Graphical abstract graphic file with name fx1.jpg [33]Open in a new tab Highlights * • TSC22D3 expression is closely associated with poor prognosis in AML * • TSC22D3 regulates inflammatory immune response in AML tumor microenvironment * • TSC22D3 inhibits the polarization of macrophage to M1-type * • TSC22D3 attenuates inflammatory milieu through NF-κB/NLRP3 pathway __________________________________________________________________ Components of the immune system; Functional aspects of cell biology; Cancer systems biology Introduction Acute myeloid leukemia (AML) is a malignancy in which myeloid blasts proliferate in the hematopoietic system and is the most predominant leukemia in young adults aged 19–39 years.[34]^1 Despite the use of advanced multimodal treatment strategies such as chemotherapy, radiotherapy, and allogeneic stem cell transplantation, AML remains one of the deadliest forms of leukemia.[35]^2 AML is a genetically heterogeneous disease,[36]^3 and chromosomal abnormalities play a crucial role in its diagnosis and treatment.[37]^4 However, the underlying molecular mechanisms and pathogenesis of AML are still unclear, and only a few biomarkers are currently available for clinical treatment and prognosis. Although molecular-targeted therapy has made significant progress, many patients still experience relapse. Therefore, the development of prognostic biomarkers for AML diagnosis and treatment is crucial. TSC22D3, which was first identified in 1997, encodes a glucocorticoid (GC)-induced anti-inflammatory leucine zipper (GILZ) protein.[38]^5 The GILZ protein consists of 134 amino acids and is implicated in interference with various signal transduction pathways, including nuclear factor-κB (NF-κB),[39]^6 activating protein-1 (AP-1),[40]^7 Raf-MEK-ERK pathway,[41]^8 and transcription factor forkhead box O3 (FoxO3).[42]^9 Previous studies indicated that TSC22D3 was a GC-induced gene expression product that indirectly promotes glucocorticoid-mediated transcriptional inhibition.[43]^10^,[44]^11 TSC22D3 was first linked to cancer in a study that identified that GILZ can disrupt the cell cycle and facilitate tumor progression by activating the AKT pathway in epithelial ovarian cancer (EOC) cells.[45]^12 In chronic myeloid leukemia, GILZ promotes apoptosis in BCR::ABL1^+ cells by inhibiting the mTORC2/AKT pathway.[46]^13 PI3K/AKT inhibition can lead to apoptosis in multiple myeloma cells by increasing TSC22D3 expression.[47]^14 Furthermore, a variant of the TSC22D3 transcript called Long-GILZ (L-GILZ), which is upregulated by GCs, has similar functions to GILZ[48]^15 and can activate the p53 pathway through interaction with p53 and murine double minute 2 (MDM2) to inhibit tumor growth.[49]^16 Additionally, L-GILZ inhibits the growth of thyroid cancer cells and promotes carcinoma differentiation.[50]^17 However, researchers have found that overexpressing TSC22D3 can eliminate the tumor growth inhibitory effects of chemotherapy, while knockout of TSC22D3 significantly attenuated the stress-induced immunosuppression in cancer and suggests that TSC22D3 is a poor prognostic gene.[51]^18 It is evident from these findings that TSC22D3 plays key roles in both tumor progression and immunity. Recently, a bioinformatic study has shown the expression and prognostic significance of TSC22D domain family genes in adult AML;[52]^19 nevertheless, the potential role and prognostic significance of TSC22D3 in AML are still unclear and need to be further clarified. In the present study, we investigated the prognostic value of TSC22D3 in AML and its effect on the immune infiltration of AML tumor microenvironment. Our results identified TSC22D3 as an immune-related prognostic biomarker for AML patients. Mechanically, TSC22D3 alleviated the pro-inflammatory tumor microenvironment by attenuating the polarization of macrophage to M1 subtype by inhibiting NF-κB/NLRP3 signaling pathway in tumor cells, suggesting that therapeutic targeting of TSC22D3 may be a potential treatment strategy for AML. Results Expression of TSC22D3 is elevated in AML To investigate the expression of TSC22D3 in tumors, we first performed a combined analysis of The Cancer Genome Atlas (TCGA) and Genotype-Tissue Expression (GTEx) database to analyze the differences between normal and tumor samples. Our analysis indicated that TSC22D3 was significantly highly expressed in several cancers, including glioblastoma multiforme (GBM), kidney renal clear cell carcinoma (KIRC), AML, brain low-grade glioma (LGG), and pancreatic adenocarcinoma (PAAD) ([53]Figure 1A). To better understand TSC22D3 expression during AML progression, we analyzed bone marrow samples from 62 non-leukemia (control), 164 myelodysplastic syndromes (MDS), and 202 AML patients in the [54]GSE15061 dataset. Our study revealed that the expression of TSC22D3 was significantly higher in patients with AML compared to both control subjects and patients with MDS ([55]Figure 1B). Similarly, we observed a high expression of TSC22D3 in AML compared to MDS in the [56]GSE13159 dataset, which contained 74 controls, 206 MDS patients, and 542 AML patients ([57]Figure 1C). To investigate the role of TSC22D3 in AML, we divided AML patients into two groups based on the median expression of TSC22D3 in the TCGA database and performed differential expression analysis. We obtained 1,154 differentially expressed genes (DEGs) of which 657 were upregulated and 497 were downregulated. Our analysis of the top 10 differential genes revealed that TSC22D3 was co-expressed with several immune-related genes, including IL1R2, SLC10A2, AREG, interleukin (IL)-10, and SDC2 ([58]Figure 1D). Our findings indicated that TSC22D3 expression is notably high in AML, which implies a potential connection between TSC22D3 and immune response. Figure 1. [59]Figure 1 [60]Open in a new tab TSC22D3 is highly expressed in AML (A) TCGA and GTEx databases combined with pan-cancer analysis showed that TSC22D3 was highly expressed in GBM, KIRC, AML, LGG, and PAAD. The full names corresponding to TCGA disease abbreviations are shown in [61]Table S3. (B and C) Using [62]GSE15061 and [63]GSE13159 datasets in the GEO database to further verify that TSC22D3 was highly expressed in AML. (D) TSC22D3 was co-expressed with the immune-related genes IL1R, SLC10A2, AREG, IL10, and SDC2; |log2(FC)| > 1, p.adj <0.05. Data were expressed as mean ± SD (A-C). Abbreviations: p.adj, adjusted p value. Wilcoxon rank-sum test (A) was used to analyze two groups, and Kruskal-Wallis’s test (B and C) was used to analyze multiple groups. Significant markers: ns, p ≥ 0.05; ∗, p < 0.05; ∗∗, p < 0.01; ∗∗∗, p < 0.001. High expression of TSC22D3 is associated with immunoinfiltration in the AML tumor microenvironment After analyzing the differences, we formed a hypothesis that TSC22D3 in AML is involved in the