Homodimer-mediated phosphorylation of C/EBPα-p42 S16 modulates acute myeloid leukaemia differentiation through liquid-liquid phase separation

Abstract

CCAAT/enhancer binding protein α (C/EBPα) regulates myeloid differentiation, and its dysregulation contributes to acute myeloid leukaemia (AML) progress. Clarifying its functional implementation mechanism is of great significance for its further clinical application. Here, we show that C/EBPα regulates AML cell differentiation through liquid-liquid phase separation (LLPS), which can be disrupted by C/EBPα-p30. Considering that C/EBPα-p30 inhibits the functions of C/EBPα through the LZ region, a small peptide TAT-LZ that could instantaneously interfere with the homodimerization of C/EBPα-p42 was constructed, and dynamic inhibition of C/EBPα phase separation was observed, demonstrating the importance of C/EBPα-p42 homodimers for its LLPS. Mechanistically, homodimerization of C/EBPα-p42 mediated its phosphorylation at the novel phosphorylation site S16, which promoted LLPS and subsequent AML cell differentiation. Finally, decreasing the endogenous C/EBPα-p30/C/EBPα-p42 ratio rescued the phase separation of C/EBPα in AML cells, which provided a new insight for the treatment of the AML.

Fig. 1. C/EBPα undergoes an abnormal particle alteration during cell differentiation.

a Experimental schematic for the establishment of the Dox/Ara-C (D/A) treated GFP + MLL-AF9 mouse model. b Representative photographs and weights of spleens isolated from the groups of GFP + MLL-AF9 mice at 7 days after treatment (n = 5 mice per group). A two-tailed unpaired Student’s t test was used for statistical analysis and data were presented as mean values  ±  SEM. c Flow cytometric analysis of GFP + MLL-AF9 leukaemia cells in BM of the control and D/A groups (n = 5 mice per group). A two-tailed unpaired Student’s t test was used for statistical analysis and data were presented as mean values  ±  SEM. d Immunofluorescence photograph analysis of the number of endogenous C/EBPα condensates in each GFP + MLL-AF9 leukaemia cell in the control and D/A groups. Quantification of C/EBPα condensates (n =  3 biologically independent experiments). A two-tailed unpaired Student’s t test was used for statistical analysis, and data were presented as mean values  ±  SEM. e Immunofluorescence photograph analysis of the number of endogenous C/EBPα condensates in each CD34+ primary AML cell treated with the control or D/A (Ara-C:50 nM; Dox:0.2 μg/mL) for 24 h. Quantification of C/EBPα condensates (n =  6 AML patients). A two-tailed unpaired Student’s t test was used for statistical analysis, and data were presented as mean values  ±  SEM. f Immunofluorescence photograph analysis of the number of endogenous C/EBPα condensates in each THP-1 cell stimulated with 100 ng/mL PMA for 0 h, 3 h, 6 h, 12 h, 24 h and 48 h. Quantification of C/EBPα condensates (n =  3 biologically independent experiments). A two-tailed unpaired Student’s t test was used for statistical analysis, and data were presented as mean values  ±  SEM. The experiments in (d, f) were observed at least 20 random fields in each experiment. For (e), at least 10 random fields were observed in each experiment. For (d, e), scar bars, 5 μm. For (f), scar bar, 2 μm. Source data are provided as a Source Data file.

Fig. 2. C/EBPα undergoes LLPS.

a Analysis of disordered regions in C/EBPα. IUPred was used to analyse the regions of C/EBPα; Regions with a score of greater than 0.5 (above the dotted line) were considered disordered. b Turbidity of purified EGFP-C/EBPα protein (20 μM) increased after treatment with crowding agents, while EGFP protein did not. c EGFP-C/EBPα (20 μM) droplets were formed after treatment with 250 mM NaCl and 10% PEG 8000 and then disappeared after 2 min addition of 10% 1,6-Hex, while EGFP did not exhibit these properties. d Representative FRAP images of EGFP-C/EBPα droplets. The dotted circle shows the bleached area. The FRAP curve of EGFP-C/EBPα droplets was analysed with different droplets (n = 8). Data are presented as mean values ±  SEM. e Fusion of EGFP-C/EBPα droplets was imaged in the fluorescence and DIC channels (n =  3 biologically independent experiments). f C/EBPα formed puncta in HEK 293 T living cells transfected with C/EBPα-EGFP after 0.1 M mannitol treatment and then disappeared after 2 min the addition of 10% 1,6-Hex (n =  3 biologically independent experiments). g Quantification of C/EBPα puncta in Fig. 2f (n = 10 cells). A two-tailed unpaired Student’s t test was used for statistical analysis, and data were presented as mean values  ±  SEM. h Representative FRAP images of C/EBPα puncta in HEK 293 T living cells transfected with C/EBPα-EGFP after 0.1 M mannitol treatment. The white arrow indicated bleached area. The FRAP curve of C/EBPα-EGFP droplets was analysed with different droplets (n = 8 cells). Data are presented as mean values ±  SEM. i Fusion of C/EBPα-EGFP droplets in the HEK 293 T living cells transfected with C/EBPα-EGFP after 0.1 M mannitol treatment was imaged (n =  3 biologically independent experiments). For (c), scar bar, 5 μm. For (d–f, h, i), scar bars, 2 μm. Source data are provided as a Source Data file.

