PHF14 enhances DNA methylation of SMAD7 gene to promote TGF-β-driven lung adenocarcinoma metastasis

Tian, Han; Liu, Chenying; Yu, Jianchen; Han, Jian; Du, Jianan; Liang, Shujun; Wang, Wenting; Liu, Qin; Lian, Rong; Zhu, Ting; Wu, Shanshan; Tao, Tianyu; Ye, Yaokai; Zhao, Jingjing; Yang, Yi; Zhu, Xun; Cai, Junchao; Wu, Jueheng; Li, Mengfeng

doi:10.1038/s41421-023-00528-0

Download PDF

Article
Open access
Published: 18 April 2023

PHF14 enhances DNA methylation of SMAD7 gene to promote TGF-β-driven lung adenocarcinoma metastasis

Han Tian ORCID: orcid.org/0000-0003-0218-7246^1,2^na1,
Chenying Liu³^na1,
Jianchen Yu^1,4^na1,
Jian Han²,
Jianan Du¹,
Shujun Liang¹,
Wenting Wang¹,
Qin Liu¹,
Rong Lian¹,
Ting Zhu⁵,
Shanshan Wu⁶,
Tianyu Tao⁷,
Yaokai Ye¹,
Jingjing Zhao⁸,
Yi Yang¹,
Xun Zhu¹,
Junchao Cai ORCID: orcid.org/0000-0001-8412-6177¹,
Jueheng Wu¹ &
…
Mengfeng Li ORCID: orcid.org/0000-0002-5103-6367^1,2

Cell Discovery volume 9, Article number: 41 (2023) Cite this article

3134 Accesses
11 Citations
1 Altmetric
Metrics details

Subjects

Abstract

Aberrant activation of TGF-β signaling plays a pivotal role in cancer metastasis and progression. However, molecular mechanisms underlying the dysregulation of TGF-β pathway remain to be understood. Here, we found that SMAD7, a direct downstream transcriptional target and also a key antagonist of TGF-β signaling, is transcriptionally suppressed in lung adenocarcinoma (LAD) due to DNA hypermethylation. We further identified that PHF14 binds DNMT3B and serves as a DNA CpG motif reader, recruiting DNMT3B to the SMAD7 gene locus, resulting in DNA methylation and transcriptional suppression of SMAD7. Our in vitro and in vivo experiments showed that PHF14 promotes metastasis through binding DNMT3B to suppress SMAD7 expression. Moreover, our data revealed that PHF14 expression correlates with lowered SMAD7 level and shorter survival of LAD patients, and importantly that SMAD7 methylation level of circulating tumor DNA (ctDNA) can potentially be used for prognosis prediction. Together, our present study illustrates a new epigenetic mechanism, mediated by PHF14 and DNMT3B, in the regulation of SMAD7 transcription and TGF-β-driven LAD metastasis, and suggests potential opportunities for LAD prognosis.

Active demethylation upregulates CD147 expression promoting non-small cell lung cancer invasion and metastasis

Article Open access 07 February 2022

MPDZ as a novel epigenetic silenced tumor suppressor inhibits growth and progression of lung cancer through the Hippo-YAP pathway

Article 09 June 2021

RGS6 suppresses TGF-β-induced epithelial–mesenchymal transition in non-small cell lung cancers via a novel mechanism dependent on its interaction with SMAD4

Article Open access 28 July 2022

Introduction

Lung adenocarcinoma (LAD) is a major histological subtype of non-small cell lung carcinoma (NSCLC), which represents one of the most common malignant cancer types worldwide^1,2. Due to the generally imperceptible symptoms at the early stages of LAD development, more than 50% of LAD patients are found with intrapulmonary or distant metastasis at their initial diagnosis, leading to approximately only 20% overall 5-year survival rate of LAD patients and less than 5% 5-year survival rate of patients with metastatic LAD³. Hence, identification of mechanisms regulating LAD metastasis, aiming at providing potential diagnostic and prognostic biomarkers as well as therapeutic targets for metastatic LAD, is key to improving the clinical outcome of the disease.

Accumulated evidence has revealed a pivotal role of TGF-β signaling in cancer metastasis⁴. The canonical TGF-β signaling cascade is initiated by binding of specific ligands to TGF-β receptors (TGFBR), which triggers oligomeric TGFBR1/2-mediated phosphorylation of the cytoplasmic effectors SMAD2/3 (receptor-regulated SMADs, R-SMADs). Subsequently, the phosphorylation-activated SMAD2/3 forms a heterotrimeric complex with SMAD4 (common SMAD, co-SMAD) and is transported to the nucleus, where the heterotrimeric complex interacts with other transcription factors/co-factors and mediates transcription of TGF-β target genes⁵. Notably, an antagonist of the TGF-β signaling, SMAD7 (also named inhibitory SMAD, I-SMAD), regulated at the transcriptional level by the R-SMADs/co-SMAD heterotrimeric complex, interacts with TGFBR1 and recruits SMURF2 or NEDD4L to promote poly-ubiquitination and proteasomal degradation of TGFBR1, eventually forming a negative feedback loop to maintain the physiological activity of TGF-β signaling^6,7. Dysregulation of SMAD7 expression by a post-transcriptional regulation leads to aberrant activation of TGF-β signaling pathway, and such a scenario has been found in the malignant progression of multiple cancer types^8,9,10,11,12. Interestingly, the mRNA level of SMAD7 was found markedly lower in LAD samples than that in paired adjacent normal tissues, and lower SMAD7 level correlated with poor prognosis of LAD patients^13,14,15. In such a context, therefore, better understanding of the underlying mechanism of restraining SMAD7 transcription in LAD may implicate promising therapeutic targets or prognostic biomarkers for metastatic LAD.

DNA methylation represents as a stable and inheritable epigenetic modification in eukaryotic cells, which mainly represses the process of transcription through disrupting the interaction between DNA elements and transcription factors¹⁶. The landscape of DNA methylation in cancers is considerably distorted, generally exhibiting hypomethylation in repeat-rich intergenic regions, resulting in genome instability, and hypermethylation in a subset of CpG-dense gene-associated regions, leading to transcriptional suppression of tumor suppressor genes¹⁷. Notably, DNMT3B, a de novo DNA methyltransferase, has been considered as an oncogene in cancer initiation and progression^18,19. Numerous clinical studies have illustrated a great potential of DNA methylation as prognostic biomarkers and therapeutic targets for human cancers^20,21,22. Nevertheless, the regulatory mechanism responsible for the role of DNMT3B in cancer malignancies remains largely unknown, and its diagnostic or prognostic effectiveness reflected in a large proportion of the registered clinical trials remains still suboptimal, suggesting that the mechanisms underlying de novo DNA methylation and the clinical significance of DNA methylation need to be further investigated and better understood.

In this current study, we find that a zinc finger protein PHF14 recruits DNMT3B to the SMAD7 gene locus, resulting in SMAD7 DNA methylation, transcriptional suppression and TGF-β signaling hyperactivition. In addition, we further demonstrate that PHF14 promotes LAD progression and metastasis through enhancing TGF-β signaling and illustrate that not only the SMAD7 DNA methylation level in LAD tissues, but also that in circulating tumor DNA (ctDNA) of liquid biopsies correlate with poor prognosis of LAD patients.

Results

SMAD7 expression is suppressed in LADs by DNA methylation

To understand the role of TGF-β signaling in lung adenocarcinoma progression, we initially found that the activity of TGF-β signaling, which is defined as the average of TGFB1 and SERPINE1 expression, significantly correlated with poor prognosis of LAD patients in the TCGA LUAD cohort (Supplementary Fig. S1a). Specifically, Kaplan-Meier analysis showed that LAD patients bearing high TGF-β activity in their lung tumors displayed a significantly shorter median survival time of 42.31 months and lower 5-year survival rate of 37.322%, in contrast to the 53.29-month median survival and 42.910% 5-year survival rate for those with low TGF-β activity (Supplementary Fig. S1a).

The antagonist of TGF-β signaling, SMAD7, which is involved in the negative feed-back loop to counteract against TGF-β hyperactivation, is a direct transcriptional target of TGF-β signaling cascade (Supplementary Fig. S1b–d). The tumor suppressive role of SMAD7 was further validated in A549 and HCC827 LAD cell lines (Supplementary Fig. S1e–h). Of note, the expression of SMAD7 was markedly suppressed in LAD as compared to the normal tissues (Fig. 1a, b), and high SMAD7 expression level was associated with favorite prognosis of the LAD patients in the TCGA dataset (> 60 months median survival time and 52.855% 5-year survival rate in high SMAD7 expression subgroup vs 45.3 months median survival time and 28.337% 5-year survival rate in low SMAD7 expression subgroup, P = 0.0057) (Fig. 1c). Interestingly, SMAD7 expression level was barely altered between LADs with higher TGF-β activity and LADs with lower TGF-β activity (Supplementary Fig. S1i), suggesting that despite the TGF-β signaling cascade mediated transcriptional activation of SMAD7, the transcriptional status of SMAD7 might also be regulated by currently unidentified factors in LAD.

