs also play a critical role in controlling various innate immune responses involved in cancers (15), inflammatory diseases including cardiovascular diseases and atherosclerosis (64,75). However, an important question remained whether Treg have innate immune response machinery. To address this question, we recently reported that Treg secretomes (in lymph nodes and spleen) and transcription factors shared with stem cells contribute to a Treg niche to maintain Treg-ness with 80% innate immune pathways, and functions of immunosuppression and tissue repair (87). Additionally, Tregs play a highly broad spectrum of versatile anti-pathophysiological roles. For example, Tregs facilitate blood flow recovery after ischemia (88), control adipose tissue inflammation, promote muscle repair (89) and maintain tissue/organ homeostasis (90). Treg's roles in maintaining self-tolerance and prevention of autoimmune responses and chronic inflammation are mediated by various mechanisms including: a) Treg killing of target cells (15); b) modulation of target cells via cell-cell contact; c) inhibition of target cells by exosome-carried microRNAs (36); and d) secretion of anti-inflammatory/immunosuppressive cytokines (38) including IL-10, IL-35 (30,71,72), and TGF-b. Therefore, cellular therapies using Tregs are currently undergoing clinical trials for the treatment of autoimmune diseases, transplant rejection and graft-versus-host disease (91).
In addition to the secretion of anti-inflammatory/ immunosuppressive cytokines (38), including IL-10, IL-35 (30,71,72), and TGF-b, Treg play a highly tissue-and contextspecific roles in angiogenesis either pro-or anti-angiogenic effects (92). Similarly, we recently reported that Treg-secreted IL-35 delays hindlimb ischemia-induced angiogenesis through regulating ROS-extracellular matrix but spares later regenerative angiogenesis (73), indicating that Treg-secreted IL-35 plays roles in a pathological process/phase-specific manner (30,37,44,45,(71)(72)(73). Recent progress suggests that Treg secretory proteins could be also much bigger than a list of cytokines and chemokines, since the secretomes from various cell types have been reported (93,94). The secretome, defined as a portion of total proteins secreted by cells to the extracellular space, secures a proper micro-environmental niche, thus maintaining tissue homeostasis (95,96). Secreted molecules are key mediators in cell-cell interactions, via autocrine, and paracrine manners, and influence the cross-talk with the surrounding tissues in addition to their endocrine functions in long-distance by hormones, growth factors, cytokines, adipokines, myokines, cardiokines (97), and chemokines (98). There is strong evidence supporting that crucial cellular functions such as proliferation, differentiation, communication and migration are regulated strictly by the cell secretome (99). Among the secretomes, we recently reported canonical secretomes of human peripheral blood mononuclear cells (28). Similarly, Treg from mouse spleen, lymph nodes, intestine and visceral adipose tissue may use canonical secretomes (all the human proteins with signal peptide sequence) to fulfill their functions (87). In addition, exosome secretomes, due to their content enriched in proteins, mRNAs and noncoding RNAs, carry out cell-cell communication and promote tissue regeneration, wound healing, extracellular matrix remodeling, immunomodulation (100), angiogenesis, anti-apoptotic activity and cell migration, proliferation and differentiation (101). Moreover, we reported that Treg may use their secretomes shared with stem cells to maintain their Treg-ness with 80% innate immune pathways (87), thus, Treg may also use innate immunity-related secretomes such as caspase-1-gasdermin D (GSDMD) dependent secretome (102), and caspase-4 (humans)/caspase-11 (mice)-GSDMD dependent secretome (103), to carry out their functions. However, it remains unknown that Treg may use canonical secretomes, non-canonical secretomes including exosome secretomes (100), innate immune caspase-1-GSDMD secretomes and innate immune caspase-4/11-GSDMD secretomes for their immune regulatory functions (100).
However, several important knowledge gaps remain: 1) are there differences in Treg transcriptomes between Tregs from normal lymphoid and non-lymphoid tissues, injured tissues, regenerative tissues, tissues from mouse carrying tumor, tumor tissues; 2) are secretomes generated in Tregs in various tissues and pathological conditions; and 3) how Treg transcription factor Foxp3, immune checkpoint receptors cytotoxic T lymphocyte associate protein-4 (CTLA-4, CD152) and programmed cell death-1 (PD-1, CD279) (19), immunometabolic pathways and innate immune memory (trained immunity) and ROS pathways regulate tissue Treg heterogeneity and Treg various secretomes. The major differences between our current study and previous reports on the roles of cytokines and chemokines in Treg are that secretome analyses provide a panoramic view on all the secreted genes in Treg, as opposed to focusing on only one or a few cytokines/chemokines (28).
An overview on the connection logic of the manuscript sections and structure in organizing this manuscript was provided as follows. In the first major part (the Results sections 1, 2, 3, and 4), we compared normal lymphoid tissue Treg, non-tumor diseased kidney Treg and splenic Treg in mouse with injured muscle in the Section 1. In the Section 2, we extended our analysis to examine splenic Treg in mouse with tumors and Treg from tumor tissues. In the Section 3, we summarized all the findings in Treg pathways in normal tissues, non-tumor diseased organs and mouse with injured muscles for comparison. In the Section 4, we identified the differences between tumor Treg and Treg from normal tissues, diseased organs and spleen from the mouse with tumor. Taken together, the sections 1-4 as the first part focused on the transcriptomic heterogeneity of Treg in healthy environments, non-tumor disease conditions and tumors. Treg suppressive functions to inhibit other immune cells depends on several effector mechanisms including secretion of anti-inflammatory/ immunosuppressive cytokines such as IL-10, IL-35 and transforming growth factor-b (TGF-b). The second major part in the sections 5, 6 and 7 focused on examining how tissue Treg heterogeneity from normal tissues, non-tumor disease conditions and tumors affect Treg secretory functions; and how Treg secretomes could also potentially affect tissue Treg heterogeneity. We attempted for the first time to introduce all the recent cellular protein secretomic findings to Treg field by examining canonical secretomes with secretory proteins carrying signal peptide, non-canonical secretomes with secretory proteins with no signal peptide such as IL-1b including caspase-1 dependent secretome, caspase-4 dependent secretome and exosome-based secretome. The third major part of this manuscript focused on examining potential molecular mechanisms underlying the Treg heterogeneity and secretomic function differences identified in the first and second major parts from four different angles. These four angles (mechanisms) include Treg-specific transcription factor Foxp3 (nuclear) dependence in the Section 8, major Treg-related immune checkpoint receptors (T cell co-inhibition receptors) on the cell membrane such as PD-1-, CTLA-4-signaling dependence in the Section 9, intracellular immune metabolism/trained immunity (innate immune memory) regulation in the Section 10, and intracellular reactive oxygen species (ROS, a newly reported metabolic sensor system) regulation in the Section 11. Although significant amounts of work in the future are required to verify the -omic data analyses-based findings presented in this paper, similar to other papers with the data generated with -omics such as RNA-sequencing (RNA-Seq) and single cell-RNA-Seq (scRNA-Seq), our results to address the three major questions provide novel insights on tissue Treg heterogeneity in normal, diseased/injured, regenerative, and tumors, Treg secretomes and immunometabolic/trained immunity and ROS regulation of tissue Tregs.

