Abstract

   Abnormal Notch1 expression has an important role in tumorigenesis.
   However, upstream control mechanisms for Notch1 are still
   insufficiently understood. Acute myeloid leukemia (AML) is one of the
   most common and lethal blood malignancies with limited possibilities
   for treatment. Thus, new therapeutic targets are urgently needed to
   improve current ineffective therapies. Herein, high Annexin A1 (ANXA1)
   expression is found correlated with hyperproliferation of AML cells,
   and then ANXA1 is identified as a novel negative regulator of Notch1
   function in AML. Mechanistically, ANXA1 directly bound to the
   intracellular domain of Notch1 (NICD) to target this tumor suppressor
   for degradation. Furthermore, NICD executed its tumor suppressive
   function through activation of the p15 promoter. Thus, ablation of the
   Notch1‐p15‐mediated tumor suppression by ANXA1 provided a novel
   mechanism of AML proliferation. In human AML patients, a mutual
   exclusive relation is discovered between ANXA1 and Notch1/p15,
   corroborating mechanistic discovery. On the basis of these results, it
   is reasonably speculated that targeting ANXA1 would provide an
   effective approach for treatment of AML. In support of this new
   therapeutic paradigm, provided proof‐of‐concept data by antagonizing
   ANXA1 using NICD inhibitory peptides.

   Keywords: acute myeloid leukemia, annexin a1, anxa1‐notch1‐p15 signal
   axis, notch1, peptide inhibitors, protein‐protein interaction
     __________________________________________________________________

   High ANXA1 expression correlates with hyperproliferation of AML cells
   because ANXA1 interacts with NICD to target this tumor suppressor for
   degradation. Based on the novel mechanism of ablation of the
   Notch1‐p15‐mediated tumor suppression in AML hyperproliferation,
   targeting ANXA1 provides an effective approach for AML treatment, in
   which antagonizing ANXA1 using NICD inhibitory peptides is an effective
   strategy for AML treatment.

   graphic file with name ADVS-11-2409726-g005.jpg

1. Introduction

   Notch1 signaling is an evolutionarily conserved pathway that regulates
   cell fate decisions in neuronal, immune, cardiovascular, and endocrine
   systems.^[ [66]^1 ^] Considering the key role of Notch1 in regulating
   cell processes, such as proliferation, differentiation, and apoptosis.
   The abnormally expressed Notch1 may be involved in the genesis and
   development of malignant tumors.^[ [67]^2 ^] Notably, the expression
   level and function of Notch1 vary in different kinds of cancers. Recent
   studies reported the positive effects of Notch1 in promoting
   proliferation and metastasis in hepatocellular carcinomas, bladder
   tumor‐initiating cells, colorectal cancers, and triple‐negative breast
   cancers, but the opposite results were found in skin and forebrain
   tumor subtypes.^[ [68]^3 ^] Furthermore, mutations and subsequent
   activation of the NOTCH1 gene occur in more than 65% of T‐cell acute
   lymphoblastic leukemia (T‐ALL) patients, but are rare in acute myeloid
   leukemia (AML) patients.^[ [69]^4 ^] Notch1 is a tumor‐suppressive gene
   in AML.^[ [70]^5 ^] Notch1 is a single‐pass transmembrane protein that
   releases its intracellular domain (NICD) via γ‐secretase activity.^[
   [71]^6 ^] The NICD then binds to the recombination signal binding
   protein for immunoglobulin kappa J region (RBPJ, also known as CSL)
   transcription factor complex after nuclear translocation, to trigger
   the activation of target genes.^[ [72]^7 ^] However, as an important
   target for disease treatment, the regulatory factors affecting NICD
   functions remain largely unexplored.

   Annexin A1 (ANXA1) is a Ca^2+‐dependent phospholipid‐binding protein
   that participates in a variety of important cell processes, such as
   inflammation, cell survival, proliferation, apoptosis, differentiation,
   and migration.^[ [73]^8 ^] ANXA1 has the ability to inhibit the
   activity of cytosolic phospholipase A2 and cyclooxygenase 2 to exhibit
   anti‐inflammatory, antipyretic, and anti‐hyperalgesic activity, as well
   as it can be induced in human cancer cells and mice by
   anti‐inflammatory drugs to inhibit the activation of NF‐κB by binding
   to its p65 subunit.^[ [74]^8 , [75]^9 ^] ANXA1 regulates the
   extracellular signal‐regulated kinase (ERK) pathway at a proximal
   location, by Src Homology 2 (SH2) domain‐independent association with
   the adapter protein Grb‐2, to exert its antiproliferative activity by
   ERK‐mediated disruption of the actin cytoskeleton and ablation of
   cyclin D1, but not by ERK‐mediated induction of p21^cip/waf, as well as
   by the constitutive activation of the mitogen‐activated protein kinase
   (MAPK)/ERK pathway linked to its phosphorylation by epidermal growth
   factor.^[ [76]^10 ^] Notably, ANXA1 has been implicated in complex and
   contradictory functions in cancers, including tumor growth, metastasis,
   invasion, and drug resistance.^[ [77]^11 ^] ANXA1 negatively regulated
   the phosphorylation of GSK3β, inhibited the epithelial‐mesenchymal
   transformation and migration and invasion of non‐small cell lung cancer
   (NSCLC) cells.^[ [78]^12 ^] Ribonucleotide reductase subunit M2
   stabilizes ANXA1 and activates the AKT pathway, thereby promoting
   sunitinib resistance.^[ [79]^13 ^] ANXA1 acts as an important
   homeostatic protein by binding to the target protein on the membrane of
   some cancer cells.^[ [80]^14 ^] As a therapeutic and diagnostic target,
   ANXA1 is a double‐edged sword, depending on the tumor type.^[ [81]^15
   ^] The current study found that among B‐cell malignancies, ANXA1 was a
   specific marker for hairy cell leukemia.^[ [82]^16 ^] However, the
   function and mechanisms of ANXA1 in AML are largely unclear.

