mRNA expression data (59 normal cases, 539 tumor cases) of LUAD were downloaded from The Cancer Genome Atlas (TCGA) and subjected to differential analysis ($|$logFC$|$ $>$1.0, false discovery rate (FDR) 0.05), which was carried out by the “edgeR” package 4.2.3 (Posit, Boston, MA, USA), to get differentially expressed mRNAs (DEmRNAs) between normal and tumor groups. Then, by combining with the literature [29], the target gene for the study was determined. Next, the pathways regulated by target gene were predicted through gene set enrichment analysis (GSEA, $|$NES$|$ $>$1.0, FDR 0.25), and transcription factors upstream of target gene were predicted by TFtarget. Intersection of transcription factors from prediction and DEmRNAs was obtained by overlapping and visualized by the upset plot generated by the R package “UpSetR” (http://github.com/hms-dbmi/UpSetR). Pearson correlation analysis was conducted between obtained transcription factors and target gene, and the transcription factor with a higher correlation was selected as the research object. JASPAR was utilized to predict motif binding site at 2000 bp upstream of target gene. The immune infiltration data of all tumor samples of TCGA were accessed from Timer2.0 (http://timer.comp-genomics.org/timer/), and the expression matrices of UBE2T in TCGA samples were integrated with the expression matrices of individual immune cells using various algorithms such as TIMER, CIBERSORT, CIBERSORT-ABS, QUANTISEQ, MCPCOUNTER, XCELL, and EPIC and other algorithms to compute immune cell infiltration in TCGA samples, integrate expression matrix of UBE2T in TGCA-LUAD samples with the infiltration matrix of individual immune cells, and conduct Pearson correlation analysis.

Human lung bronchial epithelial cells (BEAS-2B), LUAD cells (A549, H460, and H1650), 293T cells, and peripheral blood mononuclear cells (PBMCs) were purchased from ATCC (Manassas, VA, USA), and mouse LUAD cells (LA795) were purchased from Pricella (Wuhan, China). BEAS-2B cells were maintained in MEM with 10% fetal bovine serum (FBS; PAN, Aidenbach, Germany), A549 cells in F12K medium (Invitrogen, Waltham, MA, USA) containing 10% FBS (PAN, Germany), H460 cells and H1650 cells in RPMI-1640 medium (Invitrogen, Waltham, MA, USA) with 10% FBS (PAN, Germany), 293T cells in DMEM (Invitrogen, Waltham, MA, USA) with 10% FBS (PAN, Germany) and 2 mM L-glutamine (Invitrogen, Waltham, MA, USA), and PBMC cells in HBSS medium (Invitrogen, Waltham, MA, USA) containing 10% FBS (PAN, Germany). LA795 cells were placed in RPMI-1640 culture-medium (Invitrogen, Waltham, MA, USA) mixed with 10% FBS (PAN, Germany), streptomycin (100 µg/mL, Corning, Shanghai, China), penicillin G (100 U/mL, Beyotime, Shanghai, China). The cultured cells were incubated in humidified atmosphere with 5% CO${}_2$ at 37 °C. All cell lines were validated by STR profiling and tested negative for mycoplasma. Glycolysis inhibitor 2-deoxy-d-glucose (2-DG) was purchased from Warbio (Nanjing, China).

Short hairpin RNAs (shRNAs) targeting UBE2T and FOXA1 (sh-UBE2T, sh-FOXA1) and their negative controls (sh-NC) were produced by GenePharma (Shanghai, China), and the carrier pcDNA-UBE2T and its negative control pcDNA-NC were accessed from RiboBio (Guangzhou, China). As per the instructions, sh-UBE2T, sh-FOXA1, sh-NC, pcDNA-UBE2T, and pcDNA-NC were used to transfect LUAD cells using Lipofectamine 2000 (Invitrogen, Waltham, MA, USA).