Statistical Analysis of Microarray Data
We applied a statistical method similar to our previously reported meta-analysis (28,50,104). We designed a robust house-keeping gene list (Supplemental Table 1 of housekeeping genes) with help from Eisenberg and Levanon's (105) excellent work, including ACTB, GAPDH, PGK1, PPIA, B2M, YWHAZ, SDHA, HMBS, and TBP. Briefly, the mean log fold change (LogFC) of house-keeping genes between treatment and control groups vary from -1.27 to 1.28. The target genes with expression changes more than 2-folds were defined as the upregulated genes, while genes with their expression decreased more than 2-folds were defined as downregulated genes |logFC|>1).

Ingenuity Pathway Analysis and Metascape Analysis
We utilized Ingenuity Pathway Analysis (IPA, Qiagen, https:// www.qiagenbioinformatics.com/products/ingenuity-pathwayanalysis/) to characterize clinical relevance and molecular and cellular functions related to the identified genes in our microarray analysis. Differentially expressed genes were identified and uploaded into IPA for analysis. The core and pathways analysis was used to identify molecular and cellular pathways, as we have previously reported (28,104,106). For the groups with small numbers of genes modulated, the Metascape analysis (https://metascape.org/gp/index.html#/main/step1) was used for determining signaling pathways involved. We recently reported that Treg-specific transcription factor Foxp3 is expressed in trachea, thymus, spleen, mammary gland, lymph node, lung, eye, and blood (16). Six aspects involved in development, homeostasis, function, differentiation, and plasticity of Treg are modulated by 15 cytokines including  IL-2, IL-15, IL-7, TGF-b, TNF-a, IL-33, IL-4, IL-1b, IL-6, IL-21,  IL-23, IL-12, IL-27, IL-35 (71) and IFNg (82). Since these cytokine expression levels are varied among tissues, thus, various Treg populations in these tissues have been identified (90), which are similar to what we reported on tissue macrophages that 20 novel disease group-specific and 12 new shared macrophage pathways have been found in eight groups of 34 diseases including 24 inflammatory organ diseases and 10 types of tumors (1). We hypothesized that tissue Treg have different signaling pathways and Treg from non-malignant diseases have also different signaling pathways. To test this hypothesis, we collected 20 Treg microarrays and RNA-Seq datasets from two comprehensive databases such as NIH-NCBI-Geo Datasets database (https://www.ncbi.nlm.nih.gov/ geo/) and EMBL-EBI Arrayexpress (https://www.ebi.ac.uk/ arrayexpress/). As shown in Table 1, 20 datasets included one from spleen, one from lymph nodes (LN), one from brown adipose tissue (BAT), one from kidney, two from spleens in mouse with injured skeletal muscles for four days and two weeks, respectively, two from mouse injured skeletal muscles for four days and two weeks, respectively, one from kidney fibrosis, one from kidney regeneration, four from spleens of tumor bearing mouse (BL6, MC38, CT26 and TC-1), respectively, seven tumor tissue Treg from hepatocellular carcinoma, melanoma, lung tumor (TC-1), breast cancer, tumor bearing Bl6, tumor bearing MC38, and tumor bearing CT26, respectively. Of note, As the original paper reported (107), tumor infiltrating Treg and tumor splenic Treg were sorted as live, CD45+TCRb+CD4+Foxp3 +CD8-DUMP-). Tumor Treg were distinguished from tumor splenic Treg using RNA-Seq analysis (107). In addition, one of our recent manuscripts mainly focused on profile on Treg transcription markers, CD markers, kinome, of tissue Treg from mouse spleen, lymph nodes, small intestines and white adipose tissues, which are not overlapped in this study (87). As shown in Figure 2A two third of the datasets were collected from RNA-Seq, suggesting that new techniques were used. Total differentially expressed genes (DEGs) were varied among Treg datasets, ranging from 113 (brown adipose tissue Treg) to 2608 FIGURE 1 | The flow chart of CD4+Foxp3+ regulatory T cell (Treg) transcriptomic and secretomic analyses from normal tissues, diseased tissues and tumor tissues. Two most comprehensive gene expression databases were used in this study: NIH-NIBI-GEO: National Institutes of Health (NIH)-National Center for Biotechnology Information (NCBI)-Gene Expression Omnibus (GEO) databases https://www.ncbi.nlm.nih.gov/gds/; EMBL-EBI: European Molecular Biology Laboratory (EMBL)-European Bioinformatics Institute(EBI) https://www.ebi.ac.uk/arrayexpress/. The gene expression analyses were performed by using the bioinformatic tool/database GEO2R: https://www.ncbi.nlm.nih.gov/geo/geo2r/R: version 3.6.1. (regeneration kidney Treg). The upregulated genes in total DEGs among Treg datasets were ranging from 31.8% (melanoma skin Treg) to 85.0% (brown adipose tissue Treg). As shown in Figure 2B, there were more downregulated genes in Treg from tumor tissue groups than normal tissue Treg, diseased tissue Treg groups and tumor splenic Treg groups. The Ingenuity Pathway Analyses were performed based on the DEGs in Treg in comparison to that of CD4 + Foxp3 -T effector cells (Teff). As shown in Figure 2C, Treg from lymphoid tissues, LN and spleen, shared six pathways including Th2 pathway, cardiac hypertrophy signaling, TREM1 signaling, nitric oxide and reactive oxygen species, thrombin signaling and endothelin signaling. Treg from BAT (108) and kidney shared upregulation of two pathways, including T cell exhaustion signaling and IL-8 signaling, and downregulation of Th1 pathway. Of note, it has been reported that visceral adipose tissue (VAT) Treg are Foxp3 + peroxisome proliferator-activated receptor-g (PPARg) + Treg with increased lipid metabolism (109), which are enhanced by IL33-ST2 axis (110); and among 194 significantly upregulated genes in BAT Treg, 148 genes are identified as VAT Treg-specific, whereas only 10 genes are BAT Treg-specific (108). Interestingly, Treg from spleen shared five pathways with kidney, such as phospholipase C signaling, phosphoinositide 3-kinase (PI3K) signaling, sperm motility, adrenomedullin signaling, and IL-15 production. As shown in Figure 2D, Treg from the spleens of mouse with skeletal muscle injury 2 weeks, the spleen of mouse with skeletal muscle injury 4 days, kidney fibrosis, kidney regeneration shared PI3K signaling. Treg from the spleens of mouse with skeletal muscle injury 2 weeks, the spleen of mouse with skeletal muscle injury 4 days, kidney regeneration, and mouse skeletal muscle injury 2 weeks shared T cell exhaustion signaling. Treg from the spleen of mouse with skeletal muscle injury 4 days, kidney regeneration shared four pathways such as synaptic long term depression, cardiac hypertrophy, chemokine signaling and thrombin signaling. Two pathways such as TREM1 signaling and Th2 pathway showed in Treg from the spleens of mouse with skeletal muscle injury 2 weeks, and the spleen of mouse with skeletal muscle injury 4 days. Of note, it has been reported that skeletal muscle Treg promote skeletal muscle regeneration by producing amphiregulin (Areg), to member of the epidermal growth factor family whose receptor is expressed on satellite cells in the skeletal muscles (110). Other two pathways such as role of NFAT and Toll-like receptor signaling showed in Treg from the spleens of mouse with skeletal muscle injury 4 days and kidney fibrosis. Of note, Treg play significant roles in maintaining normal kidney and suppressing acute ischemic kidney injury, preconditioning prevention of ischemia reperfusion injury, inhibition of cisplatin-induced renal injury and adrianmycin nephropathy, suppression of autoimmune glomerulopathies, lupus glomerulonephritis, nephrotoxic nephritis and diabetic kidney disease (111). In addition, 17 pathways were shared by Treg from kidney fibrosis and kidney regeneration. These kidney Tregspecific pathways include: reelin signaling, white adipose tissue browning, glioblastoma multiform, STAT3, IL-8, Rho GTPases, apelin liver, role of NFAT in cardiac hypertrophy, Wnt/Ca2 + , PDGF, cholecystokinin/gastrin, adrenomedullin, IL-15 production, estrogen receptor, IL-6, phospholipase C and endothelin-1. Moreover, three pathways including natural killer cells, systemic lupus and CD40 signaling were downregulated in Treg from skeletal