Cytoplasmic Expression of TP53INP2 Modulated by Demethylase FTO and Mutant NPM1 Promotes Autophagy in Leukemia Cells

Acute myeloid leukemia (AML) with a nucleophosmin 1 (NPM1) mutation is a unique subtype of adult leukemia. Recent studies show that NPM1-mutated AML has high autophagy activity. However, the mechanism for upholding the high autophagic level is still not fully elucidated. In this study, we first identified that tumor protein p53 inducible nuclear protein 2 (TP53INP2) was highly expressed and cytoplasmically localized in NPM1-mutated AML cells. Subsequent data showed that the expression of TP53INP2 was upregulated by fat mass and obesity-associated protein (FTO)-mediated m6A modification. Meanwhile, TP53INP2 was delocalized to the cytoplasm by interacting with NPM1 mutants. Functionally, cytoplasmic TP53INP2 enhanced autophagy activity by promoting the interaction of microtubule-associated protein 1 light chain 3 (LC3) - autophagy-related 7 (ATG7) and further facilitated the survival of leukemia cells. Taken together, our study indicates that TP53INP2 plays an oncogenic role in maintaining the high autophagy activity of NPM1-mutated AML and provides further insight into autophagy-targeted therapy of this leukemia subtype.


Introduction
Acute myeloid leukemia (AML) is a hematological malignancy carrying accumulated genetic abnormalities [1]. Nucleophosmin 1 (NPM1) mutation is one of the most common genetic alterations in AML, detected in approximately 30-35% of adult AML [2]. Notably, NPM1 mutations endow NPM1 mutants with stronger nuclear export ability that results in the cytoplasmic delocalization of NPM1 mutants, which is thought to serve a critical role in leukemogenesis [3]. Accumulating studies support NPM1 mutation as a driver lesion of myeloproliferation and hematological cancer [4,5]. However, a study in an NPM1mutated transgenic mice model indicated that NPM1 mutation alone is insufficient to initiate leukemia [6], suggesting the need for exploring other underlying oncogenic events in NPM1-mutated AML. Of note, emerging studies indicate that dysregulated autophagy is a non-negligible driving factor leading to leukemogenesis and malignant progression [7].
Autophagy is an intracellular catabolic degradation process that supports metabolic adaptation and nutrient recycling [8]. Plenty of studies demonstrate that dysfunctional autophagy contributes to tumorigenesis including leukemia [9,10]. Increasing evidence indicates that targeting autophagy holds promise for enhancing the therapeutic benefit of AML treatment [11]. In particular, our group has previously reported that NPM1-mutated AML cells have high autophagy activity [12]. Partial mechanistic studies reveal that pyruvate kinase isoenzyme M2 (PKM2) [13] and TNF receptor-associated factor 6 (TRAF6) [14] participate in the autophagic activation of leukemia cells. Recently, several novel insights related Int. J. Mol. Sci. 2023, 24, 1624 2 of 21 to autophagy regulation have been proposed [15,16]. Methyltransferase METTL3-mediated N 6 -methyladenosine (m 6 A) modification inhibits cell autophagy activity by reducing the PTEN stability and further results in the chemoresistance of leukemia cells [17]. Despite these existing studies, the underlying mechanism for autophagic activation has not been fully elucidated in NPM1-mutated leukemia.
Tumor protein p53 inducible nuclear protein 2 (TP53INP2) is a newly discovered scaffold protein that plays an important role in the process of autophagy activation [18]. TP53INP2 is first identified on human chromosome band 20q11.2 by in situ hybridization [19]. It is homologous to TP53INP1 which acts as a cofactor of p53 to regulate apoptosis and cell cycle [20,21]. In fact, TP53INP2 fails to participate in the regulation of p53 like TP53INP1 [22]. As a nuclear protein, TP53INP2 mainly participates in the transcriptional regulation of nuclear hormone receptors to sustain muscle or adipose wasting [23]. Recently, Mauvezin et al. reported that nuclear TP53INP2 delocalizes to the cytoplasm under nutrient starvation, stimulating autophagosome formation and accelerating the degradation of stable proteins in mammalian and drosophila cells [24]. Up to now, accumulating studies have shown that TP53INP2-mediated autophagy participates in the occurrence of many diseases, including adiposity [25], diabetes [26], and tumors [27]. However, the role of TP53INP2 in leukemia, especially in NPM1-mutated leukemia, has been unknown.
In our study, we first determined the high expression and cytoplasmic localization of TP53INP2 in NPM1-mutated leukemic cells. Mechanistically, TP53INP2 was upregulated by demethylase FTO-mediated m 6 A modification and delocalized into the cytoplasm by interacting with NPM1-mA. More importantly, cytoplasmic TP53INP2 could promote the interaction of LC3-ATG7 to enhance autophagy activity and further facilitate leukemia cell proliferation. Collectively, our findings uncover the oncogenic role of TP53INP2 in facilitating the high autophagy activity of NPM1-mutated AML and propose further insights into the autophagy-targeted therapy of this distinct leukemic entity.