Total RNA extraction from tissues and cells was conducted by using TRIzol reagents (Invitrogen, Waltham, MA, USA), and RNA concentration and purity were quantitatively determined with a NanoDrop ND-1000 spectrophotometer (NanoDrop, Waltham, MA, USA). Revert Aid™ First Strand cDNA Synthesis Kit (Thermo Fisher, Waltham, MA, USA) was applied to synthesize cDNA from RNA. ChamQ SYBR Color qPCR Master Mix (High ROX Premixed; Vazyme, Nanjing, China) was utilized to perform qPCR on ABI 7500 PCR instrument (Applied Biosystems, Waltham, MA, USA). Total reaction volume was 20 µL. Relative expression of genes was computed by applying 2${}^{-\mathrm{\Delta }\mathrm{\Delta }\text{CT}}$. $\beta$-actin gene was taken as a control. Sequences of primers are listed in Table 1.

LUAD cells were inoculated into 96-well plates (1 $\times$ 10${}^3$ cells/well). After incubation at 37 °C for 0, 24, 48, and 72 h, medium was replaced and 20 µL MTT reagent was introduced (5 mg/mL; Sigma-Aldrich, St. Louis, MO, USA). After culture at 37 °C for 4 h, supernatant was discarded and 200 µL DMSO (Invitrogen, Waltham, MA, USA) was introduced. Absorbance at 570 nm was obtained by a microplate reader (Varioskan LUX, Invitrogen, Waltham, MA, USA).

Total protein extraction from tissues and cells was performed with RIPA buffer (Sigma, St.Louis, MO, USA), followed by protein concentration determination by a BCA protein detection kit (Thermo Fisher, Waltham, MA, USA). SDS-PAGE was run for protein isolation with 10% polyacrylamide gel (Beyotime, Shanghai, China) and transferred to a polyvinylidene fluoride membrane (Beyotime, Shanghai, China). Thereafter, membrane was incubated with primary antibodies (Abcam, Cambridge, UK) of the proteins PD-L1 (1:1000), UBE2T (1:1000), GLUT1 (1:100,000), LDHA (1:5000), HK2 (1:1000) and FOXA1(1:5000) overnight. The next step was incubation for 1 h with second antibody IgG (Abcam, Cambridge, UK) at room temperature. $\beta$-actin acted as a control, and a rabbit anti-$\beta$-actin antibody (Abcam, Cambridge, UK) was adopted. Protein bands were developed after the introduction of an ECL-based chemiluminescence substrate (Millipore, Burlington, MA, USA).

T cells were purified from PBMCs using magnetic beads from Miltenyi Biotech (Bergisch Gladbach, North Rhine-Westphalia, Germany). Then, CD8${}^{+}$T cells were activated by the addition of IL-2 (10 IU/mL, Abcam, Cambridge, UK), anti-CD3 (2.5 µg/mL; Abcam, Cambridge, UK), and anti-CD28 (2 µg/mL, Abcam, Cambridge, UK) [30]. Afterward, treated LUAD cells were mixed with activated CD8${}^{+}$T cells (the ratio of reactive agent to the stimulus was 1:1). They were then maintained in RPMI-1640 medium, which contained 10 mM HEPES, 10% heat-inactivated FBS, 100 ng/mL streptomycin and 100 U/mL penicillin, for 96 h [31].

CD8${}^{+}$T cell cytotoxicity was assayed with lactate dehydrogenase (LDH) cytotoxicity kit (Thermo Fisher, Waltham, MA, USA). Cytotoxicity was measured following: Cytotoxicity% = [(Chemical-treated LDH activity – Spontaneous LDH activity)/(Maximum LDH activity – Spontaneous LDH activity)] $\times$ 100% [32].

Interferon $\gamma$ (IFN $\gamma$), tumor necrosis factor $\alpha$ (TNF $\alpha$), and interleukin (IL) 2 levels were tested with ELISA kits (Abcam, Cambridge, UK). In short, supernatant of each co-culture group was harvested. Next, each sample well was added with 100 µL enzyme conjugate and allowed for reaction at 37 °C for 30 min. Then, each well was introduced with 100 µL horseradish peroxidase substrate solution for color development at 37 °C for 20 min. Lastly, each well was added with 50 µL termination solution and the absorbance at 450 nm was assessed within 20 min [32].