**Fig. 1: DNMT3B-catalyzed *SMAD7* DNA methylation correlates with poor prognosis of LAD patients.**

DNA methylation serves as an important epigenetic modification, which impedes gene transcription through altering chromosomal structure, DNA conformation and the interaction between regulatory DNA elements and protein factors. When the Shiny Methylation Analysis Resource Tool program was employed to identify the correlation of SMAD7 DNA methylation level with SMAD7 mRNA level and with the prognosis of LAD patients in the TCGA LUAD cohort²³, among the 26 methylation probes distributed on SMAD7 gene locus in the Illumina Infinium Human Methylation 450 K BeadChip, a panel of 6 probes was used as a probe set in the following analysis process, which detects sequences located on the 3′ UTR and CDS region of SMAD7 gene, and methylation values of the 6 CpG sites were found to inversely correlate with SMAD7 expression individually or aggregately in the TCGA LUAD dataset (Fig. 1d; Supplementary Figs. S1j, S2a and Table S1). Moreover, we found that the methylation level of SMAD7 was elevated in LAD samples as compared to their paired adjacent normal controls (Fig. 1e). Analogously, the DNA methylation level of SMAD7 was markedly increased in LAD tissues as compared to the normal lung tissues and negatively correlated with SMAD7 mRNA levels in our collected tumor tissue cohort (Fig. 1f, g). In addition, high methylation levels of SMAD7 were associated with poor prognosis (overall survival and recurrence-free survival) of LAD patients in TCGA LUAD dataset (Fig. 1h, i; Supplementary Fig. S2b). These data suggested that suppression of SMAD7 expression can be induced by DNA hypermethylation, and that importantly, SMAD7 DNA methylation levels revealed by the 6 probes predict unfavorable prognosis of LAD patients.

In the light that two de novo DNA methyltransferases DNMT3A and DNMT3B play distinct roles in lung cancer, as DNMT3B was implicated in promoting lung cancer by silencing tumor-suppressor genes, whereas DNMT3A suppressed KRAS^G12D-mediated lung cancer progression^24,25,26, we further investigated whether the oncogenic DNMT3B was involved in catalyzing de novo DNA hypermethylation of SMAD7 gene. When DNMT3A or DNMT3B was silenced or ectopically overexpressed in A549 and HCC827 LAD cells with TGF-β treatment, only silencing DNMT3B significantly promoted SMAD7 expression and diminished the DNA methylation of SMAD7, which could be partially reversed by additional ectopic overexpression of DNMT3B (Fig. 1j–l; Supplementary Fig. S2c–f). Whereas, in cells without TGF-β treatment, silencing DNMT3B markedly reduced SMAD7 DNA methylation, while the mRNA level of SMAD7 was slightly elevated (Fig. 1j–l; Supplementary Fig. S2c–f). In contrast, silencing or further overexpressing DNMT3A barely affected SMAD7 DNA methylation and gene expression in cells with or without TGF-β treatment (Fig. 1j–l; Supplementary Fig. S2c–f). Interestingly, abundant DNMT3A or DNMT3B did not enhance SMAD7 DNA methylation and had no effect on SMAD7 transcription in cells with TGF-β treatment (Fig. 1m–o; Supplementary Fig. S2g), together suggesting that DNMT3B is the one de novo DNA methyltransferase that is required but seems to be insufficient for SMAD7 DNA methylation and transcriptional suppression; therefore the detailed mechanisms for abrogation of the TGF-β-mediated SMAD7 transcription in LAD, which may through facilitating the function of DNMT3B, still need to be further understood.

PHF14 is a binding partner of DNMT3B

To gain an insight into the mechanism of DNMT3B-mediated abnormal DNA methylation on SMAD7 locus in LAD, we further examined whether previously unidentified factor might regulate DNMT3B-mediated SMAD7 DNA methylation and transcriptional suppression in LAD. The immunoprecipitation-mass spectrometry (IP-MS) assay was employed to screen uncharted binding partners of DNMT3B in A549 cell. Among the top-ranked putative binding proteins of DNMT3B, a plant homeodomain (PHD) finger-containing protein PHF14, which was recognized as an epigenetic suppressor, was identified to interact with DNMT3B (Supplementary Table S2), and the interaction was further validated by endogenous and exogenous co-IPs using LAD cells and 293FT cell (Fig. 2a–c; Supplementary Fig. S3a). Notably, the interaction between DNMT3B and PHF14 was moderately enhanced with TGF-β treatment (Supplementary Fig. S3a). Moreover, overexpressing PHF14 in LAD cells enhanced TGF-β-induced SMAD binding element (SBE)-luciferase activity with restrained SMAD7 expression by facilitating SMAD7 DNA methylation, while in cells without TGF-β treatment, the SMAD7 DNA methylation level was moderately enhanced and the SMAD7 mRNA level and TGF-β signaling activity were not markedly affected by ectopically overexpressed PHF14 (Fig. 2d–g; Supplementary Fig. S3b). In contrast, knocking out PHF14 in LAD cells impaired the TGF-β-induced SBE-luciferase activity by elevating SMAD7 expression level with depleted SMAD7 DNA methylation (Fig. 2d, h–j; Supplementary Fig. S3c). These results indicated that PHF14 is a newly identified binding partner of DNMT3B, which may promote TGF-β signaling activation through facilitating DNMT3B-mediated SMAD7 DNA hypermethylation and transcriptional suppression in LAD.

**Fig. 2: PHF14 is identified as a novel binding partner of DNMT3B.**

PHF14 facilities DNMT3B-mediated DNA methylation of SMAD7

Next, we further investigated the effect of PHF14 and DNMT3B interaction on DNA methylation of SMAD7. Immunofluorescent staining revealed that either endogenous PHF14 or ectopically overexpressed Flag-tagged PHF14 were nuclear localized protein (Fig. 3a), suggesting a possibility that PHF14 may directly regulate DNMT3B-catalyzed SMAD7 DNA methylation. Notably, chromatin immunoprecipitation-quantitative polymerase chain reaction (ChIP-qPCR) assays showed that binding of DNMT3B to SMAD7 DNA (regions of the 6 probes) could be abrogated by TGF-β treatment; however, the interaction was enhanced in different levels by PHF14 overexpression in cells with or without TGF-β treatment, but greatly diminished by knocking out PHF14 in LAD cells (Fig. 3b; Supplementary Fig. S4a, b). Simultaneously, the DNA methylation level of SMAD7 was also elevated in PHF14-overexpressing cells treated with TGF-β, but could be repressed by loss of PHF14 expression or silencing DNMT3B (Figs. 2f, i, 3c; Supplementary Figs. S3b, c, S4c). Moreover, the binding of PHF14 on SMAD7 DNA was barely altered by silencing DNMT3B, whereas the PHF14 overexpression-enhanced DNA methylation on SMAD7 DNA could be reversed when DNMT3B expression was silenced (Fig. 3c, d; Supplementary Fig. S4c, d). Meanwhile, TGF-β treatment-induced SMAD7 elevation was suppressed by PHF14 overexpression, and silencing DNMT3B could abrogate the effect of PHF14 overexpression in LAD cells (Fig. 3e, f). In addition, treatment with 5-Aza-2-deoxycytidine (Decitabine), an inhibitor of DNA methyltransferases, showed a similar effect to that caused by DNMT3B depletion on SMAD7 DNA methylation and expression level in LAD cells (Fig. 3g–i; Supplementary Fig. S4e), together suggesting that PHF14 plays a crucial role in DNMT3B recruitment to SMAD7 DNA, and is required for SMAD7 DNA methylation and transcriptional suppression.

PHF14 serves as a CG-rich motif reader

To further understand the molecular basis upon which PHF14 recruits DNMT3B to SMAD7 DNA, various Flag-tagged truncated proteins of PHF14 were constructed, as shown in the schematic diagram of the PHF14 protein domains (Fig. 4a). Subsequently, co-IP assays revealed that the ePHD and Middle domain of PHF14 interacted with DNMT3B, and interestingly, the interaction between the ePHD domain of PHF14 and DNMT3B was drastically abrogated by DNase I treatment of the whole cell lysate, suggesting that DNMT3B directly binds the Middle domain of PHF14, whereas the interaction between the ePHD domain of PHF14 and DNMT3B might rely on their recognition of and binding with chromosome (Fig. 4b, c; Supplementary Fig. S5a).

In light of the previous finding that the zinc finger PHD domains can serve as a reader of histones or DNAs, purified ePHD domain of or full-length recombinant human PHF14 protein was incubated with the MODified Histone Peptide Array, and unexpectedly, no specific histone peptide was found to interact with either the PHF14 ePHD domain or its full-length protein (Supplementary Fig. S5b). We thus further examined whether the ePHD domain of PHF14 interacts with DNA. To this end, HDOCK server program was employed to predict interaction between the PHF14 ePHD domain and various DNA probes²⁷, and our results showed that an α-helix region of the PHF14 ePHD domain (429–435aa) was responsible for mediating its interaction with, and stretching into the major groove of DNA helix, which has been recognized as a docking platform for protein recognition (Fig. 4d). Moreover, the docking score between unmethylated CG-rich oligonucleotide and the PHF14 ePHD domain was lower than that between methylated CG-rich oligonucleotide or AT-rich oligonucleotide with the PHF14 ePHD domain, indicating a more intense putative binding between the ePHD domain of PHF14 with unmethylated CG-rich DNA element (Supplementary Table S3).