immune response. To delve deeper into this theory, we conducted Gene Ontology (GO) analysis of DEGs. The findings suggested that TSC22D3-related genes significantly impact multiple biological functions. These functions encompass the regulation of inflammatory responses, leukocyte proliferation, tumor necrosis factors production, cytokine production related to immune response, myeloid leukocyte differentiation, and macrophage migration ([64]Figure 2A). To further investigate the relationship between TSC22D3 expression and immune cells, we performed immunoinfiltration analysis. TSC22D3 expression was positively correlated with macrophages, T helper-1 (Th1) cells, and Th17 cells and negatively correlated with T helper cells and mast cells ([65]Figure 2B). As the expression of TSC22D3 increased, the infiltration of macrophages ([66]Figure 2C), Th1 cells ([67]Figure 2D), and Th17 cells ([68]Figure 2E) in the tumor microenvironment increased. We further investigated the association between TSC22D3 expression and T cell immune checkpoints such as PD-1, PD-L1, and CTLA-4. TSC22D3 expression was significantly corroborated with PD-1 ([69]Figure 2F) and CTLA-4 ([70]Figure 2H) expression in AML, but not with PD-L1 ([71]Figure 2G). Additionally, to explore the functional relevance of TSC22D3-associated genes, we conducted gene set enrichment analysis (GSEA) on DEGs using the curated gene set (c2.cp.v7.2.symbols.gmt) from the Molecular Signature Database (MSigDB) ([72]https://www.gsea-msigdb.org/gsea/msigdb/index.jsp). The set of genes enriched in the TSC22D3 high-expression group included IL-18, p53, IL-4 and IL-13, IL-1 family, and NF-κB signaling pathways ([73]Figure 2I). These results demonstrated that TSC22D3 influences the infiltration of immune cells and inflammation-related signaling pathways in the AML tumor microenvironment. Figure 2. [74]Figure 2 [75]Open in a new tab TSC22D3 is highly associated with the AML immune microenvironment (A) GO analysis showed that TSC22D3-related genes were mainly involved in the biological processes of inflammation and immunity. (B) Correlation analysis between TSC22D3 and enrichment scores of 24 immune cells by Spearman correlation analysis. The size of the circle represents the degree of correlation. The greater the height of the bar (distance from 0), the higher the degree of correlation. The depth of the circle represents the p value obtained by the correlation statistical method. (C–E) The relationship between TSC22D3 and the infiltrated macrophages (C), Th1 (D), and Th17 (E) lymphocytes. (F–H) Scatterplots of the correlations between TSC22D3 expression and PD-1 (F), PD-L1 (G), and CTLA-4 (H) in AML. (I) GSEA pathway enrichment analysis showed that TSC22D3-related differential genes could cause significant changes in different signaling pathways such as inflammation. Data were expressed as mean ± SD (C–E). Abbreviations: BP, biological process; CC, cell component; MF, molecular function; NES, normalized enrichment score; FDR, false discovery rate; r, correlation coefficient. Student’s t test (C–E) was used for the analysis between the two groups. Significant markers: ∗, p < 0.05; ∗∗, p < 0.01. High expression of TSC22D3 is closely related to poor prognosis in AML patients We conducted an analysis of 148 AML patients with different cytogenetic risk profiles from the TCGA database, including 30 patients with favorable cytogenetic risk, 82 with intermediate cytogenetic risk, and 36 with poor cytogenetic risk patients, to investigate the significance of TSC22D3 expression on prognosis. The findings indicated that the TSC22D3 expression increased progressively with increasing cytogenetic risk class ([76]Figure 3A). The Kaplan-Meier (K-M) curve was utilized to evaluate the survival time of AML patients in groups with high TSC22D3 versus low TSC22D3. We observed that TSC22D3^−low patients had significantly better 8-year overall survival (OS) rate than TSC22D3^−high patients ([77]Figure 3B). Subgroup survival analysis indicated a positive correlation between TSC22D3 expression and poor prognosis in patients who lacked mutations in FLT3, IDH1 R132, IDH1 R140, IDH1 R172, RAS, and NPM1 ([78]Figures 3C–3H). However, TSC22D3 expression was not associated with poor prognosis in the FLT3, IDH1 R132, IDH1 R140, RAS, and NPM1 mutation-positive subgroups ([79]Figure S1). Furthermore, multivariate Cox regression analysis revealed that age over 60 years, intermediate and poor cytogenetic risk, and high expression of TSC22D3 were independent prognostic predictors ([80]Table S4). To investigate the predictive effect of TSC22D3 on AML prognosis, we constructed a prognostic prediction model utilizing the findings of a multivariate Cox regression analysis. The nomogram model used a ruler score to evaluate the conditions of each variable and calculate a total score to estimate the probability of the event taking place. The points assigned to each variable were adjusted on a scale of 1–100. A vertical line was drawn from the total point axis to the outcome axis to determine the probability of survival for AML patients at 1, 3, and 5 years. The model demonstrated good prediction efficiency, as evidenced by the consistency index (C-index) of 0.735 ([81]Figure S2A). Additionally, we utilized the receiver operating characteristic curve (ROC) model to assess the prognostic predictive efficacy of TSC22D3 expression. Area under curve (AUC) values for 1, 3, and 5 years were 0.615, 0.587, and 0.618, respectively. These results suggested that the predictive power of TSC22D3 expression is limited as the AUC values indicate low prediction efficiency ([82]Figure S2B). To assess the model’s performance, we generated a calibration plot that compares the predicted probabilities of survival with the actual outcomes. The calibration plot showed a bias-corrected line that closely resembled the ideal curve, which is the 45-degree line. This indicates that there is good agreement between the predictions and observations. The figure displays three lines that represent the comparison between the model’s predicted survival probabilities for 1-year (blue), 3-year (red), and 5-year (green) periods, respectively, with the actual outcomes ([83]Figure S2C). In conclusion, our findings indicated that increased TSC22D3 expression is linked to a negative outcome for AML and may serve as a potential prognostic marker. Figure 3. [84]Figure 3 [85]Open in a new tab High expression of TSC22D3 is associated with poor prognosis of AML (A) The clinical correlation between TSC22D3 expression and cytogenetic risk stratification of AML. (B) OS curves between TSC22D3-high and TSC22D3-low patients with AML. (C–H) High TSC22D3 expression was associated with poor prognosis in the subgroups of mutation-negative genes including FLT3 (C), IDH1 R132 (D), IDH1 R140 (E), IDH1 R172 (F), RAS (G), and NPM1 (H). The gene mutation-positive groups can be found in [86]Figure S1. Cox regression analysis and prognostic modeling can be seen in [87]Figure S2 and [88]Table S4. HR > 1 indicates that the variable was a risk factor, and HR < 1 was a protective factor. Data were expressed as mean ± SD (A). Abbreviations: HR, hazard ratio. One-way ANOVA test (A) was used to analyze multiple groups of overall comparisons. Cox regression (C–H) was used for survival analysis. Significant markers: ns, p ≥ 0.05; ∗, p < 0.05. TSC22D3 promotes the proliferation of AML cells and reduces Ara-c sensitivity To investigate the expression of TSC22D3 in hematological tumor cell lines, we utilized real time quantitative polymerase chain reaction (RT-qPCR) and western blotting techniques to detect mRNA and protein expression. Our findings revealed relatively high levels of TSC22D3 mRNA expression in U937, MV-4-11, Hel, and THP1 cell lines ([89]Figure 4A), and TSC22D3 protein expression was relatively high in KG-1, MV-4-11, U937, and Hel cell lines ([90]Figures 4B and 4C). To study the impact of TSC22D3 on tumor cells, Hel cells were transfected with the shTSC22D3 vector and pLKD vector, and the knockdown efficacy of TSC22D3 was verified first. Our results indicated that both the mRNA levels ([91]Figure 4D) and protein expression ([92]Figures 4E and 4F) of TSC22D3 were significantly reduced in the knockdown group compared to the control group. We investigated the regulatory role of TSC22D3 by conducting the cell counting kit-8 (CCK-8) assay to determine cell proliferation and drug sensitivity in vitro. Knockdown of TSC22D3 inhibited the growth rate of tumor cells ([93]Figure 4G). Ara-c is a commonly used chemotherapy drug for treating AML. In our study, we found that the cell concentration for 50% of the maximal effect (EC50) of Ara-c on Hel cells was 155.6 nM ([94]Figure S3A). Consistently, TSC22D3 deficiency increased the sensitivity to Ara-c of Hel cells ([95]Figure 4H). To further demonstrate the effect of TSC22D3 on cell proliferation, we examined the expression of proliferation cell nuclear antigen (PCNA) protein, a marker of cell proliferation, in the presence and absence of 100 nM Ara-c treatment for 24 and 48 h. Knockdown of TSC22D3 inhibited the expression of PCNA ([96]Figures 4I and 4J). In vivo, the mouse cell-derived xenograft (CDX) model further confirmed that TSC22D3-deficient AML cells obviously alleviated the proliferation rate and tumor invasion ([97]Figure 4K). The above results indicated that the knockdown of TSC22D3 inhibits the growth of tumor cells and increases the drug sensitivity to Ara-c. Figure 4. [98]Figure 4 [99]Open in a new tab High expression of TSC22D3 promotes the proliferation of AML cells and inhibits sensitivity to Ara-c (A) The mRNA expression levels of TSC22D3 in different hematologic tumor cell lines (n = 3). (B) Western blot analysis of TSC22D3 in different hematologic tumor cell lines. (C) The normalized quantification of TSC22D3 band density compared to β-Actin (n = 3). (D) RT-qPCR analysis knockdown efficacy of Hel cells after transduction of pLKD vector or shTSC22D3 vector (n = 6). (E) Western blot analysis knockdown efficacy of Hel cells after transduction. (F) The normalized quantification of TSC22D3 band density compared to β-Actin (n = 3). (G) CCK-8 analysis for cell viability of Hel cells after transfected with shTSC22D3 vector or pLKD vector (n = 4). (H) CCK-8 analysis for cell viability of Hel cells after transduction followed by 100 nM Ara-c treatment for 4 days (n = 4). EC50 values see [100]Figure S3. (I) Western Blot analysis of proliferation cell nuclear antigen (PCNA) in Hel cells transfected with shTSC22D3 vector or pLKD vector in the presence or absence of 100 nM Ara-c treatment for 24 h and 48 h. (J) The normalized quantification of PCNA band density compared to β-Actin (n = 3). (K) Analysis of proliferation of Hel cells after transfection in 4-6-week-old male BALB/c nude CDX treated with 20 mg/kg Ara-c every 3 days for 25 days (n = 4). Data were expressed as mean ± SD (A, C, D, F, G, H, and J) or min to max (K). Student’s t test (D, F, and J) for two groups comparison and two-way repeated measures ANOVA (G, H, and K) analysis for multi-factor comparisons. Significant markers: ∗, p < 0.05; ∗∗, p < 0.01; ∗∗∗, p < 0.001. TSC22D3 inhibits apoptosis of tumor cells and affects cell cycle As apoptosis is a crucial factor in the development and advancement of tumors, we sought to investigate whether TSC22D3 regulates the apoptosis of AML cells. To this end, we measured the apoptosis rate of cells that were treated with 100 nM Ara-c or vehicle for 2 days. Interestingly, knockdown of TSC22D3 did not cause apoptosis of tumor cells, but with the treatment of Ara-c, knockdown of TSC22D3 could promote the apoptotic rate of tumor cells ([101]Figures 5A and 5B). We consider that Ara-c exerts its antitumor effects mainly by inhibiting DNA polymerase, interfering with DNA replication, and affecting the cell cycle. We speculate whether TSC22D3 affects cell proliferation, apoptosis, and sensitivity to Ara-c by influencing the cell cycle. To test this hypothesis, we used propidium iodide (PI) staining and flow cytometry to detect the cell cycle. We found that the knockdown of TSC22D3 increased the Gap0 (G0) and Gap1 (G1) phases meanwhile inhibiting the synthesis (S) phase ([102]Figures 5C and 5D). Consistently, after co-stimulated with Ara-c, knockdown of TSC22D3 significantly increased the S stage but inhibited the Gap2 (G2) and mitosis (M) stage of AML cells ([103]Figures 5E and 5F). The role of the p53 signaling pathway in regulating the cell cycle is widely recognized. To investigate the possible mechanism by which TSC22D3 impacts the cell cycle, we conducted protein blotting to detect the expression of the p53 protein. Our findings, as revealed by western blot analysis, indicated that TSC22D3 deficiency in AML cells leads to increased expression of p53 ([104]Figures 5G and 5H), suggesting that TSC22D3 may regulate the tumor cell cycle via interaction with p53. Figure 5. [105]Figure 5 [106]Open in a new tab TSC22D3 affects apoptosis and the cell cycle (A) Representative fluorescence-activated cell sorting plots showed the frequencies of apoptotic Hel cells after transfected with shTSC22D3 vector or pLKD vector for 48 h in the presence or absence of 100 nM Ara-c treatment. (B) Quantitative statistical analysis of Hel cells apoptosis percentage (n = 3). (C) The staggered offset histogram analysis showed the percentage of different cell cycles of Hel cells after being transfected with the shTSC22D3 vector or pLKD vector. (D) Quantitative statistical analysis of Hel cell cycle percentage (n = 3). (E) The staggered offset histogram analysis showed the percentage of different cell cycles of Hel cells after