We measured cellular glycolysis capacity and mitochondrial function by extracellular flow analyzer (Seahorse Bioscience, North Billerica, MA, USA) as per instructions. The day before assay, cells were placed in cell culture microporous boards (Seahorse Bioscience, North Billerica, MA, USA). Seahorse buffer was a mixture of DMEM, 2 mM sodium pyruvate, phenol red, 25 mM glucose, and 2 mM glutamine. For ECAR value test, 10 mM glucose, 1 µM oligomycin, and 100 mm 2-DG were automatically added. Following baseline respiration monitoring, 1 µM oligomycin, 1 µM carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone (FCCP), and 1 rotenone were automatically injected into the microporous boards to assess OCR [1].

The fragment of UBE2T was inserted into pGL3 promoter vector to construct pGL3-UBE2T-promoter-WT (wild type) or pGL3-UBE2T-promoter-MUT (mutation type) vector. The luciferase reporter vector and sh-FOXA1/sh-NC were used for co-transfection of 293T cells by Lipofectamine 2000 reagent. The 293T cells were incubated for 48 h, followed by luciferase activity measurement in dual luciferase assay system.

After 37% formaldehyde (Beyotime, Shanghai, China) treatment, the cells were collected for sonication with VCX750 (SONICS, Newtown, CT, USA) at 25% power. Sonication was performed for 4.5 s each time, with 14 times in total and 9 s intervals. Next, centrifugation at 10,000 $\times$g was performed at 4 °C for 10 min, and insoluble matter was discarded. Then, an antibody against FOXA1 (Abcam, Cambridge, UK) and a control IgG antibody (Abcam, Cambridge, UK) were introduced to bind to target protein-DNA complex. Then, protein A was applied to sediment antibody-target protein-DNA complex. After washing the sediment, the sediment was eluted and de-crosslinked. Finally, the purified DNA sequences were used for qPCR [33]. PCR primer sequences were as follows: forward direction, 5${}^{\prime }$-AGTGAACATCTGGGTTGGTAAA-3${}^{\prime }$ and reverse direction, 5${}^{\prime }$-AGTGAACATCTGGGTTGGTAAA-3${}^{\prime }$.

The paraffin-embedded sample was sliced into 4 µm-thick slices using a paraffin microtome (Leica, Wetzlar, Germany). IHC staining was completed according to the standard protocol described previously [18]. After initial dewaxing, antigen repair, and sealing, the slices were incubated overnight with anti-UBE2T antibody (Abcam, Cambridge, UK) in a humid container at 4 °C. The next day, slices were treated with horseradish peroxidase-labeled secondary antibody IgG (Abcam, Cambridge, UK). DAB (Novus Biologicals, Littleton, CO, USA) was adopted to develop UBE2T staining, and hematoxylin counterstaining was conducted.

Thirty male mice (BALB/c, 4 weeks old, 18–22 g) were from SLAC Laboratory Animal Co., Ltd. (Shanghai, China). Animal experiments were conducted with approval from the Ethics Committee of Affiliated Hospital of Southwest Medical University. Mice were kept in controlled conditions with 25 $\pm$ 2 ${}^{\circ }$C, a humidity of 70%, a light-dark cycle of 12 h, and regular food and water feeding. LA795 cells (1 $\times$ 10${}^5$) with sh-NC or sh-UBE2T were subcutaneously inoculated into the right flank of each mouse. The formula of V(volume) = W(width)${}^2$$\times$ L(length)/2 was adopted for calculating tumor volume, and mice were weighed every 5 days. On day 30, mice were euthanized, and tumor weight was measured. Tumor tissues were collected from each group and the tumor growth was measured.

Data were processed using GraphPad Prism 8.0 (GraphPad Software, Inc., San Diego, CA, USA). All measured data were presented in the form of mean $\pm$ standard deviation (SD). T-test was applied for detecting differences between two groups. Each experiment was conducted at least three times. p 0.05 indicated a significant difference.

Previous studies have found that UBE2T is up-regulated and promotes cancers of various types, such as hepatocellular carcinoma [34] and esophageal squamous cell carcinoma [35]. Hence, we explored UBE2T expression in LUAD. Bioinformatics results showed a significant up-regulation of UBE2T in LUAD tissues (Fig. 1A). qRT-PCR and western blot detection also showed an evident higher UBE2T expression in LUAD cells than in normal lung epithelial cells (Fig. 1B,C). These results suggested the aberrantly high expression of UBE2T in LUAD. Since UBE2T had higher relative expression in A549 cells and lower expression in H460 cells among the tested cell lines, A549 was selected for the subsequent knockdown experiment and H460 for the overexpression experiment.