Next, electrophoretic mobility shift assay (EMSA) was performed and showed that purified ePHD domain of PHF14 increased the shifts of the biotinylated CG-rich unmethylated oligonucleotide probe, and that further addition of corresponding unmodified probes attenuated the shifted band of the biotin-labelled probes in a dose-dependent competitive manner, whereas the interaction between ePHD and methylated CG-rich oligonucleotide probe or AT-rich oligonucleotide probe was substantially weak (Fig. 4e, f; Supplementary Fig. S5c, d). Furthermore, an analogous result was obtained from the incubation of the purified full-length PHF14 protein with above mentioned three types of oligonucleotide probes (Fig. 4f; Supplementary Fig. S5c, d). In addition, the HDOCK server-predicted interaction intensity suggested that amino acid mutations of the PHF14 ePHD domain in its α-helix DNA-binding region (E430AK435A) abrogated the binding between the PHF14 ePHD domain and CG-rich DNA element (Supplementary Table S3). Consistently, transduction of PHF14 ePHD- and full-length PHF14-expressing plasmids, respectively, containing E430A and K435A mutations also revealed that the interaction between PHF14 ePHD or full-length PHF14 with methylated CG-rich probe could be drastically vanished by the α-helix mutations (Fig. 4g). Moreover, the binding constants of PHF14 ePHD domain with CG-rich or AT-rich oligonucleotide probes, respectively, have been measured by the surface plasmon resonance (SPR) assay, which illustrated that the equilibrium dissociation constant (K_D) of unmethylated CG-rich probe with ePHD domain of PHF14 was 3.60 × 10⁻⁸ M; in contrast, the K_D value of AT-rich probe with PHF14 ePHD domain or CG-rich probe with mutated ePHD domain was 2.30 × 10⁻⁷ M and 2.75 × 10⁻⁷ M (Fig. 4h; Supplementary Fig. S5e), suggesting that the ePHD domain of PHF14 specifically binds with CG-rich DNA motif in vivo.

Furthermore, the binding of PHF14 on SMAD7 gene and the recruitment of DNMT3B on selected regions by PHF14 were markedly diminished by E430AK435A mutations of PHF14, as compared with those by WT PHF14 (Fig. 4i, j; Supplementary Fig. S5f, g). Meanwhile, DNA methylation level and the expression level of SMAD7 were not altered by the E430AK435A PHF14 mutant as compared to the vector-control in cells with TGF-β treatment, and the pro-TGF-β activation effect of PHF14 was largely reversed by E430AK435A mutations (Fig. 4k–n; Supplementary Fig. S5h). These results strongly suggest that the ePHD domain of PHF14 is responsible for the binding of PHF14 with CG-rich motifs and unveils a heterotrimer complex constituted by DNMT3B, PHF14 and DNA, which supports the notion that PHF14 facilitates the DNMT3B-catalyzed de novo DNA methylation on SMAD7 locus.

PHF14 promotes TGF-β-induced LAD progression and metastasis

To determine the biological function of PHF14 in TGF-β-mediated LAD malignancies, the effects of PHF14 on in vitro cultured LAD cells were tested. As shown in Fig. 5a–f, our results showed that while TGF-β greatly enhanced colony formation, invasion and migration abilities of LAD cells, deletion of PHF14 in A549 and PC9 cells markedly impaired the TGF-β1-promoted malignant phenotype of the LAD cells as compared to vector controls (Fig. 5a–f). By contrast, overexpressing PHF14 in Calu3 and HCC827 LAD cells significantly enhanced the cell colony formation, invasion and migration abilities of the tested cells treated with TGF-β1, whereas silencing DNMT3B or overexpressing SMAD7 in PHF14-overexpressed cells potently reversed the phenotype synergistically enhanced by TGF-β1 and PHF14 co-expression (Fig. 5g–l), suggesting that the tumor promoting role of PHF14 largely depends on DNMT3B and SMAD7.

**Fig. 5: PHF14 promotes TGF-β-induced colony formation, invasion and migration of LAD cells in vitro.**

In consistence with the above observed effects of PHF14 in vitro, intracardial inoculation of LAD cells labeled with luciferase showed that loss of PHF14 drastically abrogated the abilities of TGF-β1-overexpressing A549 and PC9 cells to form distant metastases, as evidenced by weaker bioluminescent signals in mice with PHF14 knockout cells than that of mice with vector-control cells (Fig. 6a, b). Indeed, micro-computed tomography (micro-CT) imaging and histological staining revealed markedly diminished TGF-β-driven cancerous lesions and tissue invasion/disruption in the lower-limb bones of nude mice intracardially injected with TGF-β1-overexpressing and PHF14 knocked out LAD cells, whereas TGF-β1-overexpressing vector-control cells displayed a robust ability to spread to bone tissues (Fig. 6c). Consequently, nude mice intracardially inoculated with PHF14-deleted LAD cells exhibited significantly prolonged metastasis-free and overall survival as compared with those inoculated with vector-control cells (Fig. 6d). On the contrary, PHF14 markedly promoted TGF-β1-overexpressing Calu3 and HCC827 LAD cells to form systemic metastases, with enhanced multi-organ bioluminescent signals, prominent cancerous lesions, severe tissue invasion/disruption in the lower-limb bones, and shortened metastasis-free and overall survival of nude mice intracardially injected with TGF-β1 and PHF14 co-overexpressing LAD cells as compared to those inoculated with the vector-control cells (Fig. 6e–h). Notably, silencing DNMT3B or overexpressing SMAD7 robustly abrogated the ability of TGF-β1 and PHF14 co-overexpressing LAD cells to form systemic metastases and prolonged metastasis-free and overall survival of nude mice intracardially injected with these LAD cells (Fig. 6e–h). Taken together, our data clearly demonstrated a crucial and potent pro-metastatic role of PHF14 in TGF-β-induced LAD progression and metastasis in a manner dependent on the PHF14-DNMT3B-SMAD7 axis.

**Fig. 6: PHF14 promotes TGF-β-induced systemic metastases of LAD cells in vivo.**

Potential significance of SMAD7 and PHF14 as promising diagnostic and prognostic biomarkers for LAD progression

We next evaluate whether the expression of PHF14 and SMAD7 can indicate the state of clinical progression of LAD. In this context, we found that PHF14 was significantly upregulated in LAD tissues collected in the TCGA LUAD datasets, as compared with the normal tissue specimens, which include 59 pairs of lung cancer/normal tissues and 517 cases of lung cancer vs 59 cases of unpaired normal lung tissue (Fig. 7a, b). PHF14 upregulation can also be observed in our LAD tissue cohort containing 8 cases of freshly collected LAD specimens paired with adjacent non-cancerous tissues and in a panel of LAD cell lines as compared to normal lung epithelial cells (NLECs) and an immortalized normal lung epithelial cell line Beas-2B (Fig. 7c–f). Moreover, using the TCGA LAD datasets, Kaplan-Meier analysis showed that LAD patients with high PHF14 expression in their lung tumors displayed a significantly shorter median survival time of 40.57 months and lower 5-year survival rate of 33.783%, in contrast to the 56.67-month median survival time and 44.711% 5-year survival rate for those with low PHF14 expression; in parallel, high PHF14 levels correlates with shorter recurrence-free survival (26.64 months median survival time and 29.481% 5-year survival rate in high PHF14 expression subgroup vs 47.63 months median survival time and 44.936% 5-year survival rate in low PHF14 expression subgroup) (Fig. 7g, h). Importantly, the correlations between PHF14 and SMAD7 levels in DNMT3B-high and -low scenarios, respectively, were further examined in our collected patient tissue specimen cohort. Our results showed that PHF14 expression was inversely associated with the level of SMAD7 in tumor specimens with high DNMT3B expression, and by contrast, the correlation between the expression levels of PHF14 and SMAD7 was not significant in those expressing low DNMT3B (Fig. 7i, j). In addition, high PHF14 level was found to positively correlate with TGF-β-mediated transcription in TCGA LUAD dataset (Fig. 7k), further supporting a close clinical relevance of the PHF14 and SMAD7 expression in LAD progression.

**Fig. 7: SMAD7 and PHF14 are diagnostic and prognostic biomarkers for LAD progression.**

It is of particular note that when we examined SMAD7 DNA methylation by detecting ctDNA obtained through liquid biopsy from LAD patients in our collected cohorts (Supplementary Table S4). We found that the SMAD7methylation levels of ctDNA in patients bearing metastatic cancer (n = 27) were significantly higher than those in patients only with primary LAD tumor (n = 19), and the ctDNA methylation levels in these two groups were markedly greater than those in the normal controls (n = 12) (Fig. 7l). In addition, SMAD7 ctDNA methylation levels were significantly reduced in patients after LAD tumors were resected as compared with those in the same group of patients before tumor resection (n = 14) (Fig. 7m). Furthermore, there is a positive correlation between the SMAD7 methylation levels of ctDNA and poor prognosis of LAD patients. Patients with higher SMAD7 ctDNA methylation levels exhibited substantially shorter survival time as compared with those with lower SMAD7 methylation level of ctDNA (Fig. 7n). Collectively, the above analyses suggest that the SMAD7 DNA methylation level in ctDNA correlates well with tumor burden and may have the potential to be applied to indicate tumor progression and prognosis of the patients.

Discussion

It has been long noted that TGF-β signaling can be both tumor-suppressive at the initiation phase of tumorigenesis and pro-metastatic in later stages of cancer progression, known as the dual role of TGF-β during cancer development²⁸. Of particular note, therefore, monotherapy targeting TGF-β ligand–receptor interaction usually reveals suboptimal and highly variable therapeutic effects, suggesting that the biological functions of TGF-β signaling and the outcome of TGF-β targeting cancer therapies may be largely context-dependent²⁹. Also notably, accumulated evidence has indicated that TGF-β signaling potently drives SMAD-specific transcriptional program, and that orchestration of the biologically distinct roles of TGF-β signaling in cancer cells may in part rely on molecules that interact with SMADs and thereby modulate the functions of SMADs, which consequently alters the expression spectra of TGF-β target genes^{30,31,32,33,34,35}. Nevertheless, how the functions of various SMADs are changed by their interactive molecules and consequently redirect the biological effects of TGF-β signaling remains incompletely understood.