muscle injury 2 weeks and skeletal muscle injury 4 days, which were muscle injury Treg-specific. Of note, although tissue Treg heterogeneity has been identified, Treg-specific signatures remain including Treg markers such as CD4+Foxp3+. These common Treg signatures allow us to study the effects of transcriptomic differences of tissue Treg, Treg from diseased conditions, tumor environments and tumor-carrying mouse on secretomic pathways as well as Treg niche-secretomes interactions. Similar approaches were reported by other investigators (110,112). Taken together, Treg from normal lymphoid tissues such as lymph nodes (LN) and spleen have five specific pathways, Treg from brown adipose tissue have no specific pathways, Treg from kidney fibrosis and kidney regeneration have 17 specific pathways, and Treg from spleen of mouse with injured skeletal muscle have three muscle injuryspecific pathways. We then hypothesized that Treg from spleens of mouse bearing tumors and Treg isolated from tumor tissues have specific signaling pathways. To test this hypothesis, the Ingenuity Pathway Analyses (IPA) were performed based on the DEGs in Treg in comparison to that of CD4 + Foxp3 -T effector cells (Teff). As shown in Figure 2E, 12 pathways were shared by Treg from spleens of mouse bearing tumors and Treg from tumor tissues including TREM1, Vitamin C transport, T cell exhaustion, endothelin-1, PI3K signaling, Th2 pathway, thrombin, kinetochore metaphase, pancreatic adenocarcinoma, phospholipase C, integrin, and osteoarthritis. Of note, the sources of Tregs were specified in Figure 2E and were also detailed in Table 1 Taken together, our results have demonstrated that first, Treg from spleens of mouse bearing tumors share 12 pathways with Treg from tumor tissues; second, Treg from spleens of mouse bearing tumors have 11 specific pathways, suggesting that tumorigenesis affect distal splenic Treg significantly; and third, Treg from tumor tissues have eight specific pathways. Our new findings are correlated with previous reports that Treg compartmentalization and trafficking may be tissue or/and organ specific, in which distinct chemokine receptor and integrin expression may contribute to selective retention and trafficking of Treg cells to sites where regulation is required (113). These Treg trafficking receptors include chemokine receptors and other G-protein coupled receptors, integrins, as well as selectins and their ligands. These receptors enable Treg to: 1) enter appropriate tissues from the bloodstream via postcapillary venules, 2) navigate these tissues to locally execute their immune-regulatory functions, and 3) seek out the right antigen-presenting cells to interact with in order to receive the signals that sustain their survival, proliferation, and functional activity (114).

Pathways Are Upregulated and 11 Pathways Are Downregulated in Tregs From Normal Tissues, and Disease Tissues
We hypothesized that Treg from normal tissues and nonmalignant disease tissues have shared upregulated pathways and downregulated pathways. As shown in Figure 2F, 52 upregulated pathways were shared by Treg from normal tissues and non-malignant disease tissues and 11 pathways were downregulated in Treg from normal tissues and non-malignant disease tissues including Th1 pathway, RhoGDI signaling, GPCR-mediated integration of enteroendocrine signaling, PPAR signaling, antioxidant action of vitamin C, IL-23 signaling, glutamate receptor, PTEN, cross-talk between dendritic cells and natural killer cells, CD28 signaling, and Table 1 led to identification of differentially expressed genes (DEGs) in each dataset. The type of results from datasets enrolled including Array data and RNA-seq data. (Cut off: p < 0.05; |Log 2 FC| ≥ 1). (B) There were more suppressed DEGs of Treg in Tumor tissues than other 3 groups. The mean percentages of up-regulated genes in tumor tissue Treg groups were 48.22 ± 5.004, lower than that in normal, benign, and tumor spleen groups (67.71 ± 9.806, 64.20 ± 4.891, 70.04 ± 1.458, respectively). *Significantly changed. (C) The majority (73.8%, 59/80) of enriched pathways in normal tissue Treg groups were tissue Treg-specific in lymphoid and non-lymphoid tissues. The ingenuity Pathway Analysis (IPA, Qiagen, https://www.qiagenbioinformatics.com/products/ingenuity-pathway-analysis/) using the differentially expressed gene (DEG) from datasets enrolled in this study showed the activated (↑) and inhibited (↓) canonical pathways in normal tissue Treg groups (p < 0.05, |Z-score| >= 2). In addition, Venn diagram analysis was performed to find the commonly shared pathways in this group. The specifically enriched pathways in GSE37532 Lymph node (LN), E-MTAB-7961 spleen, and E-MTAB-7961 Kidney were shown in Supplementary Table 3A. In the normal Treg groups, there were a total of 80 pathways enriched including 64 activated and 16 inhibited. There were no dual regulated pathways in normal Treg groups. The most common shared pathway was Th2 Pathway, which was activated in 71.2% (5/7) datasets. TREM1 Signaling, Production of Nitric Oxide and Reactive Oxygen Species in Macrophages, Cardiac Hypertrophy Signaling (Enhanced), and Thrombin Signaling where activated only in Treg from lymphoid tissues. Whereas Th1 Pathway and tRNA Splicing were inhibited, and IL-8 Signaling was activated only in Treg in non-lymphoid tissues. (D) Non-lymphoid tissue Treg enriched more specific pathways than lymphoid tissue Treg in benign disease groups. (85 specific pathways in 4 non-lymphoid Treg datasets vs. 2 specific pathways in 2 lymphoid Treg datasets). IPA analysis using the DEG from datasets enrolled in this study showed the activated (↑) and inhibited (↓) canonical pathways in different benign disease tissue Treg groups (p < 0.05, |Z-score| >= 2). Venn diagram analysis was performed to find the commonly shared pathways in this groups. The specifically enriched pathways in various Treg groups were showed in Supplementary Table 3B. A total of 124 pathways were enriched in benign disease Treg groups, including 90 activated and 30 inhibited pathways. Cardiac Hypertrophy Signaling (Enhanced), Hepatic Fibrosis Signaling Pathway, Reactive Oxygen Species in Macrophages, and HGF Signaling were dual regulated pathways in non-lymphoid tissue Treg. The percentages of specifically enriched pathways in this group was 70.2% (87/124). 97.7% (85/87) specific pathways were in non-lymphoid tissue Treg datasets. The most commonly shared pathway in benign disease Treg groups were T Cell Exhaustion Signaling Pathway and PI3K Signaling in B Lymphocytes, activating in 67% (4/6) datasets. TREM1 Signaling and Th2 Pathway were only activated in spleen tissue Treg in this group. Whereas IL-23 Signaling Pathway, PPAR Signaling, Natural Killer Cell Signaling, Systemic Lupus Erythematosus In B Cell Signaling Pathway, and CD40 Signaling were inhibited in non-lymphoid tissue Treg. (E) Malignant Treg groups enriched more inhibited and dual pathways than normal and benign disease Treg groups. IPA analysis using the DEG from Treg datasets enrolled in this study showed the activated (↑) and inhibited (↓) canonical pathways in different malignant disease tissue Treg groups (p < 0.05, |Z-score| >= 2). Venn diagram analysis was performed to find the commonly shared pathways in this Treg group. The specifically enriched pathways in each dataset were showed in Supplementary Table 3C. There were a total of 128 pathways enriched in the malignant Treg groups, including 62 activated, 50 inhibited, and 16 dual regulated pathways. There were 19 specifically enriched pathways in 4 lymphoid Treg datasets (mean 4.8 per dataset), 52 specifically enriched pathways in 6 non-lymphoid Treg datasets (mean 8.7 per dataset). TREM1 Signaling and Vitamin-C Transport were the most commonly shared pathways. In addition, TREM1 Signaling was activated in all the lymphoid tissue Treg and dual regulated in the non-lymphoid tissue Treg. Vitamin-C Transport was activated both in lymphoid and non-lymphoid tissue Treg. (F) The Dataset GSE50096 Treg from skeletal muscle injured 4d enriched more down regulated pathways than normal and other benign disease tissue Treg. The pathways enriched in normal lymphoid and non-lymphoid tissue Treg were compared with benign tissue Treg. The 5 pathways all of which were activated in normal tissue Treg were suppressed in Treg from skeletal muscle injured 4d and 2w tissues. All the suppressed pathways in normal tissue Treg groups were also suppressed or non-statistically enriched in benign disease tissue Treg. (G) Malignant tissue Treg enriched more down regulated pathways than normal tissue Treg. The Pathways enriched in normal lymphoid and non-lymphoid tissue Tregs were compared with malignant tissue Treg. The 13 pathways all of which were activated in normal tissue Tregs were suppressed in at least one malignant Treg datasets, more specifically, in malignant non-lymphoid tissue Treg. However, only one of the 11 suppressed pathways in normal tissue Treg, PTEN Signaling, was activated in GSE116347 CT26 dataset.  iCOS-iCOSL signaling. Of note, the sources of Tregs were specified in Figure 2F and were also detailed in Table 1.