Fig. 1.

UBE2T expression is up-regulated in LUAD. (A) UBE2T expression in LUAD tissues analyzed by bioinformatics approaches. (B,C) Expression of UBE2T in LUAD cells analyzed by western blot and qRT-PCR. * indicates p 0.05. UBE2T, ubiquitin binding enzyme E2T; LUAD, lung adenocarcinoma; qRT-PCR, quantitative reverse transcription polymerase chain reaction.

UBE2T is associated with immune cell infiltration [24]. We found by bioinformatics analysis that UBE2T had a negative correlation with CD8${}^{+}$T cell infiltration (Fig. 2A). The following cell lines were constructed: sh-NC, sh-UBE2T, oe-NC, and oe-UBE2T. As qRT-PCR and MTT results revealed, sh-UBE2T substantially reduced UBE2T expression and cell viability, while oe-UBE2T significantly increased UBE2T expression and cell viability (Fig. 2B,C). The above cells were co-cultured with activated CD8${}^{+}$T cells separately. Western blot showed down-regulated PD-L1 expression after sh-UBE2T treatment and up-regulated PD-L1 expression after oe-UBE2T treatment (Fig. 2D). In addition, after sh-UBE2T treatment, CD8${}^{+}$T cells showed enhanced toxicity to A549 cells and increased expression of cytokines (IFN-$\gamma$, IL-2, TNF-$\alpha$), while after oe-UBE2T treatment, CD8${}^{+}$T cells showed reduced toxicity to H460 cells and decreased expression of cytokines (Fig. 2E,F). The above experiments demonstrated that high UBE2T expression in LUAD hindered CD8${}^{+}$T cell activity.

Fig. 2.

High expression of UBE2T inhibits the activity of CD8${}^{\mathrm{+}}$T cells. (A) Bioinformatics analysis on correlation between UBE2T and CD8${}^{+}$T cell infiltration. (B) qRT-PCR analysis on UBE2T expression in LUAD cells after knockdown and overexpression. (C) MTT assay results on LUAD cell activity after knockdown and overexpression. (D) Western blot analysis on PD-L1 expression. (E) LDH toxicity assay on cytotoxicity of CD8${}^{+}$T cells co-cultured with LUAD cells. (F) ELISA results on the expression of cytokines. * indicates p 0.05. ELISA, enzyme linked immunosorbent assay; PD-L1, programmed death ligand 1; MTT, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide; LDH, Lactate dehydrogenase; sh, short hairpin RNA; oe, overexpression; NC, negative control; OD, optical density; IFN-$\gamma$, Interferon-$\gamma$; IL-2, interleukin-12; TNF-$\alpha$, tumor necrosis factor-$\alpha$.

Next, we investigated mechanism that UBE2T affects activity of CD8${}^{+}$T cells in LUAD. As we found, UBE2T was divided into two groups based on UBE2T expression: UBE2T-high and UBE2T-low. Single gene GSEA presented that UBE2T was highly expressed in the glycolysis signaling pathway (normalized enrichment scores (NES) = 1.51, FDR = 0.141) (Fig. 3A, Supplementary Table 1), and its expression was positively correlated with genes related to glycolysis (Fig. 3B). We hypothesized that UBE2T high expression could inhibit CD8${}^{+}$T cell activity through the glycolysis pathway. To verify this conjecture, we constructed the following cell lines based on H460 cells: oe-NC (oe-NC+DMSO), oe-UBE2T (oe-UBE2T+DMSO), and oe-UBE2T+2-DG. Seahorse XF extracellular flow analyzer was applied for analyzing the effect of abnormal UBE2T expression on glycolysis. ECAR reflects aerobic glycolysis flux, and OCR demonstrates mitochondrial oxidative respiration state. As our results displayed, the ECAR level was notably increased in H460 cells with UBE2T overexpression, reflecting that UBE2T overexpression could promote glycolysis level, while the addition of 2-DG restored ECAR value to the control level (Fig. 3C). Cellular oxygen consumption indicates mitochondrial respiration level, and OCR decreases when aerobic glycolysis occurs. In contrast to ECAR results, we observed that UBE2T overexpression reduced OCR level in H460 cells, but this was reversed by the addition of 2-DG (Fig. 3D). Expression of glycolysis-related genes was analyzed by qRT-PCR and western blot. Overexpression of UBE2T significantly enhanced GLUT1, LDHA and HK2 expression, but 2-DG reversed this result (Fig. 3E,F). Then, we tested contents of pyruvate, lactic acid, citric acid, and malic acid in cells in each group using corresponding kits. The results were that these contents in cells increased after oe-UBE2T treatment, while these indicators were reversed by 2-DG treatment (Fig. 3G). In conclusion, UBE2T positively regulated aerobic glycolysis in LUAD cells.

