Fig 2: HSF1 is critical for maintenance of LSC self-renewal.a Engraftment of three different primary AML LSCs in the presence or absence of CRISPR-mediated HSF1KD (started 4 mice/AML). *p = 0.0404, **p = 0.0003, ***p = 0.0006. b ATP production in control or HSF1 knockdown AML cells (post-transplantation from a) measured by Seahorse assay (n = 3 independent replicates). AML-a: *p = 0.0046. **p = 0.0349; AML-b: *p = 0.0201. **p = 0.0124; AML-c: *p = 0.0325. **p = 0.0033; c The expression of HSP90, SDHC, and HSF1 proteins in SISU-102 treated mouse MLL-AF9 cells and the human AML cell lines MV4–11 and NOLM-1 (n = 3 independent replicates). ACTIN is used as a load control. d Cell growth of murine MLL-AF9 cells (*p = 1.6 × 10-5, **p = 3.1 × 10-5, ***p = 6.1 × 10-5) and human AML cell line MV4–11 (*p = 4.8 × 10-3, **p = 2.3 × 10-4, ***p = 1.6 × 10-4) and NOLM-1 (*p = 6.3 × 10-4, **p = 3.01 × 10-5, ***p = 6.38 × 10-5) in the presence of absence of SISU-102 (n = 3 independent replicates). e OCR in Hsf1fl/flcreER LSCs (*p = 8.87 × 10-31, **p = 1.09 × 10-7) or human AML cell line MV4–11 (*p = 4.89 × 10-13, **p = 4.97 × 10-12) and NOLM-1 (*p = 1.19 × 10-18, **p = 0.00028) treated with or without SISU-102 (n = 8 independent replicates). f Engraftment of human AML cell line MV4–11 (n = 5 mice/group, *p = 9.8 × 10-5) or three different primary AML cells in the presence or absence of SISU-102. AML-d: *p = 1.8 × 10-5, n = 4 mice/group, and AML-e: *p = 8.98 × 10-6, Con n = 4 mice, SISU-102, n = 5 mice; AML-f: *p = 0.0011, Con n = 5 mice, SISU-102, n = 4 mice. g Engraftment of human BM CD34+ HSPCs treated with (n = 4 mice) or without (n = 3 mice) SISU-102 (5 mg/kg, IP, daily). In (a, b, d–f), two-tailed t test was used, data are presented as mean values ± SEM. Source data are provided as a Source Data file.

Fig 3: HSF1 is a key regulator of oxidative phosphorylation in AML.For experiments (a–m), Hsf1fl/flcreER mouse MLL-AF9 LSPCs were treated with or without (Con) 200 nM 4-OHT for different times. a 2-NBDG uptake (n = 3 independent replicates). b Mitochondrial membrane potential (Mitotracker, n = 3 independent replicates). c Mitochondrial superoxide (Mitosox, n = 4 independent replicates), *p = 0.0017. d, e Mitochondrial ROS levels measured by H2-DCFDA (d, n = 4 independent replicates, *p = 0.0106) or CellROX (e, n = 3 independent replicates, *p = 0.0016). f, g OCR (f, *p = 5.9 × 10-18, **p = 4.3 × 10-11) and ECAR (g, *p = 5.3 × 10-7, **p = 8.8 × 10-8) measured by Seahorse (n = 8, independent replicates each). h ATP production (n = 3 independent replicates). *p = 3.57 × 10-5, **p = 0.000795. i Leukemia cell intermediate metabolites (n = 5 independent replicates) using high-resolution Gas chromatograph mass spectrometry (GC/MS). j 13C6-Glucose tracing (measuring leukemia cell intermediate metabolites) using GC/MS. Left, illustration of glucose metabolism. White circles, 12C carbons; blue circles, 13C carbons. M + 2 and M + 3 refer to the number of 13C carbons.; Right, TCA intermediates: citrate (*p = 0.00167), succinate (*p = 0.0121), fumarate, and malate (*p = 0.0011) and intracellular metabolites: pyruvate, alanine, lactate (*p = 0.0146), glutamate (*p = 0.00014), and aspartate (*p = 0.00098). Data are the isotopologue distribution relative to abundance and normalized to the control group, plotted as mean values and individual data points from n = 5 cultures. k Expression of key electron transport chain complex (ETC) I, II, III, and IV components by Western blotting (n = 3 independent replicates). l qPCR analysis of expression of Sdha (*p = 0.0008) and Sdhc (*p = 8.2 × 10-7) (n = 3 independent replicates). m ETC complex II activity (*p = 0.00575) (n = 3 independent replicates). n Expression of exogenous SDHC protein in Hsf1fl/flcreER LSPCs (n = 2 independent replicates). o, p Leukemia cell proliferation (o, *p = 0.00078) and colony formation (p, *p = 0.00024) in empty vector or SDHC transduced Hsf1fl/flcreER LSCs in the presence or absence of 200 nM of 4-OHT (n = 3 independent replicates, each). q OCR in empty vector or SDHC transduced Hsf1fl/flcreER LSPCs treated with or without 200 nM 4-OHT (n = 7 independent replicates). *p = 0.00066, **p = 2.2 × 10-10, ***p = 4.86 × 10-6. r ETC complex II activity in empty vector or SDHC transduced Hsf1fl/flcreER leukemia cells treated with or without (Con) 200 nM 4-OHT (n = 4 independent replicates). *p = 0.00056, **p = 0.0044. In (c–e, h, j, l–m, r), two-tailed t test was used, data are presented as mean values ± SEM. s Survival curve of sublethally irradiated recipient mice receiving vector or SDHC transduced Hsf1fl/flcreER LSCs treated with vehicle (n = 5 mice/group) or TAM (n = 9 mice/group). *p = 0.0001. Source data are provided as a Source Data file.

Fig 4: Fisetin blocks in vitro sumoylation of p53.In vitro sumoylation of (A,B) tumor suppressor protein p53 or (C,D) Heat Shock Factor 1 (HSF1) protein in the absence of (-, DMSO control) or presence of fisetin (+F, 300 μM dissolved in DMSO). Western blots were probed with (A,C) SUMO1, (B) p53 or (D) HSF1 antibody (Ab) as indicated. Sumoylated substrate proteins (red arrowheads), unmodified substrate proteins (open arrowheads), E1 or E2 SUMO1 conjugates (open circles) and SUMO1 (asterisk) are indicated. A SUMO1 conjugation-deficient mutant (mt) served as a negative control; NS, a no substrate control allows detection of E1 and E2-SUMO conjugates with the SUMO1 Ab. (A,B) The p53 substrate is visibly sumoylated only when fisetin is absent. No E1 or E2 conjugates are detected in the presence of fisetin. A dividing line between lanes indicates that the NS controls are from separate blots. (C,D) The HSF1 substrate is sumoylated in both the absence and presence of 300 μM fisetin. Western blots are representative of at least two independent experiments.

Fig 5: Expression of HSF1 might serve as a marker to monitor AML status.a HSF1 H-score index in AML samples at diagnosis, remission, or relapse (n = 39 primary AML samples). Box plots show median and first/third quartiles. ANOVA with LSD post-hoc analysis was used, *p = 0.0001. b–f Representative pictures showing 2 different HSF1 staining patterns in AML patients. b–c Pattern 1, HSF1 staining at remission (b) and follow up (c). d–f Pattern 2, HSF1 staining at remission (d), follow up (e) and relapse (f). Note that (b) and (d) are 2 weeks post chemotherapy, BM shows 5–10% of cellularity. The insets in (b) and (d), are HSF1 staining at AML diagnosis. Scale bar, 20 µM. Source data are provided as a Source Data file.