FIGURE 2 | (A) Comparisons of regulatory T cells (Treg) versus conventional T cells (Tconv) in tissues enrolled in
Previous report showed that systemic ablation of Treg compromises the adaptation of whole body energy expenditure to cold exposure, correlating with impairment in thermogenic gene expression and massive invasion of proinflammatory macrophages in BAT (108). The four pathways were downregulated only in Treg from BAT including glutamate receptor, cross-talk between dendritic cells and natural killer cells, CD28 signaling, and iCOS-iCOSL signaling. Our results may suggest that first, thermogenic environment in BAT may play significant roles in downregulating those four pathways; and second, these pathway downregulation in BAT Treg may play important roles. We then hypothesized that Treg from injured tissues have pathway changes from that of normal tissues. To examine this hypothesis, we compared four datasets from Treg in normal tissues such as LN, spleen, BAT and kidney to that of six Treg datasets including spleen Treg of mouse with injured muscles for 2 weeks and Treg from mouse with injured muscles for four days, kidney fibrosis, kidney regeneration, Treg from injured muscle for two weeks and Treg from injured muscle for four days. As shown in Figure 2F, five pathways were upregulated in normal tissues but downregulated in injured muscles including cardiac hypertrophy signaling (enhanced), production of nitric oxide and reactive oxygen species, xenobiotic metabolism signaling, role of pattern recognition receptors, and dermatan sulfate biosynthesis.
Taken together, our results have demonstrated that fifty two pathways are upregulated in Treg from normal lymphoid, nonlymphoid tissues and non-malignant diseases; eleven pathways are downregulated in Treg from normal lymphoid, non-lymphoid tissues and non-malignant diseases; four pathways are downregulated only in Treg from BAT; and five pathways upregulated in Treg from normal tissues are downregulated in injured muscles. It has been reported that Treg depletion alters the tumor microenvironment and accelerates pancreatic carcinogenesis (115), suggesting that Treg play critical roles in controlling tumor growth. We hypothesized that Treg transcriptomes from normal tissues are different from that of Treg from tumor tissues; and some pathways in Treg from normal tissues are different from that in malignant tissues (116). As shown in Figure 2B, upregulated gene numbers in Treg from tumor tissues in comparison to that Teff cells in the same tissues were significantly lower than that of Treg from normal tissues, nonmalignant diseases and spleens of mouse with tumors. These results have demonstrated that Treg transcriptomes undergo significant remodeling in tumor tissues.
As shown in Table 2A, canonical secretomic genes were upregulated in splenic Tregs from skeletal muscle injured for 2 weeks (48 genes) and skeletal muscle injured for 4 days (47 genes), fibrosis kidney (179 genes), regeneration kidney (265 genes), Treg from skeletal muscle injured for 2 weeks (74 genes), and Treg from skeletal muscle injured for 4 days (49 genes), respectively. Of note, the sources of Tregs were specified in the Table 2A and were also detailed in Table 1. These results showed that splenic Treg from injured skeletal muscles increased more canonical secretomic genes compared to normal splenic Treg ( Figure 4); Treg from injured skeletal muscles (2 weeks) increased more canonical secretomic genes compared to that of Treg from injured skeletal muscles (4 days); Treg from regeneration kidney increased more canonical secretomic genes compared to that of Treg fibrosis kidney. In the benign disease Treg group as shown in Table 2B, a total of 24 activated pathways were enriched in upregulated secretomic genes. P38 MAPK signaling were shared in five Treg groups including splenic Treg (injured muscle for 2 weeks), splenicTreg (injured muscle for 4 days), skeletal muscle Treg (injured muscle for 2 weeks), skeletal muscle Treg (injured for 4 days), and Treg from fibrosis kidney. STAT3 pathway was shared in two kidney Treg sets. IL-6 signaling was increased in skeletal muscle Treg (injured for 4 days). Four of the canonical secretome pathways were specific in E-MTAB-7961 Treg from kidney regeneration dataset, including hepatic fibrosis signaling pathway, PDGF signaling, VEGF signaling, and VEGF family ligand-receptor interactions, which may carry functions of tissue repair. IL-6 signaling was specifically enriched in GSE50096 Treg from muscle injury 4 days dataset, which was in the acute reaction phase after muscle injury. Table 2B, canonical secretomic genes were upregulated in Treg from tumor bearing spleen (B16, 125 genes), Treg from tumor bearing spleen (MC38, 147 genes), Treg from tumor bearing spleen (CT26, 103 genes), Treg from tumor bearing spleen (TC-1, 194 genes), Treg from melanoma skin (71 genes), Treg from TC-1 tumor (161 genes), Treg from breast tumor (83 genes), Treg from B16 tumor (173 genes), Treg from MC38 tumor (143 genes) and Treg from CT26 tumor (61 genes), respectively. Of note, the sources of Tregs were specified in the Table 2B and were also detailed in Table 1. A total of 17 activated canonical secretome pathways were enriched in Treg from malignant groups. The most common shared pathways were p38 MAPK signaling and IL-6 signaling. In addition, TREM1 signaling was shared in splenic Treg from MC38 tumor bearing mouse, and Treg from TC-1 lung tumor. Tolllike receptor signaling was shared by splenic Treg from mouse bearing MC38 tumor and splenic Treg from mouse bearing TC-1 lung tumor. Specifically enriched pathways in malignant spleen and tissues were as following: Two pathways such as HMGB1 signaling and cardiac hypertrophy signaling (Enhanced) were enriched in GSE116347 B16 splenic Treg. One pathway, type I

A B
FIGURE 3 | (A) Several secretomic pathways of Treg may contribute to the compositions of extracellular space molecules from Treg. In all the datasets, the percentages of extracellular space molecules were varied from 6.38 to 13.27%, which were much higher than the percentages of cytokines. Therefore, numerous additional non-canonical secretome molecules were existed in the extracellular space. The mechanisms of canonical and non-canonical secretome transportation were shown in Figure 1B   specified in the Table 3A and were also detailed in Table 1. As shown in Table 3A, A total of 56 exosome pathways were enriched in normal group, including 22 shared and 34 specific pathways. Two pathways including synaptic long term depression and role of NFAT in regulation of the immune response were common in Treg from lymphoid tissues. Three pathways including cardiac hypertrophy signaling, sphingosine-1-phosphate signaling and Wnt/Ca+ pathway were common in Treg from non-lymphoid tissues. Production of nitric oxide and reactive oxygen species in macrophages was enriched in GSE37532 LN Treg. Six pathways including neuroinflammation signaling pathway, type I diabetes mellitus signaling, ERK5 signaling, NF-kB signaling, synaptic long term potentiation, and endothelin-1 signaling were enriched in E-MTAB-7961 splenic Treg. Three pathways including vitamin-C transport, FGF signaling, and senescence pathway were enriched in GSE37532 VAT Treg. 24 pathways enriched in E-MTAB-7961 kidney Treg were shown in supplementary Table 3A.