Fig. 3.

High expression of UBE2T inhibits CD8${}^{\mathrm{+}}$T cell activity by activating glycolysis. (A) Bioinformatics analysis results of UBE2T enrichment pathway. (B) Pearson correlation analysis of UBE2T and glycolytic related genes. (C,D) Determination of ECAR and OCR of H460 cells after different treatments. (E,F) qRT-PCR and western blot analysis results of expression of glycolysis-related genes after different treatments. (G) Contents of pyruvate, lactic acid, citric acid and malic acid in H460 cells after different treatments. (H) Western blot analysis results of PD-L1 expression. (I) Cytotoxicity assay results of CD8${}^{+}$T cell cytotoxicity to H460 cells. (J) ELISA detection results on cytokine expression. * indicates p 0.05. ECAR, extracellular acidification rate; OCR, oxygen consumption rate; 2-DG, 2-deoxy-d-glucose.

Next, H460 cells with oe-NC, oe-UBE2T, or oe-UBE2T+2-DG were subjected to co-culture with activated CD8${}^{+}$T cells. As western blot showed, UBE2T overexpression could largely increase PD-L1 expression, but the addition of 2-DG restored PD-L1 to the control level (Fig. 3H). According to cytotoxicity test results, CD8${}^{+}$T cytotoxicity to H460 cells was weakened after UBE2T overexpression, and impact of UBE2T overexpression on CD8${}^{+}$T cell cytotoxicity was counteracted after 2-DG was added (Fig. 3I). ELISA showed similar results in terms of the expression of related cytokines. Overexpression of UBE2T resulted in decreased levels of cytokines (IFN-$\gamma$, IL-2, TNF-$\alpha$), but this was reversed by the addition of 2-DG (Fig. 3J). In conclusion, high expression of UBE2T in LUAD could repress CD8${}^{+}$T activity by promoting glycolysis levels.

To explore the transcription factors upstream of UBE2T, hTFtarget was utilized to predict UBE2T upstream potential transcription factors, which were then intersected with up-regulated DEmRNAs. Finally, a total of 17 potential transcription factors were obtained (Fig. 4A). FOXA1 is associated with the growth and cell characteristics of LUAD [36]. Further, Pearson correlation analysis demonstrated a positive correlation between UBE2T and FOXA1 (Fig. 4B). JASPAR prediction showed a binding site of UBE2T to FOXA1 in the promoter region of UBE2T (Fig. 4C). Bioinformatics analysis told high FOXA1 expression in LUAD tissues (Fig. 4D), while qRT-PCR and western blot results presented up-regulated FOXA1 level in LUAD cells (Fig. 4E,F). Thereafter, we continued to probe mechanism of influence of FOXA1 and UBE2T on LUAD. As the FOXA1 level in A549 cells was the highest among the tested cell lines, A549 was selected for subsequent experiments. Subsequent ChIP assay showed that FOXA1 antibody could significantly enrich UBE2T compared with IgG (Fig. 4G). The dual luciferase assay found that sh-FOXA1 led to a substantially reduced luciferase activity in UBE2T-WT but caused no great change in the UBE2T-MUT group (Fig. 4H). Finally, the expression of UBE2T in sh-FOXA1-treated A549 cells was assayed by qRT-PCR. UBE2T was significantly down-regulated (Fig. 4I). The above experiments showed that FOXA1 activated UBE2T transcription.

Fig. 4.