transfected with shTSC22D3 vector or pLKD vector treated with Ara-c (100 nM) for 48 h. (F) Quantitative statistical analysis of Hel cells cycle percentage treated with Ara-c (100 nM) (n = 3). (G) Western blot analysis of protein expression levels of p53 in Hel cells after transfection. (H) The normalized quantification of p53 band density compared to β-Actin (n = 3). Cell cycle model constraints were G2 peak = 2G1, G1 CV = G2 CV = 8.00. Data were expressed as mean ± SD (B, D, F, and H). Abbreviations: CV, coefficient of variation. Student’s t test (B, D, F, and H) for two groups comparison. Significant markers: ∗, p < 0.05; ∗∗, p < 0.01; ∗∗∗, p < 0.001. TSC22D3 inhibits the infiltration and polarization of M1-type macrophages To explore the specific role of TSC22D3-knocked tumor cells on the immune response, we used the co-culture of tumor cells and M0-type macrophages. Firstly, the induction of differentiation of THP1 cells into M1- and M2-type macrophages was verified ([107]Figures S3B and S3C). The downregulation of TSC22D3 resulted in increased levels of CD80 and tumor necrosis factor-α (TNF-α) mRNA expression in M1 macrophages, while simultaneously decreasing the expression of CD206 and PPARγ mRNA in M2 macrophages ([108]Figures 6A and 6B). To assess the impact of TSC22D3 knockdown in tumor cells on immunoinfiltration, we generated single-cell suspensions from CDX mouse model tumor tissues for flow cytometry analysis. The results showed that TSC22D3-knockdown tumor cells promoted infiltration of M1 macrophages in subcutaneous tumors but had no effect on the infiltration of M2 macrophages, which might be due to the limited samples ([109]Figures 6C–6F). These results indicated that TSC22D3 inhibits the polarization and infiltration of M1-type macrophages, ultimately promoting tumor progression. Figure 6. [110]Figure 6 [111]Open in a new tab High expression of TSC22D3 inhibits the polarization and infiltration of M1 macrophages (A) RT-qPCR analysis of markers of M1-type macrophages after co-culture (n = 4). (B) RT-qPCR analysis of markers of M2-type macrophages after co-culture (n = 4). (C) The staggered offset histogram analysis showed M1 macrophages infiltration in the CDX model. (D) Quantitative statistical analysis of M1 macrophages infiltration percentage (n = 4). (E) The staggered offset histogram analysis showed M2 macrophages infiltration in the CDX model. (F) Quantitative statistical analysis of M2 macrophages infiltration percentage (n = 4). Data were expressed as mean ± SD (A, B, D, and F). Student’s t test (A, B, D, and F) for two groups comparison. Significant markers: ∗, p < 0.05; ∗∗, p < 0.01; ∗∗∗, p < 0.001. TSC22D3 alleviates NF-κB and NLRP3 activation in tumor cells To investigate the underlying mechanism that TSC22D3 attenuated the polarization and infiltration of M1 macrophages, we further investigated whether TSC22D3 regulates inflammatory-related signaling pathways. We detected NF-κB and NLRP3 pathway-related proteins in TSC22D3-knockdown tumor cells and found that inhibition of TSC22D3 led to the activation of phosphorylated p65 (P-p65) and phosphorylated nuclear factor-kappa B inhibitor alpha (P-IKBα) ([112]Figures 7A and 7B). NLRP3 pathway-associated proteins (NLRP3 and pro-IL-1β) were also increased in TSC22D3-knockdown tumor cells ([113]Figures 7C–7F). To further investigate the role of NLRP3 in regulating macrophage polarization, we used cell-conditioned medium for secreted protein assays. Knockdown of TSC22D3 resulted in the increase of IL-1β cytokines in the supernatant of tumor cells compared with their wild-type cells, while there was no significant IL-18 secretion ([114]Figures 7G and 7H). Together, these results confirmed that knockdown of TSC22D3 promotes the activation of NF-κB and NLRP3-associated inflammatory signaling pathways, leading to the increase of IL-1β secretion from tumor cells, which in turn affects the polarization and infiltration of macrophages and ultimately promotes the proliferation of tumor cells. Figure 7. [115]Figure 7 [116]Open in a new tab High expression of TSC22D3 inhibits the activation of NF-κB and NLRP3 signaling pathways (A) Western blot analysis of protein expression levels of P-p65 and P-IKBα in Hel cells after transfection. (B) The normalized quantification of P-p65 and P-IKBα band density compared to vinculin (n = 3). (C) Western blot analysis of protein expression levels of NLRP3 in Hel cells after transfection. (D) The normalized quantification of NLRP3 band density compared to β-Actin (n = 3). (E) Western blot analysis of protein expression levels of NLRP3 pathway-related molecular in Hel cells after transfection. (F) The normalized quantification of NLRP3 pathway-related molecular band density compared to vinculin (n = 3). (G and H) Elisa detected IL-1β and IL-18 concentrations in the supernatant of the transfected cell supernatant medium (n = 4). Data were expressed as mean ± SD (B, D, F, G, and H). Student’s t test (B, D, F, G, and H) for two groups comparison. Significant markers: ∗, p < 0.05; ∗∗, p < 0.01; ∗∗∗, p < 0.001. Discussion Despite the extensive research conducted on TSC22D3 in tumor biology, its role and implication in AML had not been explored. Hence, this study focuses on the potential role of TSC22D3 in the immunity and prognosis of AML patients. Through bioinformatics analysis of RNA sequencing (RNA-seq) data from UCSC XENA and GEO databases, TSC22D3 was found to exhibit high expression levels, suggesting its potential as an immune-related biomarker for AML. Additionally, it has been discovered that increased levels of TSC22D3 in individuals with AML are linked to unfavorable cytogenetics risk, shorter survival rates, and a bleak prognosis. TSC22D3 is a leucine zipper protein that functions as a transcriptional regulator.[117]^20 Structurally, The TSC22D3 protein contains three main components: the N-terminal domain of the transcription start site, the leucine zipper, and the proline-rich C-terminal domain.[118]^9 TSC22D3 expression is stimulated by glucocorticoids and IL-10 and is thought to be critical in the anti-inflammatory and immunosuppressive effects of CD80 and CD86.[119]^21 Furthermore, TSC22D3 is a vital factor in controlling the differentiation, apoptosis, and cell cycle of immune cells.[120]^11 Prior research has indicated that TSC22D3 has the potential to hinder the proliferation and maturation of cancerous cells by obstructing the functions of critical inflammatory signaling agents, including CCL5 (RANTES) and macrophage inflammatory protein-1 alpha (MIP-1α).[121]^21 Macrophage-derived TSC22D3 also functions as a transcriptional co-inhibition for the pathogen-associated molecular patterns (PAMPs), Toll-like receptor 2, and NF-κB, thereby mediating the suppression of macrophage and anti-inflammatory effects.