Identification of the Autophagy-Related Genes in NPM1-Mutated AML
To identify the dysregulated autophagy-related genes in NPM1-mutated leukemia, we first investigated the differentially expressed genes (DEGs) between NPM1-mutated AML cases and NPM1-unmutated AML cases from three public datasets (TCGA, GSE15434, and Beat-AML) and took the intersection of DEGs from three datasets. A total of 110 candidate DEGs were obtained ( Figure 1A). Then, these candidate DEGs were used to intersect with the autophagy-related genes derived from the HADb database and three autophagy-related genes (TP53INP2, NKX2-3, and TUSC1) were screened out ( Figure 1B). Next, the heatmap showed that TP53INP2 and NKX2-3 were significantly gathered in NPM1-mutated AML cases whereas TUSC1 was mainly expressed in NPM1-unmutated AML cases ( Figure 1C). Additionally, upregulated TP53INP2 and NKX2-3 were identified in NPM1-mutated AML cases of GSE15434 datasets ( Figure 1D). Furthermore, gene set enrichment analysis (GSEA) showed that the positive regulation of the autophagy and autophagosome assembly process were significantly enriched in the high TP53INP2 expression group ( Figure 1E,F). However, autophagy-associated pathways were not obtained in the enrichment results of NKX2-3 and TUSC1. Finally, the Kaplan-Meier analysis indicated that AML patients with high TP53INP2 expression had poorer overall survival, while the other two genes showed no prognostic value ( Figure 1G). In addition, we investigated the expression profile of TP53INP1 which is highly homologous to TP53INP2, whereas the result showed no significant difference between NPM1-mutated and NPM1-unmutated AML patients. Thus, we focused on TP53INP2 for further investigations in this study. prognostic value ( Figure 1G). In addition, we investigated the expression profile of TP53INP1 which is highly homologous to TP53INP2, whereas the result showed no significant difference between NPM1-mutated and NPM1-unmutated AML patients. Thus, we focused on TP53INP2 for further investigations in this study. (C) Heatmaps of TP53INP2, NKX2-3, and TUSC1 genes in AML patients with NPM1 mutations and without NPM1 mutations from GSE15434 (n = 251) datasets. (D) The expression patterns of TP53INP2, NKX2-3, and TUSC1 transcripts in NPM1-mutated AML cases compared to NPM1-unmutated AML cases from GSE15434 datasets. (E,F) GSEA plots showing differential enrichment of genes related to autophagosome reassembly and positive regulation of autophagy. The data were from Beat-AML. (G) Survival analysis based on TP53INP2, NKX2-3, and TUSU1 mRNA expression level in primary NPM1-mutated AML samples of Beat-AML. Survival analysis was conducted by the use of Kaplan-Meier method.

TP53INP2 Is Aberrantly Expressed in NPM1-Mutated Leukemia Cells
To further determine the TP53INP2 expression in NPM1-mutated leukemia, we acquired the RNA-seq data of TP53INP2 in a group of leukemia-lymphoma cell lines in the CCLE database. The results showed that there is higher TP53INP2 expression in OCI-AML3 cells with NPM1 mutation type A(NPM1-mA) compared to other leukemia cell lines without NPM1-mA ( Figure 2A). Then, the expression of TP53INP2 transcripts in leukemia blasts derived from AML patients was detected. The results showed that TP53INP2 was dramatically upregulated in NPM1-mutated leukemia patients (n = 21) compared with NPM1-unmutated leukemia patients (n = 18) ( Figure 2B). Furthermore, qRT-PCR and Western blot assays indicated that TP53INP2 mRNA and proteins were highly expressed in NPM1-mA positive OCI-AML3 cells ( Figure 2C,D). Besides, the cellular localization of TP53INP2 was detected by immunofluorescence staining. The results revealed that TP53INP2 was largely expressed in the cytoplasm of OCI-AML3 cells and mostly expressed in the nucleus of OCI-AML2 and THP-1 cells ( Figure 2E). Similarly, cytoplasmic localization of TP53INP2 in OCI-AML3 cells was also confirmed by nucleus and cytoplasm fraction isolation assays ( Figure 2F-H). These observations demonstrate the high expression and cytoplasmic localization of TP53INP2 in NPM1-mutated leukemia cells.