As shown in Table 3B, tissue Treg upregulated exosome secretomic genes including splenic Treg from injured skeletal          Table 3B and were also detailed in Table 1.As shown in Table 3B, A total of 111 exosome pathways were enriched in benign disease groups including 32 shared pathways and 79 specific pathways. Among them, 71 out of 79 specifically enriched pathway were in Treg from regeneration kidney and fibrosis kidney (E-MTAN-7961 dataset). Vitamin-C transport and Th2 pathway were enriched in splenic Treg from injured skeletal muscle for 2 weeks (GSE50096). TREM1 signaling was enriched in splenic Treg from injured skeletal muscle for 4 days (GSE50096). PTEN signaling was enriched in injured skeletal muscle Treg for 4 days (GSE50096). Three pathways including dendritic cell maturation, apelin endothelial signaling pathway, and bladder cancer signaling were enriched in Treg from fibrosis kidney (E-MTAB-7961), and 71 specifically enriched pathway in Treg from regeneration kidney (E-MTAB-7961) were shown in supplementary Table 3B. As shown in Table 3C, exosome secretomic genes were upregulated in splenic Treg from B16 tumor bearing mouse (328 genes), splenic Treg from MC38 tumor bearing mouse (331 genes), splenic Treg from CT26 tumor bearing mouse (236 genes), splenic Treg from TC-1 tumor bearing mouse (475 genes), Treg from melanoma skin (121 genes), Treg from TC-1 tumor (419 genes), Treg from breast tumor (204 genes), Treg from B16 tumor (572 genes), Treg from MC38 tumor (415 genes) and Treg from CT26 tumor (188 genes), respectively. Of note, the sources of Tregs were specified in the Table 3C and were also detailed in Table 1. As shown in Table 3C, a total of 69 exosome pathways were enriched in malignant groups, including 35 common shared and 34 specific pathways. The 42 out of 69 exosome pathways were in the lymphoid tissues. Two pathways including vitamin-C transport and endothelin-1 signaling were the most commonly enriched pathways. Eight pathways including cell cycle control of chromosomal replication, BEX2 signaling pathway, cholesterol biosynthesis III (via desmosterol), superpathway of cholesterol biosynthesis, cholesterol biosynthesis II (via 24,25-dihydrolanosterol), cholesterol biosynthesis I, and glycolysis I were only common shared in Treg from malignant non-lymphoid tissues.
Taken together, our results have demonstrated that First, in comparison to that shown in Tables 3A, B, Treg from tumor bearing spleens and tumor tissues in Table 3C upregulate more exosome secretomic genes but less pathways, suggesting that Treg pathways from tumors are more focused than Treg from non-malignant disease tissues; Second, two pathways including synaptic long term depression and role of NFAT in regulation of the immune response were common in normal lymphoid tissue; three pathways including cardiac hypertrophy signaling, sphingosine-1-phosphate signaling and Wnt/Ca+ pathway were common in Treg from non-lymphoid tissues; Third, Treg from non-malignant disease groups have much less shared pathways than normal tissue Treg and malignant tissue Treg; and Fourth, vitamin-C transport and endothelin-1 signaling were most common enriched pathways; and eight pathways were enriched in Treg from malignant non-lymphoid tissues in cholesterol synthesis and cell cycle controls.

Normal Tissue Tregs, Diseased Tissue Tregs and Tumor Tregs Upregulate Novel Innate Immune Caspase-1 Secretomic Genes and Innate Immune Caspase-4 Secretomic Genes to Fulfill Their Tissue-Specific, Secretomes-Specific Functions and Shared Functions
It has been reported that NLRP3 negatively regulate Treg differentiation through Kpna-2 mediated nuclear translocation (120); that TLR2 and TLR4 signalings induce inflammasome priming in Th1-like Treg (128); and that caspase-1 canonical inflammasome pathway and caspase-4 (humans)/caspase-11 (mice) non-canonical inflammasome pathway all cleave Nterminal Gasdermin D (123), which form a protein pore/ channel on the cell membrane to release proinflammatory cytokines such as IL-1b and IL-18. It has been also reported that   secretomic genes, respectively. Of note, the sources of Tregs were specified in the Table 3D and were also detailed in Table 1. The pathway analysis results (Table 3E and Figure 5A) showed that normal tissue Treg and non-tumor diseased tissue Treg shared seven upregulated caspase-1 secretomic gene pathways including pathways in cancer, vesicle organization, negative regulation of apoptotic signaling, myeloid leukocyte activation, striated muscle cell differentiation, cell morphogenesis, and supramolecular fiber organization. Treg from non-tumor diseased tissues and tumors shared three upregulated caspase-1 secretomic pathways including extracellular structure, response to reactive oxygen species, cellsubstrate junction. Tumor spleen Treg and tumor Treg shared one upregulated caspase-1 secretomic pathway, fluid shear stress and atherosclerosis. In addition, normal spleen Treg, normal tissue Treg, non-tumor diseased spleen Treg, non-tumor diseased tissue Treg, tumor spleen Treg and tumor Treg had 2, 13, 9, 11, and 12 specific upregulated caspase-1 secretomic pathways, respectively.
Taken together, our results have demonstrated that First, upregulations of caspase-1-dependent secretomic genes and caspase-4-dependent secretomic genes in Treg are more than downregulation of those secretomic genes; Second, upregulations of caspase-4-dependent secretomic genes in Treg often are more than upregulations of caspase-1 secretomic genes in Treg; Third, most of Treg caspase-1 secretomic gene pathways are different from that of Treg caspase-4 secretomic pathways ( Figure 5C), suggesting that Treg caspase-1 secretomic genes play different roles in Treg from that of caspase-4 secretomes; and Fourth, the majority of caspase-1 secretomic pathways and caspase-4 secretomic pathways are tissue Treg-specific. These results suggest that normal tissue Treg, non-tumor diseased tissue Treg and tumor Treg upregulate novel caspase-1 secretomic genes and caspase-4 secretomic genes to fulfill their tissue-specific, secretomes-specific Treg functions and shared Treg functions.