[122]^21 Furthermore, L-GILZ, the protein isoform encoding TSC22D3, can promote the activation of p53 by competitively binding MDM2, thereby inhibiting tumor growth.[123]^16 Conversely, another study has suggested that TSC22D3 may enhance the phosphorylation and activity of AKT in ovarian cancer cells, leading to increased tumor proliferation and metastasis.[124]^12 Moreover, the increased expression of TSC22D3 induced by glucocorticoids can attenuate anti-cancer immunity and promote tumor recurrence.[125]^18 In present study, we observed a significant abundance of TSC22D3 expression in AML patients ([126]Figure 1), and high levels of TSC22D3 are linked to unfavorable clinical outcomes ([127]Figure 3), which was consistent with a recent bioinformatics study.[128]^19 Given the limited knowledge available regarding the function of TSC22D3, we conducted functional annotation based on GO and GSEA analysis. Our results showed that TSC22D3 is involved in the regulation of inflammation and immunity ([129]Figure 2), suggesting that TSC22D3 may contribute to the advancement of AML by stimulating these signaling pathways.[130]^22^,[131]^23 Meanwhile, inhibition of TSC22D3 expression significantly impeded AML cell proliferation ([132]Figure 4), affected the cell cycle via the p53 pathway, and increased cell sensitivity to Ara-c treatment ([133]Figure 5). The prognostic significance of TSC22D3 in AML was further suggested. TSC22D3 is correlated with a negative prognosis for AML and contributes to the progression of the disease. TSC22D3 encodes a protein that functions as important anti-inflammatory and immunosuppressive processes. Previous studies showed that immune cells can modify the carcinogenic properties of tumor cells and play crucial roles in the advancement of tumors.[134]^24 Among those immune cells, macrophage is an important player in regulating AML tumor microenvironment.[135]^25^,[136]^26^,[137]^27 It is noteworthy that tumor cells themselves can attract anti-inflammatory M2-type macrophages to promote the immune escape,[138]^28 while the M1-type macrophages may help to suppress the development of AML disease.[139]^29 In the present study, we found that TSC22D3 positively correlated with macrophage infiltration ([140]Figure 2B), and the knockdown of TSC22D3 promoted polarization and infiltration of M1-type macrophage ([141]Figure 6). These findings illustrated the close association between TSC22D3 and immunoinfiltration in AML, potentially leading to immune evasion by changing the tumor microenvironment. Previous studies showed that NF-κB is controversial in immune responses in tumor microenvironment,[142]^30^,[143]^31^,[144]^32^,[145]^33 which can either promote or inhibit the progression of tumors.[146]^34^,[147]^35 IL-1β, a downstream regulatory molecule of NF-κB, also has critical roles in tumor development,[148]^36 by promoting the polarization of M1 macrophages.[149]^37 This study revealed that TSC22D3 led to the inhibition of NF-κB and NLRP3 signaling pathways, which suppressed IL-1β secretion by cancer cells ([150]Figure 7), attenuating the inflammatory environment in the tumor microenvironment. As a result, M1 macrophage polarization and infiltration are inhibited and the balance between M1 and M2 macrophages is disrupted, ultimately promoting the progression of AML. In summary, our study demonstrated that high expression of TSC22D3 is significantly associated with poor survival and immune infiltration in AML patients. TSC22D3 promotes tumorigenesis and progression in AML by suppressing the inflammatory response, which may trigger the immune escape of tumor cells. Together, TSC22D3 is an immune modulator of AML, which provides a new perspective to further investigate the pathogenesis of AML and tumor immune escape. Limitations of the study Although the present study revealed the role of TSC22D3 in AML prognosis and immunoinfiltration, some limitations still require attention. Firstly, our study has not used the primary AML samples to validate our results; analysis of clinical patient specimens from newly diagnosed and relapsed patients is critical to better elucidate the role of TSC22D3 in the disease process in the future. Secondly, we need to increase the quantity of cell lines to verify the functional role of TSC22D3 in AML cells in multiple ways. Additionally, it would be beneficial to analyze RNA-seq in TSC22D3-knockdown cells to clarify possible downstream molecular mechanisms. Finally, to better elucidate the effect of TSC22D3 on immunoinfiltration, humanized mouse models should be used to construct AML disease models. STAR★Methods Key resources table REAGENT or RESOURCE SOURCE IDENTIFIER Antibodies __________________________________________________________________ Rabbit Polyclonal anti-TSC22D3 Signalway Antibody Cat#39175 Rabbit Monoclonal anti-β-Actin Signalway Antibody Cat#52901 Goat Polyclonal anti-Rabbit IgG (H&L) HRP Signalway Antibody Cat#[151]L35009 Rabbit Monoclonal anti-PCNA (D3H8P) XP® Cell Signaling Technology Cat#13110; RRID:[152]AB_2636979 Rabbit Monoclonal anti- Phospho-NF-kappaB p65 (Ser536) (93H1) Cell Signaling Technology Cat#3033; RRID:[153]AB_331284 Rabbit Monoclonal anti-IKB alpha (phospho S36) Abcam Cat#ab133462; RRID:[154]AB_2801653 Rabbit Monoclonal anti-NLRP3 Abcam Cat#ab263899; RRID:[155]AB_2889890 Rabbit Monoclonal anti-pro Caspase-1 + p10 + p12 Abcam Cat#ab179515; RRID:[156]AB_2884954 Rabbit Monoclonal anti-p53 PTM Biolabs Cat#PTM-6319 Rabbit Polyclonal anti-Caspase-1/Cleaved Caspase-1 Wanleibio Cat#WL03450 Rabbit Polyclonal anti-pro IL-1β Wanleibio Cat#WL02257; RRID:[157]AB_2894987 Rabbit Polyclonal anti-TMS1/ASC Wanleibio Cat#02462; RRID:[158]AB_2935861 Rabbit Polyclonal anti-Vinculin Wanleibio Cat#03922 Monoclonal Pacific Blue™ anti-mouse/human CD11b BioLegend Cat#101224; RRID:[159]AB_755986 Monoclonal FITC anti-mouse F4/80 antibody BioLegend Cat#123107; RRID:[160]AB_893500 Monoclonal PE anti-mouse CD80 antibody BioLegend Cat#104707; RRID:[161]AB_313128 Monoclonal APC anti-mouse CD206 (MMR) antibody BioLegend Cat#141707; RRID:[162]AB_10896057 __________________________________________________________________ Bacterial and virus strains __________________________________________________________________ Stbl3 competent cells TSINGKE Cat#TSC-C06 __________________________________________________________________ Chemicals, peptides, and recombinant proteins __________________________________________________________________ TRIzol® reagent ThermoFisher Cat#15596018 Universal antibody diluent EpiZyme Cat#PS119L Annealing buffer Beyotime Cat#D0251 Ampicillin MDbio Cat#A010 Opti-MEM ThermoFisher Cat#31985070 CCK-8 Apex Bio Cat#K1018 Cytarabine (Ara-c) Apex BIO Cat#A8405 Phorbol-12-myristate-13-acetate (PMA) MCE Cat#HY-18739 Recombinant human IFN-γ protein NovoProtein Cat#C014 Lipopolysaccharide (LPS) Bioss Cat#bs-8000P