TP53INP2 Is Upregulated by FTO-Mediated m 6 A Modification
Next, we explored the molecular mechanism underlying the high expression of TP53INP2 in NPM1-mutated leukemia cells. Considering NPM1-mA is a driver of AML [28], we first knocked down NPM1-mA to investigate whether it is involved in the regulation of TP53INP2. The results showed that NPM1 knockdown decreased TP53INP2 mRNA and protein levels in OCI-AML3 cells ( Figure 3A,B). Inversely, NPM1-mA overexpression increased TP53INP2 mRNA and protein levels in OCI-AML2 cells, but NPM1 wild-type (NPM1-wt) overexpression resulted in no significant changes ( Figure 3C,D). Nevertheless, based on the fact that NPM1 mutants cannot directly regulate gene transcription [29], we deduced that other molecular events participate in the regulation of TP53INP2 transcripts. Considering our recent studies showed that NPM1-mA can stabilize m 6 A demethylase FTO to participate in the transcriptional regulation of target genes [30], we investigated whether FTO-mediated demethylation regulates the expression of TP53INP2. Initially, several m 6 A methylated sites in the TP53INP2 mRNA were identified using the sequencebased RNA adenosine methylation site predictor (SRAMP) ( Figure 3E). Moreover, FTO was predicted to bind in the CDS region of the TP53INP2 sequence using the Starbase database ( Figure 3F). Subsequently, the interaction between TP53INP2 mRNA and FTO was verified by RNA-binding protein immunoprecipitation (RIP) assays ( Figure 3G). Next, silencing FTO significantly reduced TP53INP2 mRNA levels ( Figure 3H). Similarly, meclofenamic acid (MA) and FB23-2, two selective inhibitors for FTO activity, were applied to treat OCI-AML3 cells, and also observed that TP53INP2 mRNA levels were decreased ( Figure 3I). Furthermore, both FTO knockdown and activity inhibition increased the m 6 A modification levels in the TP53INP2 mRNA ( Figure 3J,K) and resulted in the decreased stability of TP53INP2 mRNA ( Figure 3L). These results indicate that the expression of TP53INP2 is upregulated by FTO-mediated m 6 A modification in NPM1-mutated leukemia cells. cence staining. The results revealed that TP53INP2 was largely expressed in the cytoplasm of OCI-AML3 cells and mostly expressed in the nucleus of OCI-AML2 and THP-1 cells ( Figure  2E). Similarly, cytoplasmic localization of TP53INP2 in OCI-AML3 cells was also confirmed by nucleus and cytoplasm fraction isolation assays ( Figure 2F-H). These observations demonstrate the high expression and cytoplasmic localization of TP53INP2 in NPM1-mutated leukemia cells. Cytoplasmic versus nuclear TP53INP2 in OCI-AML3, OCI-AML2, and THP-1 cells were quantified by Image J software. Scale bar: 50 µm. (F-H) Western blot analysis of cytoplasmic and nuclear TP53INP2 protein levels in OCI-AML3, OCI-AML2, and THP-1 cells. GAPDH was used as the cytoplasmic control. Histone H3 was used as the nuclear control. Data are presented as the mean ± SD of three independent results. * p < 0.05, *** p < 0.001. NPM1-mA protein levels in NPM1-mA-silenced OCI-AML3 and NPM1-mA-enforced OCI-AML2 cells. β-actin was used as the control. The bar graph presented relative protein levels. (E,F) Prediction score of m 6 A distribution in TP53INP2 mRNA sequence using SRAMP and prediction of the binding sites between TP53INP2 mRNA and demethylase FTO using Starbase. (G) RIP assay was used to determine the binding between TP53INP2 mRNA and FTO. (H,I) qRT-PCR analysis of TP53INP2 mRNA levels after silencing FTO and inhibiting FTO activity using 20 µM FB23-2 and 50 µM MA for 24 h. (J,K) m 6 A RIP assay was used to determine the m 6 A levels in TP53INP2 mRNA. (L) qRT-PCR following the addition of ActD was used to detect TP53INP2 mRNA stability. Data are presented as the mean ± SD of three independent results. * p < 0.05, ** p < 0.01, *** p < 0.001. ns. indicates no significant difference.