   Herein, we identified ANXA1 as a novel upstream negative regulator of
   Notch1 through the direct interaction between the 2,050–2,060 amino
   acid domain of NICD and the C‐terminal 214–228 amino acid domain of
   ANXA1 on cell membranes. We also identified p15, a cyclin‐dependent
   kinase inhibitor, as a novel target gene of NICD, which was regulated
   by binding to the promotor of p15. In human AML patients, we discovered
   a mutually exclusive relationship between ANXA1 and Notch1/p15, which
   confirmed our mechanistic finding. In brief, we uncovered an
   unanticipated ablation of the Notch1‐p15‐mediated tumor suppression by
   ANXA1 in AML through the ubiquitination‐proteasome degradation of NICD
   to block the transcriptional activation of p15 in vitro and in vivo. On
   the basis of these results, we speculated that targeting ANXA1 would
   provide an effective approach for treatment of AML. In support of this
   new therapeutic paradigm, we provided proof‐of‐concept data by
   antagonizing ANXA1 using NICD inhibitory peptides. Thus, these findings
   reveal a key role of ANXA1 in the negative regulation of
   Notch1‐p15‐mediated tumor suppression in AML. This mechanistic finding
   may provide a novel concept for the further development of anti‐AML
   drugs targeting direct interactions between ANXA1 and Notch1 on cell
   membranes.

2. Results

2.1. Negative Regulation of Notch1 Expression by ANXA1 in AML Cells

   To identify novel signaling molecules and therapeutic targets in AML,
   we performed a study of Gene Expression Profile (GEP) analyses from
   Gene Expression Omnibus (GEO) and The Cancer Genome Atlas (TCGA)
   datasets, according to the review of molecular biomarkers in AML.^[
   [83]^17 ^] ANXA1 was identified as one of the most highly expressed
   biomarkers in AML (Figures [84]1A,B and [85]S1, Supporting
   Information). Therefore, we focused on defining functional roles of
   ANXA1 in AML using a loss‐of‐function approach by knocking‐down ANXA1
   using shRNAs. Two shRNAs with different stem structures targeting ANXA1
   were generated, in which shANXA1#2 was more potent in mediating ANXA1
   knockdown (Figure [86]S2A, Supporting Information). The KG1a and HL60
   cell lines were derived from patients with the poorest prognosis AML
   subtype (FAB M0) and the relatively better prognosis subtype (FAB M2),
   respectively (Figure [87]S2B, Supporting Information). Depletion of
   ANXA1 using knockdown shRNA technology significantly reduced cell
   proliferation in both KG1a and HL60 cells (Figure [88]1C). To verify
   the functional specificity of ANXA1‐knockdown by shRNA, overexpression
   of ANXA1 in ANXA1‐knockdown KG1a and HL60 cells was shown to promote
   cell proliferation (Figures [89]1D and [90]S2C, Supporting
   Information).

Figure 1.

   Figure 1
   [91]Open in a new tab

   The expression level of ANXA1 is upregulated in AML patients, which
   promotes the proliferation of AML cells by down‐regulating Notch1
   expression. (A) mRNA expression levels of AML biomarkers in bone marrow
   mononuclear cells from AML patients (n = 422). (B) ANXA1 mRNA
   expression levels in peripheral blood or bone marrow samples of AML
   patients (n = 648) and non‐leukemia people (n = 274) from [92]GSE122505
   dataset. Student's t‐test, ^*** p < 0.001. (C) The cell proliferation
   ability detected by cell counting after 96 h of culture of KG1a or HL60
   cells transiently transfected with shANXA1 or shControl plasmids. (D)
   The cell proliferation ability detected by cell counting after 96 h of
   culture of KG1a‐ or HL60‐ shANXA1 cells transiently transfected with
   ANXA1 overexpression or pCDH control plasmids. (E) Gene set enrichment
   analysis of proteomic data from shANXA1 cells and shControl cells of
   KG1a. p < 0.001. (F) Western blotting of Notch1 and ANXA1 expression in
   KG1a and HL60 cells stably transfected with shANXA1 or shControl
   plasmids. Data in (C, D and F) are means ± s.e.m. of three independent
   experiments, t‐test, ^** p < 0.01, ^*** p < 0.001.

   To investigate potential molecular mechanisms underlying ANXA1‐induced
   hyperproliferation in AML cells, quantitative proteomics was performed.
   The results of gene set enrichment analysis showed that the Notch
   signal pathway was enriched in cells with stable knockdown of ANXA1
   (Figure [93]1E). Further, we confirmed that knockdown of ANXA1
   significantly increased the expression levels of Notch1 in KG1a and
   HL60 cells (Figure [94]1F).

2.2. Notch1 Interacts Directly with ANXA1 on the Cell Membrane

   Little is known about the regulators involved in upstream control of
   Notch1 function. We hypothesized that inhibition of Notch1 by ANXA1
   might entail direct physical interaction between these two proteins on
   the cell membrane. Prediction by AlphaFold2 (Figure [95]2A) and ClusPro
   (Figure [96]2B) showed the possibility of the interaction between ANXA1
   and Notch1, especially in the C‐terminal domain of ANXA1 and the
   ankyrin domain of NICD. Then, we performed laser confocal microscope
   analyses to evaluate this interaction, and we found that ANXA1 was
   co‐localized with Notch1 on the cell membrane, but co‐localization was
   lost after knocking‐down ANXA1, which was accompanied by an increased
   nuclear translocation of NICD in KG1a and HL60 cells (Figures [97]2C
   and [98]S3 and Table [99]S1, Supporting Information).

Figure 2.