Recombinant human IL-4 protein NovoProtein Cat#C050 Recombinant human IL-13 protein NovoProtein Cat#CC89 Matrigel CELLada Cat#CELLada-OM-2 Type I collagenase Sigma Cat#C0130 DNase I Acmec Cat#[163]D61780 Red blood cell lysate Beyotime Cat#C3702 7-AAD Viability Staining Solution BioLegend Cat#420404 __________________________________________________________________ Critical commercial assays __________________________________________________________________ Fast Reverse Transcription kit ESscience Cat#RT001 2 x Super SYBR Green qPCR Master Mix (with ROX) ESscience Cat#QP002 Bicinchoninic acid (BCA) Protein Assay Kit CWBIO Cat#CW0014S 12.5% SDS-PAGE Gel Fast Preparation Kit EpiZyme Cat#PG113 AgeI-HF NEB Cat#R3552L EcoRI-HF NEB Cat#R3101V T4 DNA ligase NEB Cat#M0201S Endotoxin-Free Plasmid Extractor Kit TIANGEN Cat#DP118-02 lipofectamine 3000 Invitrogen Cat#L3000015 7-AAD-APC-Annexin V Apoptosis Assay Tonbo Cat#20-6410-KIT Cell Cycle Analysis kit 4Abio Cat#FXP0211 Human IL-18 Elisa Kit MultiSciences Cat#EK118 Human IL-1β ELISA Kit MultiSciences Cat#EK101B __________________________________________________________________ Deposited data __________________________________________________________________ TCGA and GTEx databases RNA sequencing data UCSC XENA [164]https://xenabrowser.net/datapages/ [165]GSE15061 and [166]GSE13159 datasets sequencing data GEO [167]http://www.ncbi.nlm.nih.gov/geo/ Original data and statistical description This study [168]https://data.mendeley.com/datasets/tnbptc8yff/2 __________________________________________________________________ Experimental models: Cell lines __________________________________________________________________ Human: KG-1 Laboratory of Yang Liang RRID:CVCL_0374 Human: MV-4-11 Laboratory of Yang Liang RRID:CVCL_0064 Human: OCI-AML3 Laboratory of Yang Liang RRID:CVCL_1844 Human: U937 Laboratory of Yang Liang RRID:CVCL_0007 Human: Hel Laboratory of Yuanbin Song RRID:CVCL_0001 Human: THP1 Laboratory of Yuanbin Song RRID:CVCL_0006 Human: 293T Laboratory of Yang Liang RRID:CVCL_0063 __________________________________________________________________ Experimental models: Organisms/strains __________________________________________________________________ Mouse: BALB/c nude Sun Yat-sen University Cancer Center N/A __________________________________________________________________ Oligonucleotides __________________________________________________________________ RT-qPCR primers see [169]Table S1 This study N/A shRNA sequences see [170]Table S2 This study N/A __________________________________________________________________ Recombinant DNA __________________________________________________________________ pLKD-U6-MCS-CMV-EGFP-2A-Puro Umine-bio Cat#G0202296-1 pLKD-U6-shRNA (non-targeting)-CMV-EGFP-2A-Puro This study N/A pLKD-U6-shRNA (TSC22D3)-CMV-EGFP-2A-Puro This study N/A psPAX2 Laboratory of Yang Liang N/A pVSVG Laboratory of Yang Liang N/A __________________________________________________________________ Software and algorithms __________________________________________________________________ ggplot2 (v3.3.3) Wickham[171]^38 [172]https://cran.r-project.org/web/packages/ggplot2/index.html DESeq2 (v1.26.0) Love et al.[173]^39 [174]https://bioconductor.org/packages/release/bioc/html/DESeq2.html clusterProfiler (v3.14.3) Yu et al.[175]^40 [176]http://bioconductor.org/packages/release/bioc/html/clusterProfiler .html GSVA (v1.34.0) Hänzelmann et al.[177]^41 [178]https://www.bioconductor.org/packages/release/bioc/html/GSVA.html timeROC (v0.4) Blanche et al.[179]^42 [180]https://cran.r-project.org/web/packages/timeROC/index.html RStudio (v3.6.3) Posit Software Inc., USA [181]https://posit.co/download/rstudio-desktop/ ImageJ (v1.8.0) National Institutes of Health, USA [182]https://downloads.imagej.net/fiji/latest/fiji-win64.zip FlowJo (v10.8.1) BD Biosciences [183]https://www.flowjo.com/solutions/flowjo/downloads GraphPad Prism 9.5 GraphPad Software Inc., CA, USA [184]https://www.graphpad.com/ [185]Open in a new tab Resource availability Lead contact Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Lingling Shu ([186]shull@sysucc.org.cn). Materials availability This study did not generate new unique reagents. Experimental model and study participant details Human samples All RNA-seq data and clinical characteristics of human samples in this study were obtained from UCSC XENA ([187]https://xenabrowser.net/datapages/) and Gene Expression Omnibus (GEO) ([188]http://www.ncbi.nlm.nih.gov/geo/) public databases. Cell lines KG-1, MV-4-11, OCI-AML3, U937, and 293T cell lines were generously gifted by Professor Yang Liang (State Key Laboratory of South China, Sun Yat-sen University cancer center). Hel and THP1 cell lines were obtained from Professor Yuanbin Song (State Key Laboratory of South China, Sun Yat-sen University cancer center). Details of the cell lines are shown in [189]Table S5. KG-1, MV-4-11, OCI-AML3, U937, Hel, and THP1 cell lines were grown in RPMI medium (Gibco, Cat#C11875500BT). The 293T cell line was cultured using DMEM medium (Gibco, Cat#C11995500BT) and passaged with 0.25% Trypsin-EDTA (Gibco, Cat#25200072). All cell lines were cultured under controlled conditions at 37°C in 5% CO[2] and complemented with 10% Fetal bovine serum (FBS) (WISENT, Cat#086-150), as well as 1% penicillin (with a final concentration of 100 U/ml) and 1% streptomycin (with a final concentration of 100 μg/ml) (Gibco, Cat#15140122). All cell lines were identified by short tandem repeat and tested negative for mycoplasma. Mouse models The control and TSC22D3 knockdown groups were mixed with an equal volume of Matrigel (CELLada, Cat#CELLada-OM-2) and implanted by subcutaneous injection into the right dorsum of 4–6-week-old male BALB/c nude mice (5 × 10ˆ6 cells/mouse, 150 ul) (Mice were provided by Sun Yat-sen University Cancer Center) to establish a cell-derived xenograft model (CDX). When the tumor reached a measurable size, mice were treated with intraperitoneal injection of 20 mg/kg Ara-c every three days. The tumor dimensions including the longest diameter (D), shortest diameter (D), as well as its height (H), were measured every five days using vernier calipers. The formula (D∗d∗H) ∗ π/6 was used to calculate the tumor volume. Mice were euthanized if they lost more than 20% of their body weight or if the tumor size exceeded 2 cm^3. The ethics of animal experiments in this study strictly followed the relevant provisions of the Guide for the Care and Use of Laboratory Animals published by the National Academic Press and was also approved by the Laboratory Animal Ethics Committee of Sun Yat-sen University Cancer Control Center (Approval number: L025504202209001). Method details Data source, collation, and differential expression analysis RNA-seq data from TCGA and GTEx databases were downloaded from UCSC XENA database and converted to transcripts per million reads (TPM) format by Toil process uniformly.