SMAD7 functions as a critical antagonist of TGF-β signaling to prevent excessive activation of the pathway through promoting poly-ubiquitination-mediated proteasomal degradation of TGF-β receptor TGFBR1; in addition, SMAD7 also has been defined as a direct downstream transcriptional target of TGF-β signaling cascade, forming a negative feedback loop of TGF-β signaling^36,37. Mounting evidence shows that upregulated SMAD7 mRNA level, prolonged SMAD7 mRNA or protein stabilities, and dysregulated biological functions of SMAD7 in various types of cancers can be attributable to multiple transcription factors/co-factors^38,39, non-coding RNAs^8,9,11,40,41 and molecules involved in protein post-translational modifications^{10,12,13,42,43,44}. In such a context, our present study identifies a PHF14/DNMT3B-mediated epigenetic mechanism via which SMAD7 transcription is suppressed in LAD. Consistently, previous studies have found that the expression of DNMT3B and PHF14 can be elevated following TGF-β stimulation^45,46. Moreover, our data unveil that the transcriptional status of SMAD7 in LAD is jointly regulated by TGF-β signaling-mediated transcriptional activation and PHF14/DNMT3B-mediated transcriptional suppression. Only DNMT3B seems insufficient for de novo DNA methylation of SMAD7, and thus our results further elucidate an important role of PHF14 in this regulatory mechanism, underscoring that the upregulation of PHF14 expression is essential for aberrant activation of TGF-β signaling in our model. Taken together, our findings provide new insights into the molecular basis for TGF-β-driven cancer metastasis through spontaneous disruption of the negative feedback loop of TGF-β regulation. Whether such a molecular mechanism also exists and functions in other cancer types need to be further investigated.

As a component in the newly identified PHF14/DNMT3B complex, DNMT3B is known as one of the two de novo DNA methyltransferases responsible for generating 5-methylcytosine by transferring the methyl-group from S-adenosyl methionine to the fifth carbon of unmethylated cytosine⁴⁷. Previous studies have found that DNMT3B could be upregulated by TGF-β signaling activation⁴⁵, and that DNMT3B overexpression accelerates malignant transformation, promotes cancer progression and induces drug resistance in liver cancer, colon cancer and breast cancer, through enhancing DNA hypermethylation-mediated epigenetic suppression of tumor suppressor genes^{18,19,48,49,50}. Interestingly, DNMT3A, a paralogue of DNMT3B, predominantly plays a tumor-suppressive role in cancers⁵¹. DNMT3A mutations are present in approximately 10%–20% hematological malignancies, with the R882 site being a unique mutation hotspot, conferring dominant-negative inhibition against the wild-type DNMT3A enzyme and poor prognosis⁵². Moreover, allelic loss of DNMT3A gene or loss of DNMT3A expression is frequently found in multiple human somatic tumor types; and depletion of Dnmt3a gene potently promotes progression of lung cancer with Kras^G12D mutant in a mouse model²⁶. The functional discrepancies between DNMT3A and DNMT3B in stem cells have been revealed by several genome-wide studies and are thought to be attributed to distinctive preference of their genomic occupancies and specificities towards their target genes^53,54. However, the mechanistic basis for the fundamental roles of DNMT3A and DNMT3B in tumorigeneses and cancer progression remains largely unknown, and it therefore becomes important to understand the underlying mechanism determining their occupancies on chromosomes and their distinct functions.

The amino acid sequence homology between DNMT3A and DNMT3B is approximately 70%, which is mainly contributed by their PWWP domains, the ADDz domains and the site-specific DNA-cytosine methylase domains. The rest residues, therefore, might contribute to DNMT3A’s and DNMT3B’s specific interactions with their various regulatory factors. Notably, decades of research on DNMTs have shown that the ADD and PWWP domains of DNMT3B are responsible for the recruitment of DNMT3B to histone H3 N-terminal tail^{54,55,56,57,58}; and DNMT3B also interacts with histone methyltransferases SETDB1, EZH2 or transcription factors E2F6, PU.1 and NR6A1 to occupy specific regions on chromatins and silence the transcription of target genes^{59,60,61,62,63}, suggesting that the recruitment of DNMT3B on chromosomes is sophisticatedly regulated by its binding partners in a context-dependent manner. In this current study, we found that DNMT3B is required to promote SMAD7 DNA methylation and transcriptional suppression. In this context, zinc finger protein PHF14 is identified as a binding partner of DNMT3B, which recruits DNMT3B to the CpG sites distributed in the SMAD7 gene locus, leading to DNA methylation-mediated transcriptional suppression of SMAD7 and consequent dysregulated activation of TGF-β signaling. These results further suggest that the process of DNMT3B-mediated de novo DNA methylation on specific gene locus may involve uncharted epigenetic factors needed to be further identified and understood.

Our newly identified binding partner of DNMT3B, PHF14, has been previously shown to promote tumorigenesis and cancer progression by regulating cell mitosis, epithelial-mesenchymal transition, and Wnt and Akt signaling in multiple cancer types^{64,65,66,67,68}. Mechanistically, PHF14 epigenetically suppresses the transcription of cell cycle inhibitors in cancers, including p14^ARF, p15^INK4b and p16^INK4a genes⁶⁹. Here, we find that PHF14 reads the CG-rich elements of DNA through its ePHD domain, and recruits DNMT3B to the SMAD7 gene locus via its Middle domain, elucidating that PHF14 functions as a molecular scaffold linking DNMT3B and the CG-rich regions on SMAD7 gene locus to restrain the transcription of SMAD7 by enhancing its DNA hypermethylation. Recently, the PHD1-ZnK-PHD2 cassette of zebrafish PHF14 (corresponds to the PHD1-ePHD domain of human PHF14) has been found to be a reader for unmodified N-tail of histone H3⁷⁰, implicating that the function of PHF14 may be complex and largely context-dependent, and needs to be further investigated.

In summary, our findings provide an insight into the mechanistic basis for aberrant activation of TGF-β signaling and its consequent induction of LAD metastasis, and unveil a novel PHF14/DNMT3B complex involved in the epigenetic suppression of SMAD7, indicating that PHF14 plays a pivotal role in determining the biological effects of TGF-β. These results also suggest a notable significance of PHF14 and SMAD7 in the diagnosis and prognostic evaluation LAD. Furthermore, it is of great interest to note that our current study also revealed the feasibility of employing liquid biopsy, a promising noninvasive diagnostic approach in clinical oncology^20,71,72, to develop practically applicable tests for SMAD7 ctDNA methylation as a clinically useful marker for LAD progression and prognosis. Moreover, it warrants future efforts to further evaluate whether inhibiting DNA methylation and abrogating the interaction among PHF14, DNMT3B and DNA can be a potential strategy for targeting TGF-β-driven LAD progression and metastasis.

Materials and Methods

Cell culture

LAD cell lines, including A549, PC9, Calu-3, HCC827, H1975, 95D and non-cancerous HEK293FT cells were purchased from the American Type Culture Collection (ATCC; Manassas, VA, USA) and maintained in Dulbecco’s modified Eagle’s medium (DMEM for LAD cell lines and HEK293FT cells; GIBCO, Carlsbad, CA, USA), supplemented with 10% fetal bovine serum (FBS; Corning, Tewksbury, MA, USA) and 1% penicillin/streptomycin (penicillin 100 U/mL and streptomycin 10 μg/mL) (GIBCO). The Beas-2B immortalized human bronchial epithelial cell line obtained from ATCC was cultured in Bronchial Epithelial Cell Growth Medium (BEGM) as instructed by the supplier (Lonza, Walkersville, MD, USA). All cell lines were authenticated by short tandem repeat DNA profiling at the Forensic Medicine Department of Sun Yat-Sen University Zhongshan School of Medicine (Guangzhou, China), and were verified to be mycoplasma-free.