   Figure 2
   [100]Open in a new tab

   The 2,050–2,060 amino acids domain of NICD mediates the direct
   interaction with the C‐terminal 214–228 amino acid domain of ANXA1 on
   the cell membrane. (A) Model diagram of the interaction between NICD
   and ANXA1 predicted by AlphaFold2. (B) Model diagram of the docking
   prediction of NICD ankyrin domain (PDB: 1YYH) with ANXA1 (PDB: 1MCX) by
   ClusPro. (C) Representative images and profile analyses of Notch1 and
   ANXA1 colocalization in KG1a and HL60 cells or shANXA1 cells taken by
   laser scanning confocal microscope. Orange squares indicate regions of
   interest for Manders overlap coefficient analyses and white solid lines
   indicate regions of interest for profile analyses. The arrows indicate
   the colocalization and scale bar represent 25 µm. (D) Schematic diagram
   of BLI analysis. (E) BLI analysis of the interactions between NICD and
   indicated ANXA1 or C‐terminal of ANXA1. (F) Schematic diagram of the
   corresponding positions of the designed peptides, including the model
   diagram of the ankyrin domain of NICD (PDB: 1YYH). (G) 3D structural
   model of the NICD‐5l peptide and NICD‐5c peptide designed according to
   2029–2060 amino acids of ankyrin domain of NICD (PDB: 1YYH). 3D
   structural model of the TAT‐NICD‐5l peptide was predicted by
   AlphaFold2. “l” for linear peptides and “c” for cyclic peptides. (H)
   Peptide sequences of NICD‐5l, NICD‐5c, TAT‐NICD‐5l and TAT. (I) BLI
   analysis of the interactions between ANXA1 and indicated NICD‐5l
   peptide or NICD‐5c peptide. (J) Mass spectrometry results of the
   interaction region between ANXA1 and NICD identified by chemical
   cross‐linking coupled with mass spectrometry. (K) The cell
   proliferation viability of KG1a and HL60 cells was detected by CCK8
   assay after treatment with NICD‐5l peptide and NICD‐5c peptide for 72
   h, or TAT‐NICD‐5l peptide and TAT peptide for 24 h, at indicated
   concentration. Data are means ± s.e.m. of three independent
   experiments, one‐way ANOVA with Dunnett's test, compared with 0 µM
   group, ^* p < 0.05, ^*** p < 0.001.

   To further localize the domain of ANXA1 in mediating this interaction,
   we expressed and purified a GST‐tagged fusion protein of full‐length
   ANXA1, N‐terminal, and C‐terminal domains of ANXA1, by using domain
   mapping (Figure [101]S4A,B, Supporting Information). The protein
   binding experiments in vitro showed that Notch1 in the cell lysate
   could be pulled down by GST‐tagged ANXA1 (Figure [102]S4C, Supporting
   Information). In addition, more Notch1 was pulled down in sgANXA1 cell
   line than control cell line (Figure [103]S4C, Supporting Information),
   which was consistent with the results shown in Figure [104]1F. The
   results using the bio‐layer interferometry (BLI) analyses
   (Figure [105]2D) finally also showed that both full‐length and the
   C‐terminal domain of ANXA1 dose‐dependently increased the binding
   affinity with NICD (Figure [106]2E). Analyses of the binding kinetics
   yielded a K[D] of 1.84 × 10^−5 M for full‐length ANXA1 and 5.32 × 10^−6
   M for the C‐terminal domain of ANXA1 (Table [107]S2, Supporting
   Information). These results provide further evidence for the direct
   interaction between NICD and the C‐terminal domain of ANXA1.

   The ankyrin domain of NICD contains seven ankyrin repeats, in which the
   first ankyrin repeat is not involved in protein folding.^[ [108]^18 ^]
   To identify which region of the ankyrin domain‐mediated the
   interaction, we designed and synthesized peptides of 2–7 ankyrin
   repeats by solid‐phase peptide synthesis approach (Figures
   [109]S15–S31, Supporting Information), including six linear peptides
   and six cyclic peptides, which were designed according to their crystal
   conformations (Figures [110]2F–H and [111]S5A and Table [112]S3,
   Supporting Information). The BLI analyses were used to further
   characterize the interactions between ANXA1 and synthesized peptides of
   the ankyrin domain of NICD. Only NICD‐5l peptide and NICD‐5c peptide
   increased the binding affinity with GST‐ANXA1 in a dose‐dependent
   manner (Figures [113]2I and [114]S5B and Table [115]S2, Supporting
   Information), suggesting that the 2029–2060 amino acid domain of NICD
   was the region mediating the interaction with ANXA1. Furthermore,
   2050–2060 amino acid domain of NICD identified by chemical
   cross‐linking coupled with mass spectrometry may be more critical for
   interaction with the 214–228 amino acid domain of ANXA1
   (Figure [116]2J). Consistent with this, ANXA1 (Δ214–228) failed to
   restore proliferative activity in sgANXA1 cells, further the
   interaction between NICD and ANXA1 was also lost due to the deletion of
   2050–2060 amino acid domain of NICD (Figure [117]S6, Supporting
   Information). Interestingly, both NICD‐5l peptide and NICD‐5c peptide
   significantly inhibited the proliferation of KG1a and HL60 cells in a
   dose‐dependent manner (Figure [118]2K). In contrast, other peptides for
   2–7 ankyrin repeats of NICD appeared no obvious role in the
   proliferation of AML cells (Figure [119]S7, Supporting Information).
   Surprisingly, conjugation of the cell‐penetrating peptide TAT
   (RKKRRQRRR, derived from HIV‐1) to NICD‐5l (named as TAT‐NICD‐5l)
   peptide greatly enhanced its osmotic capacity and cruelly destroyed the
   proliferation and colony forming ability of KG1a and HL60 cells,
   although TAT itself barely caused cell damage (Figures [120]2K and
   [121]S8, Supporting Information). Overall, it is unexpected and
   surprising that ANXA1 was identified as a novel upstream regulator of
   Notch1 through direct interaction on cell membrane, and that the
   interaction between 2050 and 2060 amino acid domain of NICD and the
   C‐terminal 214–228 amino acid domain of ANXA1 inhibited the nuclear
   translocation of NICD.