TP53INP2 Is Delocalized into the Cytoplasm by Interacting with NPM1-mA
Followingly, we looked into the reason that TP53INP2 is largely located in the cytoplasm of NPM1-mutated leukemia cells. Considering that NPM1-mA engages in the translocation of various nuclear proteins [31], we first investigated whether the delocalization of TP53INP2 is related to NPM1-mA. The results showed that NPM1-mA overexpression increased cytoplasmic TP53INP2 and decreased nuclear TP53INP2 in OCI-AML2 and THP-1 cells ( Figure 4A,B). Meanwhile, immunofluorescence analysis showed that both NPM1-mA and TP53INP2 co-localized in the cytoplasm of OCI-AML3 cells ( Figure 4C). Furthermore, the interaction of endogenous TP53INP2 and NPM1-mA was detected by Co-Immunoprecipitation (Co-IP) analysis ( Figure 4D,E). Additionally, the combination between TP53INP2 and NPM1-mA was also confirmed in HEK293T cells co-transfected with HA-TP53INP2 and Flag-NPM1-mA plasmids ( Figure 4F,G). Finally, cytoplasmic TP53INP2 was relocalized to the nucleus following treatment with KPT-330, an inhibitor of nuclear exportation of NPM1-mA ( Figure 4H). Similar results of TP53INP2 relocalization were also shown in immunofluorescence analysis ( Figure 4I). The findings demonstrate that TP53INP2 is delocalized into the cytoplasm of leukemia cells by interacting with NPM1-mA.

TP53INP2 Facilitates Autophagy Activity by Promoting the Interaction of LC3-ATG7
To clarify the role of TP53INP2 in the autophagy process of NPM1-mutated leukemia cells, we first silenced TP53INP2 to observe the changes in autophagic level. The results showed that TP53INP2 depletion decreased LC3-II/I levels and increased p62 levels ( Figure 5A). Similar results were obtained by immunofluorescence analysis, as indicated by the accumulated LC3 puncta in OCI-AML3 cells ( Figure 5B). Conversely, TP53INP2 overexpression elevated LC3II/I ratio and diminished p62 levels in OCI-AML2 cells ( Figure 5C). In addition, an increased number of LC3 puncta was observed by immunofluorescence analysis ( Figure 5D). Given that autophagy-related gene 7 (ATG7) interacts with LC3 and is an important link in autophagosome formation [32], we explored the role of TP53INP2 on the interaction of LC3-ATG7. The String analysis indicated the potential interaction of TP53INP2, LC3, and ATG7 ( Figure 5E). Subsequently, the results of a Co-IP assay confirmed that TP53INP2 interacted with LC3 and ATG7 in OCI-AML3 cells ( Figure 5F-H). Furthermore, immunoprecipitation analysis revealed that the silencing of TP53INP2 decreased the amount of ATG7 proteins pulled down by LC3 ( Figure 5I). In contrast, overexpressing TP53INP2 increased the amount of ATG7 proteins pulled down by LC3 ( Figure 5J). In addition, immunofluorescent results showed that the depletion of TP53INP2 diminished the accumulation of LC3 and ATG7 puncta ( Figure 5K). In conclusion, these data suggest that TP53INP2 enhances autophagy by promoting the interaction of LC3-ATG7.

TP53INP2-Mediated Autophagy Is Vital for Leukemia Cell Survival
Finally, we explored whether TP53INP2-enhanced autophagy is required for leukemia cell survival. The autophagy activator rapamycin and autophagy inhibitor 3-methyladenine (3-MA) was used to treat leukemia cells, respectively. The results showed that rapamycin treatment reversed the decreased LC3-II/I levels and the increased p62 levels in TP53INP2silenced OCI-AML3 cells ( Figure 6A). Furthermore, TP53INP2 knockdown inhibited the cell proliferation, and then the effect was partially rescued by rapamycin treatment (Figure 6B,C). In contrast, 3-MA treatment disrupted the enhanced autophagy activity and cell proliferation ability in TP53INP2-enforced OCI-AML2 cells ( Figure 6D-F). In the end, we performed a rescue experiment to further probe the effect of TP53INP2 on NPM1-mAmaintained autophagy and cell survival. The data indicated that NPM1-mA knockdown reduced LC3II/I ratio and increased p62 protein levels, and then these changes were successfully rescued by TP53INP2 overexpression ( Figure 6G). Moreover, CCK-8 assays showed that overexpressing TP53INP2 partially retrieved the effect of NPM1-mA depletion on cell proliferation ( Figure 6H). Collectively, these results indicate that TP53INP2-enhanced autophagy is essential for the survival of NPM1-positive leukemia cells.