Plasmids, virus production and transfection

FLAG tagged-PHF14 (NP_055475), SMAD7 (NP_005895) and truncated mature human TGFB1 (NP_000651, aa. Ala270–Ser390) were separately cloned into the pSin-EF2-puro or pSin-EF2-neo retroviral vector (Addgene, Cambridge, MA, USA). FLAG-tagged or His-tagged full-length or truncated PHF14, HA-DNMT3A and HA-DNMT3B were separately cloned into the pCDNA3.4 TOPO vector (Invitrogen, Carlsbad, CA, USA). For depletion of PHF14, human gRNA sequences (PHF14-gRNA1: GCTGACTTGTTCCTCTGTAG; PHF14-gRNA2: GAAGTGGTCGGTTTCGTCGA) were cloned into the pLX330 vector (Genewiz, South Plainfield, NJ, USA). For silencing SMAD7, DNMT3A or DNMT3B, human shRNA sequences (SMAD7-shRNA1: GCTTTCAGATTCCCAACTTCT; SMAD7-shRNA2: GGTTTCTCCATCAAGGCTTTC; DNMT3A-shRNA: GCGTCACACAGAAGCATATCC; DNMT3B-shRNA: GCCCATTTGACTTGGTGATTG) against each of them were respectively cloned into the pSuper-retro-neo or pSuper-retro-puro vector (OligoEngine, Seattle, WA, USA). 3× SBE motifs (5′-GTCTAGAC-3′) were cloned upstream to the luciferase gene in a pGL3-Enhancer vector (Promega, Madison, WI, USA). For construction of cell lines knocked out for PHF14, pLX330-PHF14-gRNA plasmids were transfected by using Lipofectamine 3000 reagent (Invitrogen). The cells were subsequently dissociated with trypsin and re-suspended in DMEM containing 2% FBS. Then the cells were transferred into the 5 mL polypropylene round-bottom tubes (CORNING, NY, USA) on ice before being seeded into the 96-well plates by using flow cytometer (BD Influx) through GFP. For construction of stable cell lines overexpressing TGF-β1, PHF14 or SMAD7, and stable cell lines knocked down for SMAD7, DNMT3A or DNMT3B, plasmid pSin-TGFB1, pSin-PHF14 or pSin-SMAD7, respectively, was co-transfected with the pMD2G and psPAX2 packaging plasmids; and plasmid carrying specific shRNA for SMAD7, DNMT3A or DNMT3B, respectively, was co-transfected with the PIK packaging plasmid into 293FT cells using a standard calcium phosphate transfection method. Subsequently, cell culture supernatants of 293FT cells were collected to infect LAD cells for 24 h in the presence of polybrene (8 μg/mL), followed by selection with puromycin (1 μg/mL) or G418 (200 μg/mL) for 10–14 days. Transfection of plasmids or RNA oligonucleotides were performed using Lipofectamine 3000 reagent (Invitrogen) for luciferase reporter assays and molecular assays.

Antibodies

Primary antibodies used for WB analyses were: Anti-SMAD3 (phospho S423 + S425) antibody (Abcam, ab52903, 1:2000), anti-SMAD3 antibody (Abcam, ab28379, 1:800), anti-MADH7/SMAD7 antibody (Abcam, ab216428, 1:1000), anti-PHF14 antibody (Proteintech, 24787-1-AP, 1:800), anti-DNMT3A antibody (Cell Signaling, 3598 S, 1:800), anti-DNMT3B antibody (Abcam, ab13604, 1:500), anti-FLAG antibody (Sigma-Aldrich, F1804, 1:2000), anti-HA antibody (Proteintech, 51064-2-AP, 1:2000), and Normal Mouse IgG for immunoprecipitation (Sigma-Aldrich, 12–371). Blotted membranes were stripped and re-blotted with anti-GAPDH antibody (Proteintech, 60004-1-Ig, 1:10000) as a loading control.

RNA extraction and quantitative reverse transcription-PCR (qRT-PCR)

Total RNA from cultured cells or surgically resected fresh LAD tissues was isolated using TRIzol (Invitrogen) according to the manufacturer’s instruction. The first-strand cDNA was generated by reverse transcription using MMLV transcriptase (Promega) and random primers. Real-time qRT-PCR was performed on a CFX96 real-time PCR detection system (Bio-Rad, Richmond, CA, USA). Gene expression was assessed based on the threshold cycle (Ct), and relative expression levels were calculated as 2^{–[(Ct of mRNA) – (Ct of GAPDH)]} after normalization to GAPDH expression.

SBE-luciferase assay

The TGF-β signaling activity was measured by dual-luciferase reporter assay (Promega) using TGF-β signaling activity reporter, which was constructed by cloning the 3× SBE motifs (CAGACAGACAGA) into pGL3-Basic plasmid as a promoter of Firefly Luciferase gene, and the Renilla Luciferase reporter (TK plasmid) used as an internal control. Relative luciferase activity (Firefly Luciferase/Renilla Luciferase) for each treatment was calculated as the TGF-β signaling activity.

Immunoprecipitation and protein purification

Lysates were prepared from 3 × 10⁷ 293FT cells transfected with Flag-tagged full-length or truncated PHF14 or HA-tagged DNMT3B in an NP-40-containing lysis buffer supplemented with protease inhibitor cocktail (Roche), and then immunoprecipitated with FLAG or HA affinity agarose (Sigma-Aldrich) overnight at 4 °C. Beads containing affinity-bound proteins were washed six times with immunoprecipitation wash buffer (150 mM NaCl, 10 mM HEPES, pH 7.4, 0.1% NP-40), followed by elution with 0.1 M glycine (pH 3.0) twice. The eluted proteins were denatured and separated on SDS-polyacrylamide gels and stained with Coomassie blue; the indicated bands were subjected to MS analysis (Applied Protein Technology, Shanghai, China). Purification of recombinant His-tagged full-length or truncated PHF14 proteins was performed using BeaverBeads His-tag Protein Purification Kit (70501-K10, Beaver Biomedical Engineering Co., Suzhou, China).

Immunofluorescence assay

A549 and 293FT cells were grown on coverslips and transfected with the indicated plasmid. At 24 h after transfection, cells were washed once with PBS and fixed in 4% paraformaldehyde in PBS. Subsequently, cells were permeabilized with 0.2% Triton X-100 and treated for 30 min at room temperature with 10% BSA in PBS, followed by incubation with primary antibody for 1 h. Primary anti-PHF14 antibody (Proteintech, 24787-1-AP) was used. The secondary antibody was goat anti-rabbit IgG (H + L) conjugated with Alexa Fluor 594 (Thermo Fisher Scientific) in 1% BSA at room temperature for 20–90 min. The cells were rinsed 2–3 times with PBS (pH7.4) (5 min/wash) and then mounted with Gold Antifade Mountant with DAPI (Thermo Fisher Scientific). The coverslips were washed extensively and fixed on slides, and representative images were obtained with a LSM810 confocal microscope using ZEN 2012 software version 8.1 (Carl Zeiss, Oberkochen, Germany).

Bisulfite genomic sequencing

Genomic DNA from LAD cells, and clinical specimens’ ctDNA were bisulfite-modified with the Epitect Bisulfite Kit (Qiagen). Bisulfite-treated DNA was amplified with BSP primers designed in The Li Lab website (www.urogene.org) (BSP1 forward primer: 5′- AGTAAAGTAGTTTGAAGTGTGGTTTGTT -3′, reverse primer: 5′- CTATTAATACACAAAATATTCCCC -3′; BSP2 forward primer: 5′- GGGATATATTGGTTTTTTTTAAAGTAG -3′, reverse primer: 5′- AAACTTAATCCTTCTAACTAACTCC -3′; BSP3 forward primer: 5′- TTTGAAAATTTAATTTTGGGGAGTA -3′, reverse primer: 5′- TTTAACAATACTTAAAAAACTCAAAAC -3′; BSP4 forward primer: 5′- AAAATTTAGTTATAATAGAAAAATTTATTA -3′, reverse primer: 5′- AAATTAAAAATAAAACCTCTAAAAC -3′). PCR products were cloned using the pGEM-T Easy Vector System (Promega). Plasmids prepared from ampicillin-resistant bacteria clones were purified and sequenced.

Prediction of PHF14–DNA interaction

The structure of human PHF14 ePHD was generated using MOE software version 2014.09 (Chemical Computing Group, Montreal, Canada) by modifying non-conserved amino acids in zebrafish PHF14 PZP structure downloaded from the Protein Data Bank (accession code 7D86). In addition, the structures of 3 types of DNA probes were also generated by the MOE software. Subsequently, both the structures of PHF14 ePHD and DNA probes were uploaded to the HDOCK server (http://hdock.phys.hust.edu.cn/) to predict the interaction models between PHF14 ePHD and DNA probes. The predicted models and interaction intensities were further analyzed.

EMSA

EMSA was performed by using the LightShift Chemiluminescent EMSA kit purchased from Pierce Biotechnology (Rockford, IL, USA) according to the manufacturer’s standard protocol. Annealed dsDNA probes (CG-rich probe: sense, 5′- ACGGCTCGTCGCTACGGCTCGTCGCT -3′, antisense, 5′- AGCGACGAGCCGTAGCGACGAGCCGT -3′; AT-rich probe: sense, 5′- CATTAGATGATAGCATTAGATGATAG -3′, antisense, 5′- CTATCATCTAATGCTATCATCTAATG -3′, in which the positions of methyl-C are underlined) were incubated with diverse purified PHF14 proteins. Samples were then electrophoresed in a 0.5× TBE native polyacrylamide gel and the binding reactions were transferred to nylon membrane. Finally, the biotin-labeled DNAs were detected by chemiluminescence.

SPR kinetic analysis

SPR kinetic analysis was performed with a method suggested by the instruction manual using HBS-P running buffer (10 mM HEPES, pH 7.4, 150 mM NaCl, 0.05% (v/v) surfactant P20) at 25 °C and employing a series S sensor chip CM5 on the Biacore T100 SPR instrument (Cytiva). The recombinant wild-type or E430AK435A mutated FLAG-tagged PHF14 ePHD protein was diluted in the running buffer and then immobilized to a density of 600–770 response units (RU). The chemosynthetic probes at different concentrations were sequentially injected on the chip surface for 180 s at a flow rate of 20 μL/min, with the dissociation phase monitored for up to 600 s. Individual sensor grams were referenced against injection onto an empty flow cell at equivalent concentrations. Data were fitted to a 1:1 Langmuir model using the BIAevaluate 4.0.1 analysis software.