FOXA1 activates UBE2T transcription. (A) Upset plot of the intersection between hTFtarget predicted transcription factors and up-regulated DEmRNAs. (B) Pearson correlation analysis of UBE2T and FOXA1. (C) JASPAR predicted binding site between UBE2T and FOXA1. (D) FOXA1 expression in LUAD analyzed by bioinformatics means. (E,F) qRT-PCR and western blot results on FOXA1 expression in LUAD cells. (G,H) ChIP and dual luciferase assay results on the binding relationship between UBE2T and FOXA1. (I) qPCR results on UBE2T expression after sh-FOXA1 treatment. * indicates p 0.05. FOXA1, forkhead box A1; ChIP, chromatin immunoprecipitation; hTFtarget, database of human transcription factor targets; DEmRNAs, differentially expressed mRNAs; WT, wild type; MUT, mutation type.

To verify the influence of FOXA1/UBE2T on CD8${}^{+}$T cell activity in LUAD, we constructed the following cell lines based on A549 cells: sh-NC+oe-NC, sh-FOXA1+oe-NC, and sh-FOXA1+oe-UBE2T. qRT-PCR and MTT were applied for analysis of transfection efficiency and viability of cells respectively. As the results showed, sh-FOXA1 notably impaired UBE2T expression and cell viability of A549, while oe-UBE2T treatment could counteract the effects of sh-FOXA1 on UBE2T expression and cell viability (Fig. 5A,B). According to the results from Seahorse XF extracellular flow analyzer, qRT-PCR and western blot, which detected glycolysis levels and expression of glycolysis-related genes respectively, sh-FOXA1 treated A549 cells had decreased ECAR, increased OCR, and reduced GLUT1, LDHA and HK2 expression; however, oe-UBE2T treatment reversed the above results (Fig. 5C–F). Pyruvate, lactic acid, citric acid, and malic acid contents in cells were tested by corresponding kits, and according to the results, sh-FOXA1 treatment reduced the above indicators, but oe-UBE2T counteracted the inhibitory effects of sh-FOXA1 on these indicators (Fig. 5G).

Fig. 5.

FOXA1/UBE2T inhibits CD8${}^{\mathrm{+}}$T cell activity through glycolysis. (A) qRT-PCR detection on transfection efficiency. (B) Cell viability assessed by MTT. (C,D) ECAR and OCR of differently treated cells. (E,F) qRT-PCR and western blot analysis on the expression of glycolysis-related genes after different treatments. (G) Contents of pyruvate, lactic acid, citric acid and malic acid in cells. (H) Western blot results on the expression of PD-L1. (I) CD8${}^{+}$T cell cytotoxicity assessed by cytotoxicity assay. (J) ELISA assay results on the expression of cytokines. * indicates p 0.05.

These cells were subjected to co-culture with activated CD8${}^{+}$T cells. A549 treated with sh-FOXA1 significantly inhibited PD-L1 expression, promoted CD8${}^{+}$T cell cytotoxicity, and elevated expression of IFN-$\gamma$, IL-2, and TNF-$\alpha$, while oe-UBE2T reversed these results (Fig. 5H–J). These results suggested that FOXA1 activated UBE2T expression in LUAD cells and inhibited CD8${}^{+}$T cell activity through the glycolysis pathway.

We constructed an allograft LUAD model to investigate the tumorigenicity of UBE2T in vivo. Compared with sh-NC, sh-UBE2T had a significant reduction effect on tumor volume and weight (Fig. 6A–C). qRT-PCR, western blot, and IHC analyses showed decreased UBE2T levels and PD-L1 expression after UBE2T knockdown (Fig. 6D,E). According to IHC results, mice treated with sh-UBE2T cells had higher infiltration of CD8${}^{+}$T cells than sh-NC group (Fig. 6F). In summary, UBE2T knockdown suppressed tumorigenicity and promoted the immune response of LUAD cells.

Fig. 6.

UBE2T can promote lung adenocarcinoma tumor growth and inhibit CD8${}^{\mathrm{+}}$T cell infiltration. (A,B) Changes of tumor volume in mice. (C) Effects of different treatments on tumor weight of mice. (D) UBE2T expression detected by qRT-PCR. (E) PD-L1 expression detected by western blot. (F) IHC results on CD8${}^{+}$T cell infiltration. * indicates p 0.05. IHC, immunohistochemistry.