Most Tissue Treg Transcriptomes Are Controlled by Foxp3; Tumor Tregs Had Significantly Increased Foxp3 Non-Collaboration Genes With ROS and 17 Other Pathways
It has been reported that Foxp3 plays significant roles in Treg specific gene expressions (74), and induces 58 gene upregulation and 46 gene downregulation (129). We hypothesized that Foxp3 plays significant roles in promoting DEGs in Treg from normal tissues, non-tumor diseased tissues and tumor tissues. To test this hypothesis, we placed all the DEGs of Treg versus Teffector cells with Treg upregulated genes in the >1 log2 in the Y-axis, and Foxp3 depletion-suppressed genes in the >1 log2 in the X-axis ( Figure 6A). The argument was that if gene expressions are promoted by Foxp3, the genes will be showed in Foxp3 induced gene groups; otherwise, they will be showed in Foxp3 suppressed gene group. If any genes upregulated or downregulated in tissue Treg, non-tumor diseased Treg and tumor Treg are not collaborated with Foxp3 regulation, they will be placed in noncollaboration gene groups. As shown in Figure 6B, LN Treg, splenic Treg, intestine Treg, BAT Treg, VAT Treg, kidney Treg had 8, 11, 0, 2, 11, and 14 upregulated non-collaboration genes. In addition, Treg from injured skeletal muscle (2 weeks) spleen, Treg from injured skeletal muscle (4 days) spleen, Treg from fibrosis kidney, Treg from regeneration kidney, Treg from injured skeletal muscle (2 weeks), Treg from injured skeletal muscle (4 days) had 3, 4, 6, 14, 2, and 1 upregulated non-collaboration genes. Treg from B16 spleen, Treg from MC38 spleen, Treg from CT26 spleen, Treg from TC-1 spleen, Treg from melanoma, Treg from TC-1 tumor, Treg from breast tumor, Treg from B16 tumor, Treg from MC38 tumor, and Treg from CT26 tumor had 17,16,14,16,8,17,11,12,7 and 4 upregulated non-collaboration genes. Moreover, Treg from normal tissues had 0% to 12.2% non-collaboration genes, Treg from non-tumor diseased tissues had 0.6% to 7.3% non-collaboration genes, and Treg from tumor spleens and tumors had 3.6% to 36.8% non-collaboration genes ( Figures 6C,  D). Furthermore, as shown in the dot plots, Treg from tumor spleens and tumors had significantly increased Foxp3 noncollaboration genes. As shown in Figure 6E It has been reported that a Treg-specific deficiency of cytotoxic T lymphocyte antigen 4 (CTLA-4) results in development of systemic lymphoproliferation, fatal autoimmune disease, hyperproduction of immunoglobulin E, and potent tumor immunity (130). Our previous report determined the expression of 28 co-signaling receptors in 32 human tissues in physiological/pathological conditions and found that the expression of inflammasome components are correlated with the expression of co-signaling receptors (19). We also reported that among 10 immune checkpoint receptors (ICRs), lung, liver, spleen and intestine macrophages (Mj) and bone marrow (BM) Mj express higher levels of CD274 (programmed cell death-1, PDL-1) than adipose tissue Mj, presumably to counteract the  (1), suggesting that tissue immune checkpoint receptor signaling may also modulate Tregs. We hypothesized that CTLA-4 play roles in modulating DEGs in Treg from normal tissues, nontumor diseased tissues and tumor tissues. To test this hypothesis, we placed all the CLTA-4-induced Treg genes cells with Treg upregulated genes in the >1 log2 in the Y-axis, and anti-CTLA-4 antibody treated (CTLA-4 suppressed) genes in the >1 log2 in the X-axis ( Figure 7). The argument was that if gene expressions are promoted by CTLA-4, the genes will be shown in CTLA-4 induced gene groups; otherwise, they will be shown in CTLA-4 suppressed gene group. If any genes upregulated or downregulated in normal tissue Treg, non-tumor diseased Treg and tumor Treg are not collaborated with CTLA-4 regulation, they will be placed in non-collaboration gene groups. As shown in Figure 7, we found that there were a few non CTLA-4 collaboration genes upregulated in normal tissue and nontumor diseased tissue Treg in spleen (23 genes), VAT Treg (10 genes) and kidney Treg (11-16 genes). In contrast, tumor Treg had non CTLA-4 collaboration genes upregulated ranging from 10 genes to 50 genes, which were higher than that of normal tissue Treg and non-tumor diseased tissue Treg. In addition, we found, as shown in Figure 7E, that 13 pathways associated with CTLA-4 non-collaboration genes in Treg shared in various groups; 8 pathways associated with CTLA-4 non-collaboration genes in normal tissue Treg; 13 pathways associated with Treg from non-tumor diseased tissues, and 21 pathways associated with tumor Treg. Recent report showed that non-overlapping roles of PD-1 and Foxp3 in maintaining immune tolerance, which was collaborated well with our findings (131). We hypothesized that PD-1 plays significant roles in promoting DEGs in Treg from normal tissues, non-tumor diseased tissues and tumor tissues. To test this hypothesis, we placed all the PD-1-induced Treg genes cells with Treg upregulated genes in the >1 log2 in the Y-axis, and anti-PD-1 antibody treated (PD-1 suppressed) genes in the >1 log2 in the X-axis ( Figure 8A). The argument was that if gene expressions are promoted by PD-1, the genes will be shown in PD-1 induced gene groups; otherwise, they will shown in PD-1 suppressed gene group. If any genes upregulated or downregulated in tissue Treg, non-tumor diseased Treg and tumor Treg are not collaborated with PD-1 regulation, they will be placed in non-collaboration gene groups. As shown in Figure 8B, we found that there were a large numbers of PD-1 collaboration Treg genes upregulated and also significant numbers of PD-1 collaboration Treg genes downregulated in normal tissue Treg (14 -163 genes upregulated; 3 -48 genes downregulated). In addition, there were a large numbers of PD-1 collaboration Treg genes upregulated in non-tumor diseased tissue Treg (63-178 genes upregulated) and also significant numbers of PD-1 collaboration Treg genes downregulated (18 -34 genes). Moreover, there were a number of PD-1 noncollaboration Treg genes upregulated in normal and non-tumor diseased tissue Treg (0 -31 genes upregulated). Finally, there were a few numbers of PD-1 collaboration Treg genes upregulated in tumor tissue Treg (1 -8 genes upregulated). We also found 12 pathways associated with PD-1 noncollaboration genes shared by Treg in various groups ( Figure 8E); 9 pathways associated with PD-1 noncollaboration genes shared by Treg in normal tissues; 13 pathways associated with PD-1 non-collaboration genes shared by Treg in non-tumor diseased tissues; and 22 pathways with PD-1 non-collaboration genes shared by Treg in tumor tissues.