ChIP assay

ChIP assays were performed with a commercially available ChIP Assay Kit (Upstate Biotechnology, Lake Placid, NY, USA) by following the manufacturer’s instructions. In brief, cells (3 × 10⁶) cultured in 15-cm culture dishes were harvested for cross-linking and sheared by sonication. The resultant chromatin fractions were immunoprecipitated using anti-PHF14 (Proteintech, 24787-1-AP), anti-DNMT3B antibody (Abcam, ab13604), or negative control anti-IgG (Sigma-Aldrich) antibodies. After reversing the cross-links with NaCl and removing proteins with proteinase K, enriched DNA fragments were purified and isolated via phenol/chloroform extraction and ethanol precipitation. Real-time quantitative PCR of purified DNA fragments was performed with the indicated specific primers designed based on the putative PHF14 or DNMT3B binding sites in the SMAD7 gene locus (ChIP1 forward primer: 5′- GGCGGACTTGATGAAGAT -3′, reverse primer: 5′- CAATTCGGACAACAAGAGT -3′; ChIP2 forward primer: 5′- GACACACTGGCTTCTTCC -3′, reverse primer: 5′- CTTGGTCCTTCTGGCTGA -3′; ChIP3 forward primer: 5′- GCCAAGGAGCGAACTATT -3′, reverse primer: 5′- CAGTTGCTGAGACTATCGT -3′; ChIP4 forward primer: 5′- ACCAAACTTCCAACAGAGG -3′, reverse primer: 5′- GTGTGTAGGCAAAGGCTT -3′).

Colony formation assay

For colony formation assay, 1 × 10³ cells were plated in 6-well plates and cultured for 7 days, and colonies were stained with 1% crystal violet for 10 min after fixation with 10% formaldehyde for 5 min and counted.

Wound-healing assay

Cells were seeded in 6-well plates and grown to 90% confluence, followed by serum starvation for 12 h. A linear wound was created in the confluent monolayer using a pipette tip, and wounds were observed and photographed at various time points. Wound size was measured randomly at 3 sites perpendicular to the wound.

Animal experiments

Female BALB/c-nu mice (5–6 weeks of age, 18–20 g) were housed in specific pathogen-free facilities for animal studies. At least five nude mice per group were used to ensure the adequate power and each mouse with different weight was randomly allocated. For intracardiac injection, the indicated 5 × 10⁵ luciferase-expressing cells were resuspended in 0.1 mL PBS and inoculated into the left cardiac ventricle of nude mice. Metastases were monitored by bioluminescent imaging every 3 days. For bioluminescent imaging assay, 15 min prior to imaging, mice were injected intraperitoneally (i.p.) with 150 mg/kg luciferin. Following general anesthesia, images were taken and analyzed with Spectrum Living Image version 4.2 (Caliper Life Sciences, Waltham, MA, USA). At the experimental endpoint, nude mice were anesthetized and sacrificed, and lower limbs were resected, sectioned (5 μm in thickness) and histologically examined by H&E staining for the presence of metastatic lesions. All experimental procedures were approved by the Institutional Animal Care and Use Committee of Sun Yat-sen University.

Immunohistochemistry (IHC) assay

IHC assays using anti-MADH7/SMAD7 antibody (Abcam, ab216428), anti-PHF14 (Proteintech, 24787-1-AP) and anti-DNMT3B antibody (Abcam, ab13604) were separately conducted on paraffin-embedded specimens of LAD patients. The immunostaining intensities of indicated proteins was evaluated and scored by two independent observers, scoring both the proportions of positive staining tumor cells and the staining intensities. Scores representing the proportion of positively stained tumor cells was graded as: 0 (no positive tumor cells), 1 (< 10%), 2 (10%–50%) and 3 (> 50%). The staining intensity was determined as: 0 (no staining), 1 (weak staining = light yellow), 2 (moderate staining = yellow brown) and 3 (strong staining = brown). The staining index (SI) was calculated as staining intensity × percentage of positive tumor cells, resulting in scores as 0, 1, 2, 3, 4, 6 and 9. Cut-off values for high- and low-expression of protein of interest were chosen based on a measurement of heterogeneity using the log-rank test with respect to overall survival.

Clinical specimens and ethical approval

Two patient and control cohorts were employed in our current study. One cohort (i.e., the tumor and adjacent normal tissue cohort) includes LAD tissue specimens taken from 28 LAD patients, among which 8 adjacent-normal tissue specimens were also available (for Figs. 1f, g, 7c, e, i, j), and the second one was serum specimens derived from 12 healthy individuals and 46 LAD patients, respectively (for Fig. 7l–n). These two cohorts used in this study were obtained from and histopathologically diagnosed at Sun Yat-Sen Memorial Hospital. LAD tissues and paired adjacent non-cancerous lung specimens collected during surgery were frozen in liquid nitrogen and stored at –40 °C until further use. Adjacent non-cancerous specimens were obtained from a standard distance (3 cm) from the tumor margin in resected tissues of LAD patients. For the use of these clinical materials for research purposes, prior patients’ consents and approval from the Institutional Research Ethics Committee of Sun Yat-sen University were obtained. The study is compliant with all relevant ethical regulations involving human subjects.

TCGA dataset analysis

RNAseq data for PHF14 expression of 59 pairs of LAD tissues and paired adjacent non-cancerous lung tissues, and 576 cases of LAD tissues, as well as their corresponding clinical information, were mined from TCGA lung cancer datasets (https://cancergenome.nih.gov/).

Statistical analysis

All statistical analyses except sequencing data were performed using the PASW Statistics 18 version 18.0.0 (SPSS Inc., Chicago, IL, USA) software package and GraphPad Prism 8 version 8.3.0 (GraphPad Software, San Diego, CA, USA). Survival curves were analyzed by the Kaplan-Meier method and compared by the log-rank test. Comparisons between two groups were performed using Student’s t-test (two-tailed), while analyses comparing multiple treatments with a control group were performed using two-way ANOVA with post hoc Dunnett’s multiple comparisons test. All error bars represent means ± SD derived from three independent experiments. In all cases, P < 0.05 was considered to be statistically significant.

Study approval

All experimental procedures and use of LAD donors’ samples were approved by the SYSU Institutional Animal Care and Use Committee, and Research Ethics Committee. Donors provided prior written informed consent.

Data availability

All data generated or analyzed during this study are included in this article and supplementary information files. The TCGA Lung Adenocarcinoma (TCGA-LUAD) sequencing data used in this study are available in a public repository from the GDC Data Portal (https://portal.gdc.cancer.gov/).