2.3. Modulation of Notch1 Function by ANXA1 Through the Promoting of NICD
Degradation via the Ubiquitin‐Proteasome System

   To determine how ANXA1 inhibited the expression level of NICD, the
   proteasome inhibitor MG132 was used to block the proteasome degradation
   system. We found that pretreating AML cells with MG132 significantly
   reversed the ANXA1‐induced NICD degradation after 3 h of treatment in
   the control group, while MG132 minimally affected the expression level
   of NICD in the ANXA1 knockdown group (Figure [122]3A,B). We further
   immunoprecipitated NICD and found that ubiquitin was bound to NICD
   significantly in the shControl cells after MG132 treatment rather than
   ANXA1 knockdown cells (Figure [123]3C). WWP2 is a recently discovered
   E3 ligase of NICD.^[ [124]^19 ^] Knocking down WWP2 can significantly
   restore the expression level of NICD in KG1a and HL60 cells
   (Figure [125]3D). The results showed that ANXA1 promoted the proteasome
   degradation of NICD by inducing ubiquitination, thereby inhibiting the
   activation of Notch1 signaling.

Figure 3.

   Figure 3
   [126]Open in a new tab

   ANXA1 promotes the proteasome degradation of NICD and down‐regulates
   the expression level of p15. (A) Western blotting of NICD expression in
   KG1a and HL60 cells stably transfected with shANXA1 or shControl
   plasmids at 0, 3 and 6 h after MG132 treatment. (B) Statistical graph
   of NICD expression after MG132 treatment. Data are means ± s.e.m. of
   three independent experiments, one‐way ANOVA with Dunnett's test,
   compared with 0 h group, ^* p < 0.05, ^** p < 0.01, ^*** p < 0.001. (C)
   Western blotting of NICD‐bound ubiquitin in immunoprecipitation
   complexes of KG1a‐ and HL60‐ shANXA1 or shControl cells treated with
   MG132 for 0 or 6 h. (D) Western blotting of WWP2 and NICD expression in
   KG1a and HL60 cells stably transfected with shWWP2 or shControl
   plasmids. (E) Biological pathway enrichment analysis of proteomic data
   from shANXA1 and shControl cells of KG1a. (F) Cell cycle analysis of
   KG1a and HL60 cells stably transfected with shANXA1 or shControl
   plasmids. Data are means ± s.e.m. of three independent experiments,
   t‐test, ^* p < 0.05. (G) Western blotting of cyclin‐dependent kinase
   inhibitors in KG1a and HL60 cells stably transfected with shANXA1 or
   shControl plasmids. (H) Statistical graph of cyclin‐dependent kinase
   inhibitors expression. Data are means ± s.e.m. of three independent
   experiments, t‐test, compared with shControl group, ^** p < 0.01.

2.4. Ablation of the Notch1‐p15‐Mediated Tumor Suppression by ANXA1 Mediates
the Proliferation of AML Cells

   To identify the downstream molecular mechanisms involved in the direct
   interaction between ANXA1 and Notch1 to promote the proliferation of
   AML cells, we analyzed proteomics data as shown in Figure [127]1E.
   Biological pathway enrichment analysis of the proteomics data indicated
   that knockdown of ANXA1 in AML cells induced a significant enrichment
   in the progression of the cell cycle (Figure [128]3E). Consistent with
   the proteomics results, the results of flow cytometry validated the
   significant arrest of G0/G1 phase distribution in ANXA1 knockdown cells
   (Figures [129]3F and [130]S9, Supporting Information), accompanied by a
   significant increase of the cyclin‐dependent kinase (CDK) inhibitor,
   p15, instead of p16, p21, and p27 (Figure [131]3G,H).

   To determine whether the increased expression of p15 induced by ANXA1
   knockdown was mediated by NICD signals, we constructed stable NICD
   over‐expression and Notch1 knockdown cell lines by lentivirus
   infection. We found that p15 was up‐regulated or down‐regulated
   synchronously with the expression level of NICD (Figure [132]4A), while
   over‐expression of NICD significantly decreased the cell proliferation
   of AML cells (Figure [133]4B). Furthermore, we knocked down the
   expression levels of Notch1 and p15 in the shANXA1 cell line (Figure
   [134]S10, Supporting Information), which significantly recovered cell
   proliferation inhibited by knockdown of ANXA1 expression
   (Figure [135]4C). As the release of NICD depends on the cleavage of
   Notch1 by γ‐secretase.^[ [136]^20 ^] We blocked the cleavage by
   RO4929097,^[ [137]^21 ^] an inhibitor of γ‐secretase which inhibited
   the expression of p15 in KG1a and HL60 cells with ANXA1 knockdown
   (Figure [138]4D), thereby promoting the proliferation of AML cells
   (Figure [139]4E), showing that the transcription activation of p15
   depended on the nuclear translocation of NICD.

Figure 4.

   Figure 4
   [140]Open in a new tab

   p15 is activated by NICD to directly inhibit cell proliferation. (A)
   Western blotting of Notch1 and p15 in KG1a and HL60 cells stably
   transfected with NICD over‐expression or shNotch1 plasmids. (B) The
   cell proliferation ability detected by cell counting after 96 h of
   culture of KG1a or HL60 cells stably transfected with NICD
   over‐expression or pLVX control plasmid. Data are means ± s.e.m. of
   three independent experiments, t‐test, ^* p < 0.05, ^** p < 0.01. (C)
   The cell proliferation ability detected by cell counting for
   KG1a‐shANXA1 and HL60‐shANXA1 cells transfected with shNotch1, shp15 or
   shControl plasmids. Data are means ± s.e.m. of three independent
   experiments, one‐way ANOVA with Dunnett's test, compared with shControl
   group, ^* p < 0.05, ^** p < 0.01. (D) Western blotting of p15 in KG1a
   and HL60 cells stably transfected with shANXA1 or shControl plasmids
   after RO4929097 treatment for 24 h. (E) The cell proliferation ability
   of KG1a‐shANXA1 and HL60‐shANXA1 cells was detected by cell counting
   after RO4929097 treatment for 24 h. Data are means ± s.e.m. of three
   independent experiments, one‐way ANOVA with Dunnett's test, compared
   with 0 µm group, ^* p < 0.05, ^** p < 0.01, ^*** p < 0.001. (F)
   Schematic diagram of a predicted binding model of NICD/RBPJ complex on
   the promoter of CDKN2B. (G) The site with the highest sequence
   frequency among the transcription factor binding sites predicted by
   JASPAR. (H) Relative luciferase activities of CDKN2B promoter‐reporter
   constructs in KG1a cells stably transfected with shANXA1 or NICD
   over‐expression plasmids. Data are means ± s.e.m. of three independent
   experiments, t‐test, ^*** p < 0.001.