Fig. 3. C/EBPα-p30 inhibits the LLPS of C/EBPα.

a Domain structure analysis of C/EBPα and C/EBPα-p30. b Purified EGFP-C/EBPα-p30 protein (20 μM) showed no turbidity. c EGFP-C/EBPα-p30 protein (20-40 μM) droplets did not form after the addition of 250 mM NaCl and 10% PEG 8000 (n =  3 biologically independent experiments). d No C/EBPα-p30-EGFP puncta was observed after 0.1 M mannitol treatment in HEK 293 T living cells transfected with C/EBPα-p30-EGFP (n =  3 biologically independent experiments). e The diagram of the interaction between C/EBPα and C/EBPα-p30 through the LZ domain. f FLAG-tagged C/EBPα-p30 or C/EBPα-p30ΔLZ was constructed into HEK 293 T cells together with C/EBPα-EGFP. The interaction was analysed by co-immunoprecipitation with anti-FLAG and anti-GFP antibodies followed by western blotting (WB) (n =  3 biologically independent experiments). g Live-cell images indicated that the number of C/EBPα-EGFP puncta in mannitol-treated HEK 293 T cells over-expressed with C/EBPα-EGFP to mCherry-tagged C/EBPα-p30 or C/EBPα-p30ΔLZ at a mass ratio of of 1:9. Quantification of C/EBPα-EGFP puncta (n =  3 biologically independent experiments). A two-tailed unpaired Student’s t test was used for statistical analysis, and data were presented as mean values  ±  SEM. The experiments in (g) were observed at least 20 random fields in each experiment. For (c, g), scar bars, 5 μm. For (d), scar bar, 2 μm. Source data are provided as a Source Data file.

Fig. 4. The LZ fragment inhibits the LLPS of C/EBPα-p42 by disrupting the formation of its homodimers.

a Native gel electrophoresis show the the existence of dimer (n =  3 biologically independent experiments). The 0.8 μg of mCherry-tagged C/EBPα-p30 or LZ plasmids was constructed into HEK 293 T cells together with 1.2 μg of C/EBPα plasmids. Blue, pink, Black, red and purple arrows indicated the mCherry-C/EBPα-p30 homodimers, mCherry-C/EBPα-p30:C/EBPα-p42 heterodimers, C/EBPα-p42 homodimers, mCherry-LZ:C/EBPα-p42 heterodimers and mCherry-LZ homodimer respectively. “**” and “*” indicated the residual C/EBPα-p42 homodimers and C/EBPα-p42 after anti-stripping treatment; b mCherry-tagged LZ and C/EBPα-EGFP plasmids were co-constructed into HEK 293 T cells. The interaction was analysed by co-immunoprecipitation with anti-mCherry and anti-GFP antibodies (n =  3 biologically independent experiments). c The number of C/EBPα-EGFP puncta in mannitol-treated HEK 293 T cells transfected with C/EBPα-EGFP to mCherry-vector or LZ at a mass ratio of 1:9 (n =  3 biologically independent experiments). d Diagram of the small peptide TAT-LZ structure. e Changes in C/EBPα-EGFP droplets after TAT-LZ was introduced into mannitol-treated HEK 293T- C/EBPα-EGFP living cells at −10 min, 0 min and 5 min. “x-axis” and “y-axis” indicated the time after mannitol treatment and the duration of TAT-LZ peptide pre-treatment respectively. f, Quantification of C/EBPα-EGFP puncta in Fig. 4e (n = 8 cells). g The endogenous C/EBPα condensates in THP-1 cells transduced with Ctrl or LZ under the 100 ng/mL PMA for 24 h. h Quantification of C/EBPα condensates for Fig. 4g (n =  3 biologically independent experiments). i The expression of CD11b and CD68 in THP-1 cells transduced with Ctrl or LZ lentivirus under the 100 ng/mL PMA for 48 h (n =  3 biologically independent experiments). j The protein level of the CD11b and CD68 in THP-1 cells transduced with Ctrl or LZ lentivirus under the 100 ng/mL PMA for 24 h (n =  3 biologically independent experiments). For (f, h–j), a two-tailed unpaired Student’s t test was used for statistical analysis, and data are presented as mean values ±  SEM. For (c, e), scar bars, 5 μm. For (g), scar bars, 2 μm. Source data are provided as a Source Data file.