Our results have demonstrated that normal tissue Treg and non-tumor diseased tissue Treg have quite low CTLA-4 dependent but high PD-1 dependent transcriptomic changes, suggesting that CTLA-4 does not, but PD-1 does, play significant Foxp3-non-overlapped (131) roles in promoting Treg upregulated genes in normal and non-tumor diseased tissue Treg; and tumor spleen Treg and tumor Treg have certain transcriptomic changes in CTLA-4, and PD-1, non-collaboration manners with innate immune dominant pathways. Our results were correlated well with others reports on the effects of immune checkpoint receptors in Treg in pregnancy (132), Treg in autoimmunity (133), T cells in heart (134), and effects of immune checkpoint inhibitors on tumors (135)(136)(137).

Tumor Tregs Downregulate More Immunometabolic and Trained Immune Metabolic Enzyme Genes Than Tregs From Other Groups
It has been reported that while glycolysis fuels the biosynthetic and bioenergetic needs necessary for proliferation and migration of human Treg, suppressive capacity is mainly maintained by oxidative metabolism (138); that several metabolic pathways have been reported to be involved in immunometabolism (139) including glycolysis, fermentation, pentose phosphage pathway (PPP), tricarboxylic acid cycle (TCA)/mitochondrial electron transport chain (ETC), mitochondrial regulation, fatty acid metabolism, amino acid metabolism, and metabolic regulation and signaling; and that termed "trained immunity", allows innate immune cells such as macrophages, monocytes, natural killer cells and endothelial cells (50) to shown enhanced responsiveness when they re-encounter pathogens and danger associated molecular patterns (140,141), in which metabolic reprogrammings such as glycolysis, acetyl-CoA generation and mevalonate synthesis pathways (50,51). We hypothesized that the expressions of immunometabolic enzyme genes can be found in Treg upregulated genes and downregulated genes in normal, non-tumor diseased and tumor tissues. To test this hypothesis, we collected 191 immunometabolic enzyme genes (50,138,139). As shown in Figure 9A, 0 -7 genes, 1 -8 genes and 2 -25 immunometabolic enzyme genes were upregulated in normal tissue Treg, non-tumor diseased tissue Treg and tumor Treg, respectively. In addition, immunometabolic enzyme genes were downregulated more in tumor Treg than normal tissue Treg and non-tumor diseased tissue Treg ( Figure 9B). When we zoomed in the expression changes of trained immunity metabolic enzyme genes in Treg, we found that, as shown in Figure 10A, trained immunity metabolic enzyme genes in Treg from fibrosis kidney (8 genes), mouse B16 tumor (21 genes) and mouse MC38 tumor (12 genes) were increased. In comparison, trained immunity metabolic enzyme genes in Treg from normal tissue Treg, nontumor diseased tissue Treg and tumor Treg had a few genes upregulated. Once again, trained immunity metabolic enzyme genes were downregulated more in tumor Treg than normal tissue Treg and non-tumor diseased tissue Treg ( Figure 10B). These results have demonstrated that the expressions of some trained immunity metabolic enzyme genes and global immunometabolic enzyme genes are downregulated in tumor Treg, suggesting that some innate immune metabolic pathways are suppressed in tumor Treg than normal tissue Treg and nontumor diseased Treg. via ROS regulated by glutathione (GSH)-S-transferase P (GSTP1). Importantly, interfering with GSTP1 and ROS pathways re-invigorates the proliferation and function of aged Tregs (142). Another report also showed that elimination of excessive ROS by treating Treg with ROS scavengers such as Nacetyl-L-cysteine (NAC) and GSH can rejuvenate the aged Treg and reinvigorate their suppressive function (143), suggesting that DCAF-regulated ROS is the culprit underlying Treg aging and aberrant function in immune senescence and inflammaging (143). TGF-b/SRY-Box Transcription Factor 4 (SOX4) axismediated upregulation of ecto-nucleoside triphosphate diphosphohydrolase (NTPDase, CD39) is counteracted by ROS-driven autophagy (144). We hypothesized that ROS regulatomes are regulated in tissue Treg, non-tumor diseased tissue Treg and tumor Treg. As shown in Figure 11A, the expressions of some of 165 ROS regulatomic genes were differentially regulated in tissue Treg. The upregulations of ROS regulatomic genes were found in LN Treg (4), splenic Treg ( (12) and CT26 tumor Treg (4), respectively. As shown in Figure 11B, there were no statistical differences between Treg groups in upregulated ROS regulatomic genes. However, tumor tissue Treg groups significantly downregulated some ROS regulatomic genes, which may be responsible for increased Treg suppressive functions in Treg from tumors. Recently, we proposed a new concept that ROS are an integrated sensing system for metabolic homeostasis and alarming metabolic dangers (145). Nicotinamide adenine dinucleotide phosphate (NADPH) oxidase 2 (NOX2) inhibits the suppressive ability of Tregs by limiting NF-kB and Foxp3 activation (146). Dietary extra virgin olive oil attenuates kidney injury in pristane-induced SLE model via activation of HO-1/ Nuclear factor (erythroid-derived 2)-like-2 (NRF2) antioxidant pathway and suppression of JAK/STAT, NF-kB and MAPK activation (147). Epigallocatechin-3-gallate prevents lupus nephritis development in mice via enhancing the NRF2 antioxidant pathway and inhibiting NLRP3 inflammasome activation (148). We then hypothesized that ROS modulate Treg transcriptomic genes. To examine this hypothesis and identify ROS promoted genes and ROS suppressed genes, in Figure 12A, we analyzed the catalytic, membrane-bound subunit of NOX2 KO microarray dataset (GSE100671) (146) and antioxidant transcription factor NRF2 KO microarray dataset (GSE7810). As we reported recently (28), in Figure 12A, if NOX2 KO upregulated genes were co-expressed with NRF2  KO-downregulated genes, these genes were classified as ROS promoted genes. If NOX2 KO downregulated genes were coexpressed with NRF2 KO upregulated genes, those genes were classified as ROS-suppressed genes. Among 2,330 genes identified in NOX2 KO and NRF2 KO, 1384 genes were ROS promoted genes whereas 936 genes were ROS-suppressed genes, and 10 genes were ROS uncertain genes (did not follow the classification). We found that Treg in normal tissues upregulated ROS promoted genes in the range of 6 -89 genes. Treg in nontumor diseased tissues upregulated ROS promoted genes in the range of 22 -108 genes. Treg in tumor tissues upregulated ROS promoted genes in the range of 33 -99 genes. As shown in Figures 12B-D, the average numbers of upregulated ROSpromoted genes were more in Treg from benign disease tissue, Treg from tumor spleens and tumor tissues than other groups, however, the differences were not statistically significant. The average numbers of downregulated ROS-suppressed genes were more in the Treg from tumor tissues than that of Treg from other groups, p < 0.05 ( Figure 12E). These results suggest that ROSsuppressed immunosuppressive genes are downregulated more in tumor Treg than Treg from other groups.