   NICD as an interacting protein of the transcription factor RBPJ, which
   indirectly regulates the activation promoter of target genes.^[ [141]^7
   ^] To identify the potential binding site of the NICD/RBPJ complex
   interacting with the promoter region of CDKN2B (the gene encoding p15),
   the JASPAR database was used to predict transcription factor binding
   sites of RBPJ on the CDKN2B promoter, in which the TGGGAA site
   possessed the highest sequence frequency (Figure [142]4F,G). The
   luciferase reporter assay was subsequently used to determine the
   promotor gene activity of CDKN2B. We found that both knockdown of ANXA1
   or overexpression of NICD significantly enhanced the activity of the
   CDKN2B promoter (Figure [143]4H). Together, the results identified
   CDKN2B as a novel downstream target gene regulated by NICD binding on
   the promotor of p15, which modulated cell proliferation.

2.5. A Mutual Exclusive Relation Exists Between ANXA1 and Notch1/p15 in Human
Patients

   In order to further corroborate our mechanistic discovery in clinical
   AML patients, 24 AML patients and 16 volunteers (Table [144]S4,
   Supporting Information) were enrolled to detect the expression levels
   of ANXA1 using immunocytochemical staining (Figures [145] 5A and
   [146]S11A, Supporting Information). Positive immunostaining for ANXA1
   protein was detected in 95.83% of AML patients, including 25% AML
   patients with weak expression, 41.67% with moderate expression, and
   29.17% with strong expression, while there were just only 6.25% of
   volunteers with weak expression of ANXA1 (Figure [147]5B). Using the
   same patients’ samples, our results showed that Notch1 and p15 were
   significantly overexpressed in volunteers, but rarely expressed in AML
   patients (Figures [148]5C–F and [149]S11B,C, Supporting Information). A
   combined autocorrelation method also demonstrated that ANXA1 was
   significantly overexpressed in AML patients, while Notch1 and p15 were
   significantly overexpressed in volunteers (Fisher's exact test, p <
   0.0001, Table [150]S5, Supporting Information). Combining the
   immunocytochemistry scores and the correlation indexes using Fisher's
   exact test, we found that the expression level of ANXA1 was
   significantly and negatively correlated with the expression levels of
   Notch1 and p15, respectively. While the expression level between Notch1
   and p15 was positively correlated (Figure [151]5G,H; Table [152]S5,
   Supporting Information).

Figure 5.

   Figure 5
   [153]Open in a new tab

   The expression level of ANXA1 was up‐regulated in peripheral blood
   leukocytes of AML patients, which was negatively correlated with the
   expression levels of Notch1 and p15. (A, C and E) Representative
   immunocytochemical images of ANXA1 (A), Notch1 (C) and p15 (E) ‐stained
   peripheral blood from AML patients and volunteers. The red and blue
   arrows indicate positive and negative stained white blood cells,
   respectively. Scale bar, 50 µm. (B, D, and F) Pie chart overview of
   ANXA1 (B), Notch1 (D) and p15 (F) immunostaining scores in AML samples
   (n = 24) and volunteers’ samples (n = 16). (G) Allred intensity score
   statistics for the immunocytochemical images of Figure [154]S11
   (Supporting Information). Student's t‐test, ^*** p < 0.001. (H) Linear
   regression analyses among immunostaining intensity scores of ANXA1,
   Notch1 and p15, p < 0.001.

2.6. Targeting the ANXA1‐Notch1 Signaling Provides an Effective Approach for
Treatment of AML In Vivo and Ex Vivo

   We further assessed the efficacy of ANXA1 in vivo. We have established
   an in vivo model of KG1a‐shControl and KG1a‐shANXA1 cells in BALB/c
   nude mice upon subcutaneous injection to form solid tumors that allow
   quantitative tumor volume measurements for the study of tumor growth in
   vivo. Tumors were clearly observed after 5 days of subcutaneous
   inoculation (defined as day 0) of 9 × 10^6 cells in 5‐week‐old female
   BALB/c mice, then the tumor volumes were measured every 4 days for
   another 32 days (Figure [155]6A). On day 4, all 6 mice in the shControl
   group developed tumors with the volume of 290.21 ± 39.25 mm^3, while
   only 3 mice in the shANXA1 group developed tumors with the volume of
   126.67 ± 31.98 mm^3 (Figure [156]6B), and the body weight of mice in
   the two groups started to exhibit significant variances
   (Figure [157]6C). Subsequently, the tumors in the shControl group
   continued to grow and the body weight began to lose after day 8, while
   the tumors in the shANXA1 group grew significantly slower than those in
   the shControl group and were still present in only two mice after day
   24, and the body weight of mice in the shANXA1 group was significantly
   increased (Figure [158]6B–E). On day 32, the tumor volume of shControl
   group was 1945.31 ± 110.11 mm^3, and the tumor volume of shANXA1 group
   was 684.70 ± 179.60 mm^3, we immediately executed all mice humanely
   (Figure [159]6B). Also, we found the protein abundance of ANXA1 in the
   tumors of shANXA1 group was significantly decreased, followed by an
   obvious increase of the expression levels of Notch1 and p15 protein
   (Figure [160]6F). Moreover, we also established an in vivo model in
   B‐NDG mice by tail vein injection of KG1a‐shControl and KG1a‐shANXA1
   cells, respectively, in order to simulate human AML (Figure [161]6G).
   Compared with the shControl group, the transplantation levels of AML
   cells represented by hCD45^+ cells in the bone marrow and peripheral
   blood were significantly reduced in shANXA1 cells (Figure [162]6H,I),
   and the survival of the mice was significantly prolonged
   (Figure [163]6J), along with the significantly increase of body weight
   in shANXA1 group after day 24 (Figure [164]6K). We analyzed the organs
   of the mice in each group and found that there was no significant
   difference in the size and weight of the heart, liver, spleen, lung,
   kidney, and brain (Figure [165]S12A,B, Supporting Information), and the
   percentage of hCD45^+ in the spleen also showed no significant
   difference (Figure [166]S12C, Supporting Information). Therefore, these
   results showed that ANXA1 promoted the progression of human AML in mice
   in vivo.