Discussion
Accumulating evidence has supported that dysfunctional autophagy is closely associated with leukemogenesis and development [10]. However, the mechanism of high autophagic levels in NPM1-mutated leukemia has not yet been fully clarified. Here, we used bioinformatics analysis to screen out the autophagy-associated dysregulated gene TP53INP2 in NPM1-mutated leukemia. Subsequent data revealed that TP53INP2 was upregulated by demethylase FTO-mediated m 6 A modification and delocalized to the cytoplasm by interacting with NPM1-mA. Importantly, cytoplasmic TP53INP2 enhanced autophagy activity by promoting the interaction of LC3-ATG7, thereby maintaining leukemia cell survival (Figure 7). cell growth in TP53INP2-silenced OCI-AML3 cells treated with 5 μM rapamycin for 10 h and TP53INP2-enforced OCI-AML2 cells treated with 2 mM 3-MA for 10 h, respectively. (C,F) EdU analysis of cell proliferation in TP53INP2-silenced OCI-AML3 cells and TP53INP2-enforced OCI-AML2 cells. The bar graph showed the percentage of EdU-positive cells. Scale bar: 100 μm. (G) Western blot analysis of LC3II/I and p62 expression in the NPM1-mA-silenced OCI-AML3 cells following TP53INP2 overexpression. (H) CCK-8 analysis of cell growth ability in NPM1-mA-silenced OCI-AML3 cells following TP53INP2 overexpression. The data are presented as the mean ± SD of three independent experiments. * p < 0.05, ** p < 0.01, *** p < 0.001.