Fig. 5. Phosphorylation of C/EBPα at S16 stabilizes its LLPS through homo-dimerization.

a Mass spectrometry analysis indicated the C/EBPα p-S16 in HEK 293T-C/EBPα-EGFP under 0.1 M mannitol treatment for 10 min. b Western blot showed the level of endogenous C/EBPα p-S16 in THP-1 cells treated with 100 ng/mL PMA for 0–24 h. The samples derive from the same experiment and that gels were processed in parallel. GAPDH was run on the same blot. Quantification of the relative p-S16 level (n =  3 biologically independent experiments). c Western blot analysis showed the level of C/EBPα p-S16 in THP-1 cells transfected with p30, p30-ΔLZ, and LZ plasmids together with C/EBPα-Flag at a mass ratio of 9:1 after 100 ng/mL PMA treatment for 24 h. The samples derive from the same experiment and that gels were processed in parallel. GAPDH was run on the same blot. Quantification of the relative p-S16 level (n =  3 biologically independent experiments). d Representative images of THP-1 living cells transfected with C/EBPα-S16E-EGFP, C/EBPα-EGFP or C/EBPα-S16A-EGFP with 100 ng/mL PMA. e Quantification of C/EBPα condensates for 4d (n =  3 biologically independent experiments); At least 20 random fields were observed in each experiment. f The expression of the CD11b and CD68 in THP-1 cells transfected with C/EBPα-S16E-EGFP, C/EBPα-EGFP or C/EBPα-S16A-EGFP with 100 ng/mL PMA (n =  3 biologically independent experiments). g The percent of CD11b+ and CD68+ in 100 ng/mL PMA-treated THP-1 cells transfected with C/EBPα-S16E-EGFP, C/EBPα-EGFP or C/EBPα-S16A-EGFP (n =  3 biologically independent experiments). h The RFP tagged control, C/EBPα-EGFP, C/EBPα-S16E, C/EBPα-S16A, C/EBPα-p30 lentiviruses were transferred into MLL-AF9 cells respectively and then injected into C57/BL6 mice. Representative photographs and weights of spleens isolated from RFP + GFP + MLL-AF9 mice in the five groups on day 10 (n = 3 per group). i Flow cytometric analysis of RFP+ leukaemia cells in BM of the five groups. Quantification of RFP+ leukaemia cells (n = 3 per group). For (b, c, e–i), a two-tailed unpaired Student’s t test was used for statistical analysis, and data are presented as mean values ±  SEM. For (e), scar bars, 5 μm. Source data are provided as a Source Data file.

Fig. 6. Decreasing the endogenous C/EBPα-p30/C/EBPα ratio rescues the function of C/EBPα through its LLPS.

a Western blot analysis showed the ratio of endogenous C/EBPα-p30 to C/EBPα in THP-1 cells treated with different concentrations of rapamycin. Relative C/EBPα-p30/C/EBPα levels (n =  3 biologically independent experiments). b mScarlet tagged control and LZ lentivirus were transducted into THP-1 cells. Immunofluorescence photograph analysed the number of endogenous C/EBPα condensates in THP-1-Ctrl or THP-1-LZ cells treated with PMA and/or PMA + RAPA (PMA + R) for 24 h. Quantification of the number of C/EBPα condensates (n =  3 biologically independent experiments). It was observed at least 20 random fields in each experiment. c mScarlet tagged control, LZ or C/EBPα-p30 lentivirus were transducted into THP-1 cells. qRT-PCR indicated the expression of CD11b and CD68 in THP-1-Ctrl, THP-1-LZ or THP-1-C/EBPα-p30 cells treated with PMA and/or PMA + RAPA (PMA + R) for 24 h (n =  3 biologically independent experiments). d Flow cytometry was performed to measure the level of the CD11b and CD68 in THP-1-Ctrl, THP-1-LZ or THP-1-C/EBPα-p30 cells treated with PMA and/or PMA + RAPA (PMA + R) for 24 h (n =  3 biologically independent experiments). e Experimental schematic for the establishment of the mScarlet+ GFP + MLL-AF9 mouse model. f Representative photographs and weights of spleens isolated from mScarlet+ GFP + MLL-AF9 mice in the four groups on day 14 (n = 5 per group). g Flow cytometric analysis of mScarlet+ leukaemia cells in BM of the four groups. Quantification of mScarlet+ leukaemia cells (n = 5 per group). h, Immunofluorescence photograph analysis of the number of endogenous C/EBPα condensates in the control, D/A, D/A + R and D/A + R + LZ groups (n = 5 per group). Quantification of endogenous C/EBPα condensates in each cell. For (b, h), at least 20 random fields were observed in each experiment. For (a–d, f–h, e–i), a two-tailed unpaired Student’s t test was used for statistical analysis, and data are presented as mean values ±  SEM. For (b, h), scar bars, 5 μm. Source data are provided as a Source Data file.

Fig. 7. Model for homodimer-mediated phosphorylation of C/EBPα-p42 modulating AML cell differentiation by LLPS.

Homodimer-mediated phosphorylation of C/EBPα-p42 enhances AML cell differentiation through LLPS. However, C/EBPα-p30 and LZ inhibit AML differentiation by destroying the dimerization of C/EBPα-p42 and eliminating its LLPS.