Figure 6.

   Figure 6
   [167]Open in a new tab

   Targeting ANXA1 inhibits the proliferation of AML cells in vivo and ex
   vivo. (A) Schematic diagram of the course of the subcutaneous xenograft
   tumors experiment. (B) Statistical chart of tumor volume in the
   subcutaneous xenograft mice (n = 6). (C) Statistical chart of body
   weight of the subcutaneous xenograft mice (n = 6). (D) Statistical
   chart of the tumor weight of the subcutaneous xenograft mice. (E)
   Images of the tumors isolated from shControl and shANXA1 groups of the
   subcutaneous xenograft mice. (F) Western blotting of ANXA1, Notch1 and
   p15 in tumors isolated from subcutaneous xenograft mice. (G) Schematic
   diagram of the course of AML mouse model experiment. (H) Representative
   images of flow cytometry analysis of the transplantation levels in bone
   marrow and peripheral blood. (I) The proportion of hCD45^+ cells in
   bone marrow and peripheral blood of AML mice. (J) Survival analysis of
   AML mice. Log‐rank test (n = 8), p = 0.0304. (K) Body weight analysis
   of AML mice (n = 8). (L) Schematic diagram of the course of the peptide
   therapy experiment in vivo. (M) Statistical chart of tumor volume in
   the subcutaneous xenograft mice (n = 8). (N) Images of the tumors
   isolated from the subcutaneous xenograft mice. (O) Statistical chart of
   the tumor weight of the subcutaneous xenograft mice. (P) Statistical
   chart of body weight of the subcutaneous xenograft mice (n = 8). (Q)
   The cell proliferation viability of bone marrow mononuclear cells and
   peripheral blood mononuclear cells from AML patients detected by
   luminescent assay after treatment with TAT‐ NICD‐5l peptide or TAT
   peptide for 24 h, at indicated concentration. Data in (B, C, D, I, K, M
   and O) are means ± s.e.m., t‐test, compared with control group, ^* p <
   0.05, ^** p < 0.01, ^*** p < 0.001. Data in (Q) are means ± s.e.m.,
   one‐way ANOVA with Dunnett's test, compared with 0 µm group, ^* p <
   0.05, ^** p < 0.01, ^*** p < 0.001.

   Furthermore, we evaluated the drug sensitivity of TAT‐NICD‐5l peptide
   in vivo and ex vivo. We established the subcutaneous solid tumor model
   using KG1a cells to evaluate the therapeutic effect of the TAT‐NICD‐5l
   peptide in vivo (Figure [168]6L). After 8 days of continuous
   injections, the tumor volumes in TAT‐NICD‐5l peptide treatment groups
   with 5 mg kg^−1 and 10 mg kg^−1 were 1081.13 ± 120.58 mm^3 and 465.86 ±
   54.41 mm^3, respectively, which was significantly reduced compared to
   control mice with 1434.54 ± 177.90 mm^3 (Figure [169]6M). The isolated
   tumors also showed significant dose‐dependent efficacy after
   TAT‐NICD‐5l peptide treatment (Figure [170]6N,O). There was no
   significant difference in body weight among the different groups
   (Figure [171]6P), indicating that TAT‐NICD‐5l peptide has no potent
   toxicity at the indicated dosages in vivo. Ex vivo experiment also
   showed that TAT‐NICD‐5l peptide had the ability to significantly
   inhibit the cell proliferation of bone marrow mononuclear cells and
   peripheral blood mononuclear cells isolated from AML patients
   (Figure [172]6Q).

   Based on the above results in vitro, in vivo, and ex vivo, we proposed
   a novel molecular mechanism for Notch1 in regulating cell proliferation
   and a new therapeutic paradigm targeting the ANXA1‐Notch1 signaling of
   human AML as shown in Figure [173]7 .

Figure 7.

   Figure 7
   [174]Open in a new tab

   Schematic diagram of ANXA1‐Notch1‐p15 signal axis regulating
   hyperproliferation in human AML cells. In healthy leukocytes, NICD is
   cleaved by γ‐secretase and translocated to the nucleus, to activate the
   transcription of the CDKN2B gene, which expresses p15 protein to arrest
   the cell cycle in G[0]/G[1] phase and then inhibits cell proliferation.
   In AML cells with abnormally high expression of ANXA1, 2050–2060 amino
   acids of NICD directly interacts with the C‐terminal 214–228 amino acid
   domain of ANXA1 on cell membrane. This interaction promotes
   ubiquitination of NICD by WWP2 and subsequent degradation of NICD by
   the proteasome, resultant of loss of p15 and hyperproliferation of AML
   cells. However, NICD inhibitory peptides competitively bind ANXA1 and
   restore Notch1 function.