References

Siegel, R. L., Miller, K. D., Fuchs, H. E. & Jemal, A. Cancer statistics, 2021. CA Cancer J. Clin. 71, 7–33 (2021).
Article PubMed Google Scholar
Travis, W. D. et al. The 2015 World Health Organization classification of lung tumors: impact of genetic, clinical and radiologic advances since the 2004 classification. J. Thorac. Oncol. 10, 1243–1260 (2015).
Article PubMed Google Scholar
Sung, H. et al. Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J. Clin. 71, 209–249 (2021).
Article PubMed Google Scholar
Batlle, E. & Massague, J. Transforming growth factor-beta signaling in immunity and cancer. Immunity 50, 924–940 (2019).
Article CAS PubMed PubMed Central Google Scholar
Massague, J. TGFbeta in cancer. Cell 134, 215–230 (2008).
Article CAS PubMed PubMed Central Google Scholar
Miyazawa, K. & Miyazono, K. Regulation of TGF-beta family signaling by inhibitory Smads. Cold Spring Harb. Perspect. Biol. 9, a022095 (2017).
Article PubMed PubMed Central Google Scholar
Hata, A. & Chen, Y. G. TGF-beta signaling from receptors to Smads. Cold Spring Harb. Perspect. Biol. 8, a022061 (2016).
Article PubMed PubMed Central Google Scholar
Zhao, S. et al. miR-4775 promotes colorectal cancer invasion and metastasis via the Smad7/TGFbeta-mediated epithelial to mesenchymal transition. Mol. Cancer 16, 12 (2017).
Article PubMed PubMed Central Google Scholar
Xia, H., Ooi, L. L. & Hui, K. M. MicroRNA-216a/217-induced epithelial-mesenchymal transition targets PTEN and SMAD7 to promote drug resistance and recurrence of liver cancer. Hepatology 58, 629–641 (2013).
Article CAS PubMed Google Scholar
Li, H. et al. HERC3-mediated SMAD7 ubiquitination degradation promotes autophagy-induced EMT and chemoresistance in glioblastoma. Clin. Cancer Res. 25, 3602–3616 (2019).
Article CAS PubMed Google Scholar
Huang, W. et al. Circular RNA cESRP1 sensitises small cell lung cancer cells to chemotherapy by sponging miR-93-5p to inhibit TGF-beta signalling. Cell Death Differ. 27, 1709–1727 (2020).
Article CAS PubMed Google Scholar
Zhang, Z. et al. Breast cancer metastasis suppressor OTUD1 deubiquitinates SMAD7. Nat. Commun. 8, 2116 (2017).
Article PubMed PubMed Central Google Scholar
Tong, L. et al. Proteasome-dependent degradation of Smad7 is critical for lung cancer metastasis. Cell Death Differ. 27, 1795–1806 (2020).
Article CAS PubMed Google Scholar
Tang, J., Li, X., Cheng, T. & Wu, J. miR-21-5p/SMAD7 axis promotes the progress of lung cancer. Thorac. Cancer 12, 2307–2313 (2021).
Article CAS PubMed PubMed Central Google Scholar
Dai, Z. T. et al. Integrated TCGA and GEO analysis showed that SMAD7 is an independent prognostic factor for lung adenocarcinoma. Medicine 99, e22861 (2020).
Article CAS PubMed PubMed Central Google Scholar
Jones, P. A. Functions of DNA methylation: islands, start sites, gene bodies and beyond. Nat. Rev. Genet. 13, 484–492 (2012).
Article CAS PubMed Google Scholar
Klutstein, M., Nejman, D., Greenfield, R. & Cedar, H. DNA methylation in cancer and aging. Cancer Res. 76, 3446–3450 (2016).
Article CAS PubMed Google Scholar
So, J. Y. et al. Induction of DNMT3B by PGE2 and IL6 at distant metastatic sites promotes epigenetic modification and breast cancer colonization. Cancer Res. 80, 2612–2627 (2020).
Article CAS PubMed PubMed Central Google Scholar
Lin, Z. et al. FOXC1 promotes HCC proliferation and metastasis by upregulating DNMT3B to induce DNA hypermethylation of CTH promoter. J. Exp. Clin. Cancer Res. 40, 50 (2021).
Article CAS PubMed PubMed Central Google Scholar
Luo, H., Wei, W., Ye, Z., Zheng, J. & Xu, R. H. Liquid biopsy of methylation biomarkers in cell-free DNA. Trends Mol. Med. 27, 482–500 (2021).
Article CAS PubMed Google Scholar
Koch, A. et al. Analysis of DNA methylation in cancer: location revisited. Nat. Rev. Clin. Oncol. 15, 459–466 (2018).
Article CAS PubMed Google Scholar
Yang, X. et al. Gene body methylation can alter gene expression and is a therapeutic target in cancer. Cancer Cell 26, 577–590 (2014).
Article CAS PubMed PubMed Central Google Scholar
Li, Y., Ge, D. & Lu, C. The SMART App: an interactive web application for comprehensive DNA methylation analysis and visualization. Epigenetics Chromatin 12, 71 (2019).
Article CAS PubMed PubMed Central Google Scholar
Soejima, K., Fang, W. & Rollins, B. J. DNA methyltransferase 3b contributes to oncogenic transformation induced by SV40T antigen and activated Ras. Oncogene 22, 4723–4733 (2003).
Article CAS PubMed Google Scholar
Lin, T. S. et al. An association of DNMT3b protein expression with P16INK4a promoter hypermethylation in non-smoking female lung cancer with human papillomavirus infection. Cancer Lett. 226, 77–84 (2005).
Article CAS PubMed Google Scholar
Gao, Q. et al. Deletion of the de novo DNA methyltransferase Dnmt3a promotes lung tumor progression. Proc. Natl. Acad. Sci. USA 108, 18061–18066 (2011).
Article CAS PubMed PubMed Central Google Scholar
Yan, Y., Tao, H., He, J. & Huang, S. Y. The HDOCK server for integrated protein-protein docking. Nat. Protoc. 15, 1829–1852 (2020).
Article CAS PubMed Google Scholar
Massague, J. TGFbeta signalling in context. Nat. Rev. Mol. Cell Biol. 13, 616–630 (2012).
Article CAS PubMed PubMed Central Google Scholar
Colak, S. & Ten Dijke, P. Targeting TGF-beta signaling in cancer. Trends Cancer 3, 56–71 (2017).
Article CAS PubMed Google Scholar
David, C. J. et al. TGF-beta tumor suppression through a lethal EMT. Cell 164, 1015–1030 (2016).
Article CAS PubMed PubMed Central Google Scholar
Huang, Y. H. et al. ID1 mediates escape from TGFbeta tumor suppression in pancreatic cancer.Cancer Discov 10, 142–157 (2020).
Article CAS PubMed Google Scholar
Su, J. et al. TGF-beta orchestrates fibrogenic and developmental EMTs via the RAS effector RREB1. Nature 577, 566–571 (2020).
Article CAS PubMed PubMed Central Google Scholar
Aragon, E. et al. Structural basis for distinct roles of SMAD2 and SMAD3 in FOXH1 pioneer-directed TGF-beta signaling. Genes Dev. 33, 1506–1524 (2019).
Article CAS PubMed PubMed Central Google Scholar
Ying, Z. et al. CCT6A suppresses SMAD2 and promotes prometastatic TGF-beta signaling. J. Clin. Invest. 127, 1725–1740 (2017).
Article PubMed PubMed Central Google Scholar
Xue, J. et al. Sustained activation of SMAD3/SMAD4 by FOXM1 promotes TGF-beta-dependent cancer metastasis. J. Clin. Investig. 124, 564–579 (2014).
Article CAS PubMed PubMed Central Google Scholar
Nakao, A. et al. Identification of Smad7, a TGFbeta-inducible antagonist of TGF-beta signalling. Nature 389, 631–635 (1997).
Article CAS PubMed Google Scholar
Hayashi, H. et al. The MAD-related protein Smad7 associates with the TGFbeta receptor and functions as an antagonist of TGFbeta signaling. Cell 89, 1165–1173 (1997).
Article CAS PubMed Google Scholar
Sun, N. et al. ATXN7L3 positively regulates SMAD7 transcription in hepatocellular carcinoma with growth inhibitory function. EBioMedicine 62, 103108 (2020).
Article CAS PubMed PubMed Central Google Scholar
Salot, S. & Gude, R. MTA1-mediated transcriptional repression of SMAD7 in breast cancer cell lines. Eur. J. Cancer 49, 492–499 (2013).
Article CAS PubMed Google Scholar
Yu, J. et al. MicroRNA-182 targets SMAD7 to potentiate TGFbeta-induced epithelial-mesenchymal transition and metastasis of cancer cells. Nat. Commun. 7, 13884 (2016).
Article CAS PubMed PubMed Central Google Scholar
Li, Y. et al. Metastatic heterogeneity of breast cancer cells is associated with expression of a heterogeneous TGFbeta-activating miR424-503 gene cluster. Cancer Res. 74, 6107–6118 (2014).
Article CAS PubMed Google Scholar
Cai, C. et al. PRMT5 enables robust STAT3 activation via arginine symmetric dimethylation of SMAD7. Adv. Sci. 8, 2003047 (2021).
Article CAS Google Scholar
Zhou, F. et al. Nuclear receptor NR4A1 promotes breast cancer invasion and metastasis by activating TGF-beta signalling. Nat. Commun. 5, 3388 (2014).
Article PubMed Google Scholar
Kit Leng Lui, S. et al. USP26 regulates TGF-beta signaling by deubiquitinating and stabilizing SMAD7. EMBO Rep. 18, 797–808 (2017).
Article PubMed PubMed Central Google Scholar
Tang, X. et al. Autocrine TGF-beta1/miR-200s/miR-221/DNMT3B regulatory loop maintains CAF status to fuel breast cancer cell proliferation. Cancer Lett. 452, 79–89 (2019).
Article CAS PubMed PubMed Central Google Scholar
Yang, B. et al. PHF14: an innate inhibitor against the progression of renal fibrosis following folic acid-induced kidney injury. Sci. Rep. 7, 39888 (2017).
Article CAS PubMed PubMed Central Google Scholar
Okano, M., Xie, S. & Li, E. Cloning and characterization of a family of novel mammalian DNA (cytosine-5) methyltransferases. Nat. Genet. 19, 219–220 (1998).
Article CAS PubMed Google Scholar
MacKenzie, D. J. et al. DNMT3B oncogenic activity in human intestinal cancer is not linked to CIMP or BRAFV600E mutation. iScience 23, 100838 (2020).
Article CAS PubMed PubMed Central Google Scholar
Lai, S. C. et al. DNMT3b/OCT4 expression confers sorafenib resistance and poor prognosis of hepatocellular carcinoma through IL-6/STAT3 regulation. J. Exp. Clin. Cancer Res. 38, 474 (2019).
Article CAS PubMed PubMed Central Google Scholar
Yu, J. et al. Telomerase reverse transcriptase regulates DNMT3B expression/aberrant DNA methylation phenotype and AKT activation in hepatocellular carcinoma. Cancer Lett. 434, 33–41 (2018).
Article CAS PubMed Google Scholar
Chen, B. F. & Chan, W. Y. The de novo DNA methyltransferase DNMT3A in development and cancer. Epigenetics 9, 669–677 (2014).
Article PubMed PubMed Central Google Scholar
Yang, L., Rau, R. & Goodell, M. A. DNMT3A in haematological malignancies. Nat. Rev. Cancer 15, 152–165 (2015).
Article CAS PubMed PubMed Central Google Scholar
Challen, G. A. et al. Dnmt3a and Dnmt3b have overlapping and distinct functions in hematopoietic stem cells. Cell Stem Cell 15, 350–364 (2014).
Article CAS PubMed PubMed Central Google Scholar
Rinaldi, L. et al. Dnmt3a and Dnmt3b associate with enhancers to regulate human epidermal stem cell homeostasis. Cell Stem Cell 19, 491–501 (2016).
Article CAS PubMed Google Scholar
Baubec, T. et al. Genomic profiling of DNA methyltransferases reveals a role for DNMT3B in genic methylation. Nature 520, 243–247 (2015).
Article CAS PubMed Google Scholar
Morselli, M. et al. In vivo targeting of de novo DNA methylation by histone modifications in yeast and mouse. Elife 4, e06205 (2015).
Article PubMed PubMed Central Google Scholar
Ooi, S. K. et al. DNMT3L connects unmethylated lysine 4 of histone H3 to de novo methylation of DNA. Nature 448, 714–717 (2007).
Article CAS PubMed PubMed Central Google Scholar
Zhang, Y. et al. Chromatin methylation activity of Dnmt3a and Dnmt3a/3L is guided by interaction of the ADD domain with the histone H3 tail. Nucleic Acids Res. 38, 4246–4253 (2010).
Article CAS PubMed PubMed Central Google Scholar
Lehnertz, B. et al. Suv39h-mediated histone H3 lysine 9 methylation directs DNA methylation to major satellite repeats at pericentric heterochromatin. Curr. Biol. 13, 1192–1200 (2003).
Article CAS PubMed Google Scholar
Vire, E. et al. The Polycomb group protein EZH2 directly controls DNA methylation. Nature 439, 871–874 (2006).
Article CAS PubMed Google Scholar
Velasco, G. et al. Dnmt3b recruitment through E2F6 transcriptional repressor mediates germ-line gene silencing in murine somatic tissues. Proc. Natl. Acad. Sci. USA 107, 9281–9286 (2010).
Article PubMed PubMed Central Google Scholar
de la Rica, L. et al. PU.1 target genes undergo Tet2-coupled demethylation and DNMT3b-mediated methylation in monocyte-to-osteoclast differentiation. Genome Biol. 14, R99 (2013).
Article PubMed PubMed Central Google Scholar
Sato, N., Kondo, M. & Arai, K. The orphan nuclear receptor GCNF recruits DNA methyltransferase for Oct-3/4 silencing. Biochem. Biophys. Res. Commun. 344, 845–851 (2006).
Article CAS PubMed Google Scholar
Miao, L., Liu, H. Y., Zhou, C. & He, X. LINC00612 enhances the proliferation and invasion ability of bladder cancer cells as ceRNA by sponging miR-590 to elevate expression of PHF14. J. Exp. Clin. Cancer Res. 38, 143 (2019).
Article PubMed PubMed Central Google Scholar
Zhang, L. et al. A novel PHD-finger protein 14/KIF4A complex overexpressed in lung cancer is involved in cell mitosis regulation and tumorigenesis. Oncotarget 8, 19684–19698 (2017).
Article PubMed PubMed Central Google Scholar
Huang, Q. et al. Depletion of PHF14, a novel histone-binding protein gene, causes neonatal lethality in mice due to respiratory failure. Acta Biochim. Biophys. Sin 45, 622–633 (2013).
Article CAS PubMed Google Scholar
Zhao, Y. et al. PHF14 promotes cell proliferation and migration through the AKT and ERK1/2 pathways in gastric cancer cells. Biomed. Res. Int. 2020, 6507510 (2020).
PubMed PubMed Central Google Scholar
Wu, S. et al. Silencing expression of PHF14 in glioblastoma promotes apoptosis, mitigates proliferation and invasiveness via Wnt signal pathway. Cancer Cell Int. 19, 314 (2019).
Article CAS PubMed PubMed Central Google Scholar
Park, J. E. et al. Pulsed SILAC-based proteomic analysis unveils hypoxia- and serum starvation-induced de novo protein synthesis with PHD finger protein 14 (PHF14) as a hypoxia sensitive epigenetic regulator in cell cycle progression. Oncotarget 10, 2136–2150 (2019).
Article PubMed PubMed Central Google Scholar
Zheng, S. et al. Molecular basis for bipartite recognition of histone H3 by the PZP domain of PHF14. Nucleic Acids Res. 49, 8961–8973 (2021).
Article CAS PubMed PubMed Central Google Scholar
Wan, J. C. M. et al. Liquid biopsies come of age: towards implementation of circulating tumour DNA. Nat. Rev. Cancer 17, 223–238 (2017).
Article CAS PubMed Google Scholar
Schwarzenbach, H., Hoon, D. S. & Pantel, K. Cell-free nucleic acids as biomarkers in cancer patients. Nat. Rev. Cancer 11, 426–437 (2011).
Article CAS PubMed Google Scholar