Discussion
Accumulating evidence has supported that dysfunctional autophagy is closely associated with leukemogenesis and development [10]. However, the mechanism of high autophagic levels in NPM1-mutated leukemia has not yet been fully clarified. Here, we used bioinformatics analysis to screen out the autophagy-associated dysregulated gene TP53INP2 in NPM1-mutated leukemia. Subsequent data revealed that TP53INP2 was upregulated by demethylase FTO-mediated m 6 A modification and delocalized to the cytoplasm by interacting with NPM1-mA. Importantly, cytoplasmic TP53INP2 enhanced autophagy activity by promoting the interaction of LC3-ATG7, thereby maintaining leukemia cell survival (Figure 7). In the current study, we first utilized bioinformatics analysis to ascertain the dysregulated genes associated with autophagy in NPM1-mutated leukemia and obtained three genes including TUSC1, NKX2-3, and TP53INP2. Given that survival analysis showed the high expression of TP53INP2 was related to poor prognosis, we mainly focused on TP53INP2 in this study. Then, we confirmed the high expression of TP53INP2 in OCI-AML3 cells and primary blasts. Indeed, RNA-seq data from another study also indicated the upregulation of TP53INP2 in NPM1-mutated leukemia cells [33]. Followingly, we explored the mechanism of high TP53INP2 expression. Initially, we revealed that knockdown and overexpression of NPM1-mA affected the level of TP53INP2 transcripts. Considering NPM1-mA cannot directly regulate the transcription of target genes [29] but NPM1-mA can stabilize demethylase FTO to participate in transcriptional regulation [30], In the current study, we first utilized bioinformatics analysis to ascertain the dysregulated genes associated with autophagy in NPM1-mutated leukemia and obtained three genes including TUSC1, NKX2-3, and TP53INP2. Given that survival analysis showed the high expression of TP53INP2 was related to poor prognosis, we mainly focused on TP53INP2 in this study. Then, we confirmed the high expression of TP53INP2 in OCI-AML3 cells and primary blasts. Indeed, RNA-seq data from another study also indicated the upregulation of TP53INP2 in NPM1-mutated leukemia cells [33]. Followingly, we explored the mechanism of high TP53INP2 expression. Initially, we revealed that knockdown and overexpression of NPM1-mA affected the level of TP53INP2 transcripts. Considering NPM1-mA cannot directly regulate the transcription of target genes [29] but NPM1-mA can stabilize demethylase FTO to participate in transcriptional regulation [30], we investigated whether FTO-mediated demethylation modulates TP53INP2 expression. Then, the binding of TP53INP2 mRNA and FTO was confirmed by RIP experiments. Moreover, m 6 A RIP and RNA stability assays further identified that FTO upregulated TP53INP2 expression by reducing the m 6 A modification in the TP53INP2 transcripts. In line with our finding, a recent study indicated that the inhibition of FTO activity reduces PPARG mRNA levels in bone marrow mesenchymal stem cells [34]. In fact, up to 40% of all mRNA expression is subject to regulation by FTO-mediated demethylation in AML [35]. It has been well documented that m 6 A modification must be identified by reader proteins, which leads to different destinies of target RNA [36,37]. Qing et al. reported that reader protein YTHDF2 recognizes m 6 A modification in PFKP and LDH, and further upregulates their post-transcriptional expression in AML [38]. Thus, further studies are required to determine which reader protein identifies the m 6 A modification in TP53INP2. In this work, we also observed that cytoplasmic TP53INP2 was more abundant in leukemia cells with NPM1-mA than those without NPM1-mA. Furthermore, the results of Co-IP assays confirmed that NPM1 mutants interacted with TP53INP2 and gave rise to the cytoplasmic delocalization of TP53INP2. Indeed, plenty of studies indicated that NPM1 mutants act as a chaperone protein to induce cytoplasmic delocalization of many nuclear proteins including ARF [39], CTCF [40], HEXIM1 [41], PU.1 [42]. Of note, the chaperone activity of NPM1 mutants relies on its hydrophobic N-terminal region [31]. Therefore, the exact domain necessitated for the binding between NPM1 mutants and TP53INP2 demands to be further investigated.
Next, we attempted to explore the biological function of TP53INP2 in NPM1-mutated leukemia cells. First, we observed that TP53INP2 knockdown reduced the LC3 II/I ratio and increased p62 levels while TP53INP2 overexpression performed the opposite effect, which indicated that TP53INP2 contributed to the autophagy activation. Our idea was supported by Yang et al. and reported that the downregulation of TP53INP2 induced by oxidative stress suppresses the autophagy degradation pathway in bone mesenchymal stem cells [43]. Additionally, it is said that TP53INP2 overexpression induces the activation of colon cancer cell autophagy via the HEDGEHOG signaling pathway [27]. Next, we explored the molecular mechanism by which TP53INP2 activated autophagy in leukemia. The results demonstrated that TP53INP2 could directly interact with ATG7 and further stabilize the binding of ATG7 and LC3, contributing to autophagosome formation. Parallelly, Liu et al. reported that TP53INP2 acts as a scaffold protein to promote the interaction of ATG7 and LC3 in HEK293T cells [44]. In addition, it is quoted that TP53INP2 chaperones deacetylated nuclear LC3 to the cytoplasm to promote macroautophagy [45]. Recently, TP53INP2 has been identified as a novel autophagic adaptor that recruits ubiquitin and ubiquitinated proteins to autophagosomes for degradation [46]. Nowak et al. proposed that TP53INP2 could interact with the transmembrane protein VMP1 to promote the formation of autophagosomes [47]. It is ordinarily recognized that autophagy is a multistep process in which distinct sets of proteins control different steps [48]. Therefore, further studies are needed to investigate whether TP53INP2 regulates other autophagy-related components to engage in the autophagy process in NPM1-mutated leukemia.
In the end, we assessed the role of TP53INP2-mediated autophagy in maintaining the survival of leukemia cells. The data uncovered that the autophagy inhibitor 3-MA treatment could abrogate the enhancement of TP53INP2-mediated autophagic activation on leukemic cell survival. Furthermore, NPM1-mA knockdown impaired autophagy activity and cell proliferation, and TP53INP2 overexpression partially reversed the effect of NPM1-mA deficiency. These observations confirmed that TP53INP2-modulated autophagy was essential for the survival of NPM1-mutated leukemia cells. In keeping with our findings, TP53INP2 is reported to induce autophagy activation to promote pancreatic cancer progression [49]. Additionally, Hu et al. claimed that TP53INP2-mediated basal autophagy makes liposarcoma cells more resistant to bortezomib-induced growth suppression [50]. Apart from being involved in autophagy, TP53INP2 can modulate epithelial-to-mesenchymal transition via the GSK-3β/β-Catenin/Snail1 pathway in bladder cancer cells [51]. Moreover, a recent study has uncovered a novel role for TP53INP2, which acts as an accelerator of caspase-8 activation, favoring death receptor-mediated apoptosis [52]. Therefore, a deeper exploration of other biological functions of TP53INP2 contributes to revealing its distinct role in NPM1-mutated leukemia. Of course, further work is necessitated to clarify the role of TP53INP2-mediated autophagy in mouse knock-in models that mimic human NPM1-mutated leukemia. In a word, our study first reveals that the high expression of TP53INP2 mediated by FTO and cytoplasmic delocalization of TP53INP2 modulated by NPM1-mA jointly facilitates autophagic pro-survival, which provides a new theoretical basis for NPM1-mutated AML therapy.