3. Discussion

   AML is one of the most common hematological malignancies resulting from
   genetic and epigenetic alterations that perturb key processes like
   self‐renewal, proliferation, and differentiation.^[ [175]^22 ^] In the
   United States, an estimated 20 800 cases of AML are expected in 2024,
   and 11 600 deaths due to AML.^[ [176]^23 ^] Chemotherapy and
   hematopoietic stem cell transplantation are still the main treatment
   methods for AML. Although, the targeted therapy and immunotherapy have
   improved the efficacy and prognosis of AML, but the mortality of
   patients is still high and the 5‐year survival rate is still less than
   30%.^[ [177]^24 ^] Given the poor prognosis of AML, it is essential to
   improve our mechanistic understanding involved in the proliferation and
   development of AML.

   Notch signaling is evolutionarily conserved from Drosophila to Homo
   sapiens, which has been extensively investigated as a regulator of cell
   fate decisions in a variety of organisms and tissues. Dysregulation of
   the Notch signaling is involved in many hematological and solid
   malignancies.^[ [178]^5 , [179]^25 ^] Notch1 is a well‐characterized
   oncoprotein in T‐ALL, while Notch1 is believed to be a tumor suppressor
   in AML.^[ [180]^5 , [181]^26 ^] In T‐ALL, there are several reports
   proposing different molecular mechanisms involved in the leukemogenesis
   of Notch1. The DEP‐domain containing mTOR‐interacting protein is a
   direct Notch1 target, which promotes cell proliferation and survival in
   T‐cell leukemia.^[ [182]^27 ^] A Notch1‐driven MYC enhancer promotes
   oncogenesis and highlights the importance of the Notch1‐MYC regulatory
   axis as a therapeutic target for T‐ALL patients.^[ [183]^28 ^] Notch1
   regulates the expression of PTEN and the activity of the PI3K‐AKT
   signaling pathway of the leukemogenic program in T‐ALL.^[ [184]^29 ^]
   Similarly, NF‐κB is one of the major mediators of Notch1‐induced
   transformation, and the NF‐κB pathway is a potential target of future
   therapies for T‐ALL patients.^[ [185]^30 ^] In contrast with the
   comprehensive research of Notch1 in T‐ALL, the molecular mechanisms of
   Notch1 in inhibiting cell proliferation in myeloid leukemia remain
   incompletely understood. Moreover, little is known of the regulatory
   mechanisms involved in the upstream control of Notch1 function and
   activity.

   The expression level and function of ANXA1 vary according to tumor
   type. Previous reports showed the contrasting roles of ANXA1 as a tumor
   suppressor protein or an oncogenic protein in different tumors/cancers,
   or even in the same cancer types.^[ [186]^15 ^] In breast cancer cells,
   ANXA1 promotes metastasis and facilitates an epithelial‐mesenchymal
   transition‐like switch by enhancing TGF‐β/Smad signaling.^[ [187]^11c
   ^] In head and neck cancers, downregulation of ANXA1 results in
   phosphorylation of the epidermal growth factor receptor (EGFR) and
   activation of PI3K‐AKT signaling, in addition to reduced exosome
   production and release of phosphorylated EGFR, thereby inhibiting tumor
   progression.^[ [188]^31 ^] However, the expression level, function, and
   mechanisms of ANXA1 in myeloid leukemia are still unclear. At the
   beginning of the present study, we aimed to identify the role and
   mechanisms of ANXA1 during AML progression. We found that ANXA1 was
   overexpressed in AML patients, and that ANXA1 promoted the
   proliferation of AML cells. However, we excluded the classical
   mechanisms of ANXA1 involved in cancer progression reported in previous
   studies, such as the signal pathways of MAPK, NF‐κB, Hippo, and TGF‐β
   (Figure [189]S13, Supporting Information). Unexpectedly and
   surprisingly, we first identified ANXA1 as a novel upstream regulator
   of Notch1 by its direct protein‐protein interaction on cell membranes,
   to promote the degradation of Notch1 through the ubiquitin‐proteasome
   pathway. The inactivation of Notch1 has been considered to be
   associated with p53 mutation.^[ [190]^32 ^] However, in our present
   research, KG1a cells and HL60 cells are TP53 mutant and TP53 deficient,
   respectively, thus our study identified a novel mechanism of
   inactivation of Notch1 through the direct regulation by ANXA1.
   Phospholipids are present in high levels in p53 mutant cells, which may
   also influence the effect of ANXA1 through their interaction on the
   membrane.^[ [191]^33 ^] These raise the question of whether ANXA1 is a
   biomarker of p53 mutant malignancies.

   The N‐terminal domain of ANXA1 serves as a ligand of formyl peptide
   receptor (FPR) and triggers different signaling pathways in a
   dose‐dependent manner in human neutrophils, and the activity of
   N‐terminal ANXA1 peptides (Ac2‐26 and Ac2‐12) can be blocked by the FPR
   antagonists, Boc1 and Boc2.^[ [192]^34 ^] However, our present findings
   showed that neither the FPR antagonist nor the N‐terminal ANXA1 peptide
   Ac2‐26 affected the proliferation of AML cells (Figure [193]S14,
   Supporting Information), indicating that endogenous ANXA1 exerted the
   proliferative effect through the full‐length form or C‐terminal domain
   (from Pro^44 to Gly^344) in AML cells, mediated by somehow mechanisms
   instead of a ligand of FPR.^[ [194]^11b ^] Furthermore, our results
   showed that ANXA1‐Notch1 interaction contributed to the
   hyperproliferative effect of AML cells. The ANXA1‐Notch1 interaction
   sites were located at the C‐terminal domain of ANXA1, inhibiting Notch1
   activation by γ‐secretase. According to the crystalline structure, we
   designed and synthesized 12 functional peptides of the NICD ankyrin
   domain to characterize the peptide‐ANXA1 interactions using the BLI
   analyses, which was further reconfirmed by cross‐linking mass
   spectrometry experiments. Our mass spectrometry results showed that the
   2,050–2,060 amino acid domain of NICD directly interacted with the
   C‐terminal 214–228 amino acid domain of ANXA1. Furthermore, peptides
   designed on the basis of protein‐protein interface inhibited the
   proliferation of AML cell lines, bone marrow mononuclear cells and
   peripheral blood mononuclear cells of AML patients in a dose‐dependent
   manner.