Download references

Acknowledgements

This work was supported by the National Natural Science Foundation of China (82030088, 81820108025, 82172927, 81972170, 81922050, 81972730); the Guangdong Basic and Applied Basic Research Foundation (2019A1515011174, 2020A1515010452, 2020A1515010130, 2019A1515110493), the Fundamental Research Funds for the Central Universities (22ykqb11), the Open Project of State Key Laboratory of Respiratory Disease (SKLRD-OP-202316, SKLRD-OP-202201), the Science and Technology Program of Guangzhou City (202201011315), the Medical Scientific Research Foundation of Guangdong Province (No. A2020633). We thank Ms. Xuanhong Zhang (Core Facilities for Medical Science, Zhongshan School of Medicine, Sun Yat-sen University) for assisting us performing the SPR kinetic analysis.

Author information

These authors contributed equally: Han Tian, Chenying Liu, Jianchen Yu.

Authors and Affiliations

Zhongshan School of Medicine, Sun Yat-sen University, Guangzhou, Guangdong, China
Han Tian, Jianchen Yu, Jianan Du, Shujun Liang, Wenting Wang, Qin Liu, Rong Lian, Yaokai Ye, Yi Yang, Xun Zhu, Junchao Cai, Jueheng Wu & Mengfeng Li
Cancer Institute, Southern Medical University, Guangzhou, Guangdong, China
Han Tian, Jian Han & Mengfeng Li
Department of Breast Pathology and Lab, Key Laboratory of Breast Cancer of Breast Cancer Prevention and Therapy, National Clinical Research Center of Cancer, Tianjin Medical University Cancer Institute and Hospital, Tianjin, China
Chenying Liu
School of Chemistry, South China Normal University, Guangzhou, Guangdong, China
Jianchen Yu
Department of Laboratory Medicine, Affiliated Cancer Hospital & Institute of Guangzhou Medical University, Guangzhou, Guangdong, China
Ting Zhu
Department of Biology, School of Basic Medical Science, Guangdong Medical University, Zhanjiang, Guangdong, China
Shanshan Wu
Department of Gastrointestinal Surgery, The First Affiliated Hospital, Sun Yat-sen University, Guangzhou, Guangdong, China
Tianyu Tao
Department of Cardiology, The First Affiliated Hospital, Sun Yat-sen University, Guangzhou, Guangdong, China
Jingjing Zhao

Authors

Han Tian
View author publications
You can also search for this author in PubMed Google Scholar
Chenying Liu
View author publications
You can also search for this author in PubMed Google Scholar
Jianchen Yu
View author publications
You can also search for this author in PubMed Google Scholar
Jian Han
View author publications
You can also search for this author in PubMed Google Scholar
Jianan Du
View author publications
You can also search for this author in PubMed Google Scholar
Shujun Liang
View author publications
You can also search for this author in PubMed Google Scholar
Wenting Wang
View author publications
You can also search for this author in PubMed Google Scholar
Qin Liu
View author publications
You can also search for this author in PubMed Google Scholar
Rong Lian
View author publications
You can also search for this author in PubMed Google Scholar
Ting Zhu
View author publications
You can also search for this author in PubMed Google Scholar
Shanshan Wu
View author publications
You can also search for this author in PubMed Google Scholar
Tianyu Tao
View author publications
You can also search for this author in PubMed Google Scholar
Yaokai Ye
View author publications
You can also search for this author in PubMed Google Scholar
Jingjing Zhao
View author publications
You can also search for this author in PubMed Google Scholar
Yi Yang
View author publications
You can also search for this author in PubMed Google Scholar
Xun Zhu
View author publications
You can also search for this author in PubMed Google Scholar
Junchao Cai
View author publications
You can also search for this author in PubMed Google Scholar
Jueheng Wu
View author publications
You can also search for this author in PubMed Google Scholar
Mengfeng Li
View author publications
You can also search for this author in PubMed Google Scholar

Contributions

M.L., H.T. and C.L. designed the study. H.T., C.L., J.Y., J.H., J.D., S.L., W.W., Q.L., R.L., T.Z., S.W., T.T. and Y. Ye performed the experiments and analyzed data. J.Z., Y. Yang, X.Z., J.C. and J.W. provided advices and help on the concept and experiments of this project. H.T. and C.L. wrote the manuscript. M.L. edited and approved the manuscript.

Corresponding author

Correspondence to Mengfeng Li.

Ethics declarations

Conflict of interest

The authors declare no competing interests.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Cite this article

Tian, H., Liu, C., Yu, J. et al. PHF14 enhances DNA methylation of SMAD7 gene to promote TGF-β-driven lung adenocarcinoma metastasis. Cell Discov 9, 41 (2023). https://doi.org/10.1038/s41421-023-00528-0

Download citation

Received: 04 July 2022
Accepted: 07 February 2023
Published: 18 April 2023
DOI: https://doi.org/10.1038/s41421-023-00528-0

This article is cited by

Novel genome-wide DNA methylation profiling reveals distinct epigenetic landscape, prognostic model and cellular composition of early-stage lung adenocarcinoma
- Junwen Gan
- Meng Huang
- Qingdong Cao
Journal of Translational Medicine (2024)
SMAD7 expression in CAR-T cells improves persistence and safety for solid tumors
- Sixin Liang
- Rui Zheng
- Bo Yan
Cellular & Molecular Immunology (2024)
Linderapyrone analogue LPD-01 as a cancer treatment agent by targeting importin7
- Takahiro Kitagawa
- Takahiro Matsumoto
- Tetsushi Watanabe
Journal of Natural Medicines (2024)
Single cell dynamics of tumor specificity vs bystander activity in CD8+ T cells define the diverse immune landscapes in colorectal cancer
- Daniel Morales Borràs
- Sara Verbandt
- Abhishek D. Garg
Cell Discovery (2023)