Data Analysis of TCGA, GEO, Beat-AML, CCLE, and HADb Databases
The RNA-seq data and clinical characteristics of leukemia patients were obtained from The Cancer Genome Atlas (TCGA, n = 170, http://www.cancergenome.nih.gov, accessed on 12 October 2021), the Beat-AML database (n = 448, http://vizome.org/additional_figures_ BeatAML.html, accessed on 12 October 2021), and the NCBI Gene Expression Omnibus (GEO) under the accession number GSE15434 (n = 251, https://www.ncbi.nlm.nih.gov/gds, accessed on 12 October 2021). The expression data of target genes in leukemia cell lines were downloaded from the Cancer Cell Line Encyclopedia (CCLE, https://depmap.org/portal/, accessed on 15 November 2021). The autophagy-associated genes were obtained through the Human Autophagy Database (HADb, http:// www.autophagy.lu/, accessed on 18 October 2021). We utilized R software (Version 3.6.3) to analyze the above data. First, the limma package was used to acquire differentially expressed genes (DEGs) (p value < 0.05 as a cut-off criterion). Then, these DEGs were visualized by R packages. The Veen diagrams were obtained using the R package VennDiagram. The heatmap was created using pheatmap and ggplot2 packages. For Gene Set Enrichment Analysis (GSEA), R package clusterProfiler was used to visualize the autophagy-associated signal pathways. At last, the Kaplan-Meier survival analyses were performed by R package survival.

Clinical Samples
The newly diagnosed AML patient samples derived from bone marrow and peripheral blood were provided by the Chongqing University Cancer Hospital. The study was approved by the Ethics Committee of Chongqing Medical University, according to the Declaration of Helsinki. Mononuclear cells were acquired by the use of Ficoll Lymphocyte Separation Solution (Hao Yang Biological Manufacture Co, Tianjin, China, #TBD2013CHU05), referring to the manufacturer's guidance. The clinical characteristics of leukemia patients were shown in Table 1.

Plasmid Constructs and Cell Transfection
The plasmids coding Flag-NPM1-mA were bestowed by Dr. CJ Sherr (Genetics and  Tumor

Quantitative Real-Time PCR (qRT-PCR)
The total RNAs were extracted by the use of the TRIzol reagent (Takara, Kyoto, Japan) and transcribed into cDNA using PrimeScript TM RT Master Mix (Takara, #RR036A). A quantitative real-time PCR (qRT-PCR) was performed using TB Green TM Premix Ex Taq TM II (Tli RNaseH Plus) (Takara, #RR820A) on a CFX Connect TM real-time system (Bio-Rad, Hercules, CA, USA), according to the manufacturer's protocols. Cycling conditions were the 30 s at 95 • C for the preliminary denaturation, and amplification was operated with 39 cycles of 5 s at 95 • C, 30 s at 58 • C, 20 s at 72 • C, and lastly 10 min at 72 • C for the extension. The mRNA levels were evaluated through the 2 −∆∆Ct method and normalized by β-actin. Details about the primer sequences were shown in Table 2. Table 2. The PCR primer sequences for each gene used in this study.

Western Blot Analysis
To inspect the target protein expression, Western blot analysis was carried out. The collected cells were cleaned three times and cracked in cell lysis buffer (Beyotime, #P0013C) with protease inhibitors (Bimake, Houston, TX, USA, #B14001). Insoluble material was removed through centrifugation at 13,300 rpm at 4 • C for 30 min, and the protein concentrations were measured using the BCA Protein Assay Kit (Beyotime, #P0010S). Subsequently, total proteins from each sample were isolated on a 12% sodium dodecyl sulfatepolyacrylamide gel electrophoresis (SDS-PAGE) and then shifted to polyvinylidene fluoride (PVDF) membranes (Bio-Rad, #1620177). The membranes were blocked in 5% non-fat milk. The PVDF membranes were incubated with the primary antibodies overnight at 4 • C. HRPconjugated goat anti-rabbit IgG (Biosharp, Beijing, China, #BL003A) and HRP-conjugated goat anti-mouse IgG (Biosharp, #BL001A) were used as secondary antibodies at 1:5000 dilution. Detection was visualized by using an enhanced chemiluminescence solution.

Immunofluorescence
Cell fluorescence was detected with a double-label multiplex immunofluorescence kit (AiFang Biological, Hunan, China, #AFIHC023). The cells were stained with TYR-520 or TYR-570 fluorescent dye for 15 min and washed three times with PBS according to the manufacturer's instructions. 4 ,6-diamino-2-phenylindole (DAPI, Beyotime, China) was utilized for the nucleus staining. Cells were detected utilizing a fluorescence microscope (Nikon, Tokyo, Japan).