   Many target gene systems can be activated by Notch1, such as
   transcriptional repressors (the HES1 and HEY1 families), oncogenic
   pathways (NF‐κB and MYC), cell cycle progression pathways (CCND1/3),
   inhibition of apoptosis (BCL2), and cell migration/metastasis (CCR7),
   depending on the cell type and ligand‐receptor interactions at the cell
   surface.^[ [195]^5 , [196]^25 ^] In the present study, cyclin‐dependent
   kinase inhibitor‐2B (CDKN2B, also known as p15) was identified as a
   novel target of Notch1 by binding to the promotor of p15, to arrest
   cell cycle progression and inducing the suppression of proliferation
   when ANXA1 was knocked‐down in AML cells. We also identified a mutually
   exclusive relation that existed between ANXA1 and Notch1/p15 in AML
   patient samples, which further confirmed that ANXA1 promoted the
   proliferation of AML cells through functional suppression of Notch1 and
   p15.

   In conclusion, we discovered a mutually exclusive relationship between
   ANXA1 and Notch1/p15 in the same AML patients. Furthermore, endogenous
   ANXA1 promoted the proliferation of AML cells in vitro and in vivo
   through the ablation of Notch1‐p15‐mediated tumor suppression. In
   detail, we revealed that the C‐terminal 214–228 amino acid domain of
   ANXA1 interacted directly with the amino acid residues 2050–2060 of
   NICD to inhibit the expression of p15. On the basis of these novel
   mechanistic findings, we reasonably speculate that targeting ANXA1
   would provide an effective approach for the treatment of AML. In
   support of this new therapeutic paradigm, we provided proof‐of‐concept
   data by antagonizing ANXA1 using NICD inhibitory peptides in vitro, in
   vivo and ex vivo. Thus, our present results provide new insights into
   the development of novel AML drugs, such as stapled peptides,
   peptidomimetics, small molecules, and antibodies, to treat AML based on
   the direct interaction of ANXA1 and Notch1 on the cell membrane.

4. Experimental Section

Cell Lines and Peptides

   Human acute myeloid leukemia KG1a and HL60 cell lines were obtained
   from the First Affiliated Hospital, Zhejiang University School of
   Medicine. The peptides were synthesized on Wang resin (Sunresin, Xi'an,
   China) by the solid phase technique. Details of these procedures are
   described in the Materials and Methods ([197]Supporting Information).

Plasmid Construction and Protein Expression

   The insertion sequences, vectors, and restriction sites of all plasmids
   are provided in the Materials and Methods ([198]Supporting
   Information). 293T cells were used to package lentivirus. BL21
   chemical‐competent E. coli cells were used to induce the expression of
   targeted proteins.

Protein‐Protein Interaction

   AlphaFold2 and ClusPro servers were used to dock NICD and ANXA1.^[
   [199]^35 ^] The predictions were tested by GST‐pull down assay,
   bio‐layer interferometry (BLI) analysis, and chemical cross‐linking
   coupled with mass spectrometry. Details of these procedures are
   described in Materials and Methods ([200]Supporting Information).

Mouse Models

   The animal experiment protocol was performed following the guide for
   the Care and Use of Laboratory Animals. Five‐week‐old female BALB/c
   nude mice with similar body weight were equally divided into two groups
   and subcutaneously inoculated with KG1a‐shControl or KG1a‐shANXA1 cells
   (9 × 10^6 cells per mouse) into the posterior back regions. The tumor
   volumes and body weights in each group were recorded at regular
   intervals. Six‐week‐old female B‐NDG mice were inoculated with
   KG1a‐shControl or KG1a‐shANXA1 cells (3 × 10^6 cells per mouse) via
   tail‐vein injection. Peripheral blood white cells, blast cells, and
   spleen cells were collected and stained with FITC‐hCD45 and PE‐mCD45
   antibodies to identify the proportion of tumor cells. For peptide
   therapy experiment in vivo, TAT‐NICD‐5l peptide was intratumorally
   injected once a day after tumors reached 150 mm^3. Details of these
   procedures are described in the Materials and Methods ([201]Supporting
   Information).

Statistical Analysis

   All statistical data are presented as the means ± standard error of the
   mean (s.e.m.). Student's t‐test, one‐way ANOVA, log‐rank test, simple
   linear regression analysis, and Fisher's exact test were used for
   statistical analyses. p values of ^* < 0.05, ^** < 0.01, and ^*** <
   0.001 were considered significant. Details of these procedures are
   described in the Materials and Methods ([202]Supporting Information).

Study Approval

   Clinical sample experiments were approved by the Ethics Committee of
   The First Affiliated Hospital of Zhejiang University School of Medicine
   (No. 2021‐601). BALB/c nude and B‐NDG mice were used for animal work,
   which was approved by the ethics committee of animal experiments at
   Zhejiang Sci‐Tech University (No. 20210908‐1).

Conflict of Interest

   The authors declare no conflict of interest.

Author Contributions

   G.S., X.W., Y.Z., and J.M. contributed equally to this work. C.F.
   served as the lead contact for conceptualization, methodology,
   supervision, project administration, funding acquisition, and
   writing—review & editing. Y.C. and J.K. contributed to writing—review &
   editing. G.S. handled investigation, validation, data curation, funding
   acquisition, and wrote the original draft. H.W. provided resources,
   investigation, and validation. X.W., Y.Z., J.M., L.W., Z.Y., Z.S.,
   S.Z., H.W., Y.L., H.H., J.L., T.Z., B.Y., N.W., T.C., X.G., and Y.J.
   were involved in investigation and validation.

Supporting information

   Supporting Information
   [203]ADVS-11-2409726-s001.docx^ (11.1MB, docx)

Acknowledgements