DISCUSSION
Since Treg have a great therapeutic potential for numerous diseases (149) including cardiovascular disease (150) secretomic genes to fulfill their tissue-specific, secretomesspecific functions and shared functions. This is the first report that Treg may use canonical and innate immune noncanonical secretomes including exosomes to fulfill their immunosuppressive and tissue regeneration functions; 8) Most tissue Treg transcriptomes are controlled by Foxp3; Tumor Tregs had significantly increased Foxp3 non-collaboration genes with ROS and 17 other pathways, suggesting that ROS and 17 other pathways may promote tumor Treg in increasing Foxp3 noncollaboration genes; 9) PD-1 does, but CTLA-4 does not, play significant roles in promoting Treg upregulated genes in normal and non-tumor diseased tissue Tregs; and tumor splenic and tumor Tregs have certain CTLA-4-, and PD-1-,

A B
FIGURE 10 | (A) Tumor tissue Treg groups upregulated more Trained immunity enzymes than normal tissue Treg groups and benign disease tissue Treg groups. The expressions of total of 102 trained immunity enzyme genes were screened in all the Treg groups. Upregulated Trained immunity enzyme genes in tumor spleen Treg groups were more than in the normal tissue Treg groups and benign disease tissue Treg groups. However, the differences were not statistically significant. Downregulated, and total Trained immunity enzyme genes were more in tumor tissue Treg groups than in other Treg groups. (B) There were more differentially expressed trained immunity enzyme genes in tumor tissue Treg groups than other Treg groups. The total regulated trained immunity enzyme genes were more in tumor tissue Treg groups than that in normal tissue Treg group and diseased tissue Treg groups, p = 0.06. The upregulated trained immunity gene numbers had no differences between Treg groups. The downregulated trained immunity genes in tumor tissue Treg groups were more than other Treg groups, p < 0.05. non-collaboration transcriptomic changes with innate immune dominant pathways; 10) Tumor Tregs downregulate more immunometabolic and trained immune metabolic enzymes genes than Tregs from other groups; and 11) ROS significantly regulate Treg transcriptomes; and ROS-suppressed genes are downregulated more in tumor Treg than Treg from other groups.
Recently, we reported that while histone deacetylase 6 (HDAC6) and follicular Th cell-specific transcription factor Bcell lymphoma 6 (Bcl6) are important regulators of Treg plasticity, Th2-specific transcription factor GATA Binding Protein 3 (GATA3) determine the fate of plastic Treg by controlling whether it will convert in to either Th1-Treg or antigen-presenting cell (APC)-Treg (16); and Treg from spleen, lymph nodes, intestine and visceral adipose tissues promote tissue repair by generating secretomes similar to that of stem cells; and sharing TFs aryl hydrocarbon receptor (AHR), ETS variant transcription factor 5 (ETV5), early growth response 1 (EGR1), and Kruppel like factor 4 (KLF4) with stem cells, and Treg canonical secretomes and transcriptomes may be regulated by 1176 cytokines, 1706 canonical secretomes, kinome (complete list of human genome-encoded 651 kinases), cell surface receptors such as the complete list of 373 clusters of differentiation (CDs), the complete list of 1496 transcription factors, and 305 cell death regulatome (87). However, several important knowledge gaps presented in the INTRODUCTION remained unknown: 1) are there differences in Treg transcriptomes between Tregs from normal lymphoid and non-lymphoid tissues, injured tissues, regenerative tissues, tissues from mouse carrying tumor, tumor tissues; 2) are secretomes generated in Tregs in various tissues and pathological conditions; and 3) how Foxp3, immune checkpoint receptors CTLA-4 and PD-1 (19), immunometabolic pathways, trained immunity metabolic reprogramming (51,141) and ROS pathways (145) regulate tissue Treg heterogeneity and Treg various secretomes.
Based on our results, we propose a new working model. First, as shown in Figure 13, Tregs from normal lymphoid, non-lymphoid, injured muscle, fibrosis kidney, regeneration kidney, spleens from mouse with injured muscles and tumors, tumors have significant transcriptomic heterogeneity and signaling pathways; Second, in addition to immunosuppressive  Table 5) were screened in all the Treg groups. The upregulated ROS regulatome genes in tumor spleen Treg groups were more than that in the normal tissue Treg and benign diseased tissue Treg groups. However, the differences were not statistically significant. The downregulated ROS regulatome gene numbers were more in tumor tissue Treg groups than in other Treg groups (B) The differentially expressed ROS regultome genes in tumor tissue Treg groups were more than other Treg groups. The total regulated ROS regulatome gene numbers were more in tumor tissue Treg groups than that in normal tissue Treg groups, p = 0.16. The upregulated ROS regulatome gene numbers had no statistical differences between Treg groups. The downregulated ROS regulatome gene numbers were statistically more than other Treg groups, p < 0.05.
cytokines such as IL-10, TGF-b and IL-35, Treg may use much large secretomes such as canonical secretomes with proteins carrying signal peptide, non-canonical secretomes including caspase-1-GSDMD-dependent secretomes and innate immune caspase-4/11-GSDMD-dependent secretomes and exosome secretomes to act on target cells and Tregs themselves to maintain Treg niches for Treg heterogeneity and Treg-ness (87), fulfill immunosuppressive, tissue regeneration, tumor immunomodulatory functions; and Third, immunometabolism, innate immune memory (trained immunity) metabolic reprogramming, and ROS together with three well-characterized master regulators such as Treg transcription factor Foxp3, immune checkpoint receptors PD-1 and CTLA-4 regulate tissue transcriptomes and secretomes. Our findings have provided novel insights on tissue Treg niches for Treg heterogeneity and Treg-ness, and new therapeutic targets for immunosuppression, tissue repair, cardiovascular diseases, chronic kidney disease, autoimmune diseases, transplantation, and cancers.
One limitation of the current study is that due to the low throughput nature of verification techniques in the laboratories, we could not verify every result we identified with the analyses of high throughput data (see Table 1 of Dr. Lai's paper (1), and Table 10 of Dr. Zhang's paper (28) for explanations). We acknowledge that carefully designed in-vitro and in-vivo experimental models will be needed to verify all the findings further and underlying mechanisms. Nevertheless, our findings provide novel insights on the roles of tissue Treg in controlling immune responses, and promoting tissue repair and regeneration as well as novel targets for the future therapeutic interventions for immunosuppression, cardiovascular diseases, autoimmune diseases, transplantation, cancers and tissue repair. oxidase oxidase 2 (NOX2) deficient microarray and anti-oxidant transcription factor Nuclear factor (erythroid-derived 2) factor 2 (NRF-2) deficient microarray datasets. All the ROS regulated genes were classified into 3 categories: 1) 1384 ROS-promoted genes, 2) 936 ROS suppressed gene, and 3) 10 ROS regulation-uncertain genes. All the DEGs in all the Treg groups were examined in NOX2 deficiency-regulated genes (GSE100671) and Nrf2 (GSE7810) deficiency-regulated -datasets.
(B) There were no differences of ROS regulated genes between Normal spleen Treg groups and normal tissue Treg groups. (C) The benign disease tissue Treg groups had more ROS-regulated genes than the benign diseased spleen Treg groups. Both ROS-promoted genes and ROS-suppressed gene numbers in benign diseased non-lymphoid tissue Treg groups were more than that of lymphoid tissue Treg groups, p < 0.05. (D) There were no differences of ROS-regulated genes between tumor spleen Treg groups and tumor tissue Treg groups. Tumor tissue Treg groups and tumor spleen Treg groups had no significant differences in both ROS-promoted genes and ROS-suppressed genes. (E) Tumor tissue Treg groups had more downregulated reactive oxygen species (ROS)-regulated genes than other Treg groups. The average numbers of upregulated ROS promoted genes were more in the benign diseased tissue Treg groups, tumor spleen Treg groups and tumor tissue Treg groups than other Treg groups. However, the differences were not statistically significant. The average numbers of downregulated ROSsuppressed genes were more in the tumor tissue Treg groups than other Treg groups p < 0.05.