RNA-Binding Protein Immunoprecipitation (RIP) Assays
The EZMagna RIP Kit (Millipore, Burlington, MA, USA) was applied to investigate the interaction between TP53INP2 mRNA and FTO. The cells were rinsed twice with PBS and lysed in RIP lysis buffer supplemented with protease and RNase inhibitors. Then the extraction was incubated overnight at 4 • C with magnetic beads that were coated with an anti-FTO Rabbit pAb (Cell Signaling Technology, 3687S), an anti-m 6 A Rabbit pAb (Abcam, #286164) or a control IgG (Cell Signaling Technology, #2729S). Next, the protein A/G beads were cleaned, and the complex was mixed with Proteinase K to remove proteins. Finally, the target RNA was separated from the immunoprecipitated assay, and qRT-PCR was conducted.

mRNA Stability Assays
To examine the mRNA stability in leukemia cells, 5 µg/mL of Actinomycin D (Med-ChemExpress, Monmouth Junction, NJ, USA, #HY-17559) was added into the cell culture medium to disturb RNA transcription. The cells were collected at selected time points, and RNA was extracted for qRT-PCR.

Nucleus and Cytoplasm Fraction Isolation Assays
The nuclear and cytoplasmic fractions of the leukemic cells were separated by the use of the Nuclear and Cytoplasmic Protein Extraction Kit (Beyotime, #P0028). The harvested and washed cells were lysed in the Cytoplasmic Protein Extraction Reagent A of the Kit. After being vortexed and rested on ice for 20 min, the Cytoplasmic Protein Extraction Reagent B was added. The cytoplasmic protein was obtained by centrifugation at 15,000× g at 4 • C for 10 min. After the supernatant was isolated, the sediment was lysed with Nuclear Protein Extraction Reagent on ice for 30 min. The nuclear protein was obtained by centrifugation at 15,000× g at 4 • C for 10 min. Finally, the extracted product was detected by western blot assay, respectively.

Co-Immunoprecipitation Assays
The cells were lysed using the IP lysis buffer (Beyotime, #P0013) with protease inhibitors after being harvested and washed. The total cell lysate was obtained after centrifugation at 14,000× g at 4 • C for 30 min. Next, the A/G beads were washed twice with PBST and then incubated with a negative control IgG and special antibodies including anti-TP53INP2 antibody (Lifespan Biosciences, #LS-C743322), anti-NPM1 mutant antibody (Thermo Fisher Scientific, #PA1-46356), anti-LC3B antibody (ZEN-BIOSCIENCE, #350140) and anti-ATG7 antibody (Bimake, #A5121) for 1 h. After being washed with PBST three times again, the protein A/G beads coated with antibodies were incubated with total cell lysate on a rotating shaker at 4 • C all night. Protein A/G beads bound to immune complexes were washed twice with PBST and then boiled in 2 × SDS-PAGE Loading Buffer (Beyotime, #P0015B) for five minutes. The product was obtained and immunoblotted after the beads were removed by magnetic adsorption.

Cell Counting Kit-8 (CCK-8) Assays
The proliferation of cells was detected by CCK-8 assay. The lentivirus-infected cells or plasmid-transfected cells were plated into 96-well plates at 4 × 10 3 cells per well and then react with corresponding reagents. Next, cells were cultured in RPMI-1640 medium including 10% FBS. At selected time points (0-72 h) after seeding, 10 µL CCK8 solution (Solarbio, Beijing, China, #CA1210) was put into each well, and the plate was put into a constant temperature water box at 37 • C for 2-3 h in darkness. The absorbance value was detected with a microplate reader (BioTeck, Palo Alto, CA, USA).

5-Ethynyl-2 -Deoxyuridine (EdU) Assays
The cell proliferation was analyzed using a 5-Ethynyl-2 -Deoxyuridine (EdU) assay using the EdU Cell Proliferation Kit with Alexa Fluor 594 (Beyotime, #C0078S) to detect cell viability. First, the cells were seeded to a 6-well plate at 1 × 10 6 cells per well and then 10 µM EdU was added in each well at 37 • C for 3 h. After being rinsed twice, the cells were prepared as a suspension and spread on a slide, and then fixed in 4% paraformaldehyde for 15 min Next, the slides were covered with a click reaction solution for 30 min in the dark. Subsequently, the cells were treated using 0.5% Triton X-100 for 15 min, and DAPI was applied to label the cell nuclei for 30 min. Cell proliferation was evaluated by using the ratio of EdU-stained cells (with red fluorescence) to DAPI-stained cells (with blue fluorescence). Images were analyzed using a fluorescence microscope (Nikon, Tokyo, Japan) at 100× magnification.

Statistical Analysis
The results were verified by three independent experiments at least. The data were normally distributed and expressed as the mean ± standard deviation (SD). The statistical analysis was conducted with GraphPad software. (Version 8.0). The statistical significance was evaluated using unpaired Student's t-tests (two groups) and the one-way analysis of variance (more than two groups). The p-value < 0.05 was assessed as statistically significant (* p < 0.05, ** p < 0.01, *** p < 0.001). No significance was defined as ns.