[190]^43 The [191]GSE15061 and [192]GSE13159 datasets were obtained from GEO database. All data were converted to log2, and duplicate samples and samples with missing gene expression data were removed. The ggplot2 (v3.3.3) package was used for data visualization.[193]^38 Based on the TCGA datasets, DESeq2 (v1.26.0) package was used to detect DEGs between patients with high versus low TSC22D3.[194]^39 Genes that met the criteria of having an adjusted |log2(FC)| > 1 and p.adj < 0.05 were considered statistically relevant. Gene Ontology (GO), immunoinfiltration, and gene set enrichment analysis (GSEA) The org.Hs.eg.db (v3.10.0) package and clusterProfiler (v3.14.3) package were used to perform GO analysis.[195]^40 Immunoinfiltration analysis of TSC22D3 was performed by single sample GSEA (ssGSEA) with the GSVA (v1.34.0) package.[196]^41 Inferring the relationship between TSC22D3 and infiltration of 24 immune cells in the tumor microenvironment using Spearman correlation analysis.[197]^44 Additionally, GSEA and visualization of DEGs with clusterProfiler (v3.14.3) package.[198]^40 The condition that the False Discovery Rate (FDR) < 0.25 and p.adj < 0.05 was considered statistically significant enrichment by repeating the calculation 1000 times. Establishment and prediction of a prognostic model Kaplan-Meier (K-M) curves analysis of 8-year Overall survival (OS) differences between high versus low TSC22D3 expression groups and various mutation genotype subgroups (Subgroups with a total sample number of less than 5 were not included in the analysis). Clinical characteristics with p-values less than 0.1 in univariate Cox regression analysis were considered for multivariate Cox regression analysis. All statistical p-values less than 0.05 were considered significant. The timeROC (v0.4) package[199]^42 and ggplot2 (v3.3.3) package[200]^38 were used to conduct the ROC model. Calibration plots and nomograms were constructed with the RMS (v6.2-0) and survivor (v3.2-10) packages. RT-qPCR The cell samples underwent two washes with phosphate-buffered solution (PBS) (MCE, Cat#C10010500BT) at 1000 rpm for 5 minutes. Total RNA was extracted from cells according to the standard protocol using TRIzol® reagent (ThermoFisher, Cat#15596018). RNase-free water was utilized to dissolve RNA, and the RNA quality and concentration were determined with the NanoDrop microspectrophotometer (ThermoFisher, Cat#840-317500). Reverse transcription of 1ug total RNA to cDNA using the Fast Reverse Transcription kit (ESscience, Cat#RT001) according to the standard protocol. The 2 x Super SYBR Green qPCR Master Mix (with ROX) (ESscience, Cat#QP002) and Bio-Rad Quantitative real-time PCR instrument (Bio-Rad, Cat#CFX96 Touch) were used for conducting quantitative real-time PCR. The appropriate annealing temperature was set according to the primer temperature. The primer sequences used to identify gene expression in human GAPDH, TSC22D3, CD68, CD80, tumor necrosis factor-α (TNF-α), interleukin-1β (IL-1β), CD206, PPARγ, and CCL22 are detailed in [201]Table S1. Statistical analysis of the results was performed using the 2ˆ-ΔΔCt method. Western blotting The cells underwent two washes with PBS and were subsequently lysed on ice for 40 minutes in RIPA buffer, which contained 1% phenyl methane sulfonyl fluoride (PMSF) (UBIO, Cat#UW0103) and 1% Phosphatase Inhibitor Cocktail (CWBIO, Cat#CW2383S). The protein concentration was measured using the bicinchoninic acid (BCA) Protein Assay Kit (CWBIO, Cat#CW0014S) following the standard protocol and detected with a microplate reader at a wavelength of 562 nm. Adding 5 × loading buffer (Fudebio, Cat#FD006) to the protein samples and heat at 95°C for 15 minutes in a metal bath. Using the 12.5% SDS-PAGE Gel Fast Preparation Kit (EpiZyme, Cat#PG113) to separate protein samples, pre-marker proteins (ThermoFisher, Cat#26616) were used to mark protein locations. Polyvinylidene fluoride (PVDF) membranes with 0.22 μm (for molecular weights less than 20 KDa) (Biosharp, Cat#BS-PVDF-22) or 0.45 μm (for molecular weights greater than 20 KDa) (Merck Millipore, Cat#IPVH00010) pore size was activated by 100% methanol (Macklin, Cat#M813895) for 5 minutes. The isolated proteins were then transferred from the gel to the PVDF membrane using the BIO-RAD wet transfer system and immobilized. The PVDF membranes were blocked using 5% skim milk (Wako, Cat#190-12865) dissolved in TBST (20 mM Tris base (Sigma, Cat#V900483), 150 mM NaCl (Biofroxx, Cat#1249GR500), 0.1% Tween 20 (Solarbio, Cat#T8220), and pH = 7.6) buffer for 90 minutes at 25°C. Primary antibodies were incubated at 4°C overnight, while secondary antibodies were incubated at 25°C for 2 hours. Detailed dilution ratios of all antibodies see [202]Table S6. Universal antibody diluent (EpiZyme, Cat#PS119L) was used to dilute all western blotting antibodies. The BIO-RAD chemiluminescence imaging system was used to detect protein expression, and ImageJ (National Institutes of Health, USA, v1.8.0) was used to analyze protein relative grayscale values. Short hairpin RNA (shRNA) knockdown The pLKD-U6-MCS-CMV-EGFP-2A-Puro (Umine-bio, Cat#G0202296-1) plasmid was used to construct the non-targeting shRNA vector (Designated as pLKD) and the targeting human TSC22D3 shRNA vector (Designated as shTSC22D3). The targeting sequences are shown in [203]Table S2. The empty plasmid was digested with AgeI-HF (NEB, Cat#R3552L) and EcoRI-HF (NEB, Cat#R3101V) restriction enzymes according to the standard protocol to create sticky ends and formed linear plasmids. The single-stranded cDNA was formed double-stranded DNA with sticky ends according to the standard protocol under annealing buffer (Beyotime, Cat#D0251). Double-stranded oligo and linear plasmids formed pLKD-U6-shRNA (non-targeting)-CMV-EGFP-2A-Puro and pLKD-U6-shRNA (TSC22D3)-CMV-EGFP-2A-Puro recombinant plasmids under the action of T4 DNA ligase (NEB, Cat#M0201S) according to the standard protocol. Stbl3 competent cells (TSINGKE, Cat#TSC-C06) were used to amplify the plasmids. The positive monoclonal screening was then performed on Luria-Bertani (LB) solid medium (containing 10 mg/ml Tryptone (Oxoid, Cat#LP0042B), 5 mg/ml Yeast extract power (Oxoid, Cat#LP0021B), 5 mg/ml NaCl, and 100 ug/ml Ampicillin (Amp) (MDbio, Cat#A010)) following the standard protocol. The Endotoxin-Free Plasmid Extractor Kit (TIANGEN, Cat#DP118-02) was used to extract and purify positive plasmids following the standard protocol. The plasmid’s quality and concentration were determined with the NanoDrop microspectrophotometer (ThermoFisher, Cat#840-317500). Lentiviral production and transduction 293T cells were cultured to produce viral fluid in Opti-MEM (ThermoFisher, Cat#31985070) containing DNA (Recombinant plasmids: psPAX2: pVSVG = 3: 2: 1) and lipofectamine 3000 (Invitrogen, Cat#L3000015) for 8 hours following the standard protocol. After replacing the complete medium, the cells were incubated for an additional 36, 48, and 72 hours. The conditioning medium was filtered through a 0.45 μm filter and infected with Hel cells for 24 hours. EGFP-positive cells were observed and sorted using a fluorescence microscope and the Beckman Moflo Astrios EQ system after 4 days of infection. Cell viability assay EC50, proliferation, and viability were detected using the CCK-8 (Apex Bio, Cat#K1018) method. The control and TSC22D3 knockdown groups cells were cultured in 96-well plates (Corning, Cat#3599) at a density of 1 x 10ˆ4 cells/well for cell proliferation and at a density of 5 x 10ˆ4 cells/well for cell viability and EC50 assays. The cells were treated with 100 nM Cytarabine (Ara-c) (Apex BIO, Cat#A8405) for 4 days, with or without treatment. EC50 was determined using isometrically diluted concentrations of Ara-c for 48 hours. Optical density (OD) was then measured using a Biotech-Epoch microplate reader with a wavelength of 450 nm. Annexin V and 7-AAD staining 6-well plates (Corning, Cat#3516) were seeded with cells from the control and TSC22D3 knockdown groups with a density of 1 × 10ˆ6 cells in each well. Cells were harvested and washed three times with ice-cold PBS for 5 minutes at 1000 rpm after 2 days of treatment with 100 nM Ara-c. Apoptotic rates were analyzed using the 7-AAD-APC-Annexin V Apoptosis Assay (Tonbo, Cat#20-6410-KIT) according to standard protocols and the Beckman Cytoflex lx system. Results were analyzed using FlowJo (BD Biosciences, v10.8.1) software. Cell cycle analysis The control and TSC22D3 knockdown group cells were cultured in Opti-MEM medium for 24 hours to synchronize the cell cycles. The medium was subsequently altered, and cells were incubated for 48 hours with or without 100 nM Ara-c. The cell cycle was analyzed with the Cell Cycle Analysis kit (4Abio, Cat#FXP0211) and the Beckman Cytoflex lx system following the standard protocol. The results were analyzed using FlowJo (BD Biosciences, v10.8.1) software, and cell cycle model constraints were G2 peak = 2 G1, G1 coefficient of variation (CV) = G2 CV = 8.00. THP1-induced differentiation macrophages THP1 cells were seeded in 6-well plates at a density of 1 x 10ˆ6 cells/well and treated with 185ng/ml phorbol-12-myristate-13-acetate (PMA) (MCE, Cat#HY-18739) for 6 hours to induce differentiation into M0 macrophages. To induce differentiation of cells into M1 macrophages, interferon-γ (IFN-γ) (NovoProtein, Cat#C014) at a concentration of 20 ng/ml and lipopolysaccharide (LPS) (Bioss, Cat#bs-8000P) at a concentration of 100 ng/ml were added. This was followed by incubation for 48 hours in the presence of PMA. Similarly, the cells were differentiated into M2 macrophages with 20 ng/ml interleukin-4 (IL-4) (NovoProtein, Cat#C050) and 20 ng/ml interleukin–13 (IL-13) (NovoProtein, Cat#CC89) stimulation for 48 hours. RT-qPCR to assess the expression levels of mRNA markers for M1 and M2 macrophages. Co-culture The direct co-culture system was used to evaluate the impact of TSC22D3-knocked tumor cells on macrophage polarization. THP1 cells were seeded in 6-well plates at a density of 1 x 10ˆ6 cells/well. The cells were then differentiated into M0 macrophages through PMA stimulation for a period of 6 hours. TSC22D3 knockdown Hel cells and their relative control were cultured with macrophages at a 1: 1 ratio with M1 polarizing stimulators (IFN-γ and LPS) or M2 polarizing stimulators (IL-4 and IL-13) for 48 hours in the continued presence of PMA. The Hel cell suspension and medium were then removed, and macrophages were washed three times with PBS before RNA extraction. RT-qPCR analysis was adopted to analyze the mRNA abundance of M1-or M2 macrophage-related genes. Preparation of single-cell suspension of mouse tumor tissue Tumor tissue was minced into small pieces of 1-3 mm^3 and then incubated in the complete medium that contained type I collagenase (final concentration 1 mg/ml) (Sigma, Cat#C0130) and DNase I (final concentration 100 U/ml) (Acmec, Cat#[204]D61780) at 37°C with constant shaking (100 rpm) for 4 hours. The tissue suspension was washed three times with PBS at 1000 rpm for 5 minutes each after passing through a 70 μm cell strainer. Cells were lysed using a red blood cell lysate (Beyotime, Cat#C3702) and cells were allowed to stand on ice for 5 minutes. The single-cell suspension was obtained after three washes using PBS at 1000 rpm for 5 minutes. Macrophages infiltration analysis The samples underwent three rounds of washing with ice-cold PBS for 5 minutes at 1000 rpm. Flow cytometry staining buffer (PBS with 3% FBS) containing different flow cytometry antibodies was added. The 7-AAD Viability Staining Solution (BioLegend, Cat#420404, used 1: 100) was used to distinguish between living and dead cells. Myeloid cells were identified using CD11b antibody, while macrophages were labeled using F4/80 antibody. M1-type macrophages were marked with CD80 antibody, and M2-type macrophages were labeled with CD206 antibody. Detailed dilution ratios of all antibodies see [205]Table S6. Macrophage infiltration in mouse subcutaneous tumor tissues was analyzed using the Beckman Cytoflex lx and FlowJo (BD Biosciences, v10.8.1) software. Enzyme-linked immunosorbent assay (Elisa) The control and TSC22D3 knockdown group cells were plated at 5 × 10ˆ5 cells per well in 6-well culture plates. After 4 days of culture, the supernatant was collected via 5-minute centrifugation at 1000 rpm. The Human IL-18 ELISA Kit (MultiSciences, Cat#EK118) and Human IL-1β ELISA Kit (MultiSciences, Cat#EK101B) were used to detect the cytokine levels in the media supernatant according to the standard protocol. Absorbance at 450 nm and 630 nm was measured using a Biotech-Epoch microplate reader. Statistics were analyzed using calibrated OD values, acquired by deducting the OD value at 630 nm from the OD value at 450 nm. Quantification and statistical analysis Shapiro-Wilk test was conducted to assess the normality of the data. Additionally, the homogeneity of variance was evaluated using Levene's test. Differences between the two groups were compared using Student's t-test or Wilcoxon rank sum test. The overall comparison of multiple groups was analyzed using one-way ANOVA test or Kruskal-Wallis’s test. Meanwhile, the two-way repeated measures ANOVA was employed for two-factor group analysis. Bioinformatics analysis was performed using RStudio (Posit Software Inc., USA, v3.6.3) software for statistical analysis. All experimental data were statistically analyzed and plotted using GraphPad Prism 9.5 (GraphPad Software Inc., CA, USA) software. All experiments were repeated independently at least three times (n ≥ 3). Acknowledgments