Chemical structures of parthenolide and compounds found through a gene expression based search of GEO. A query of GEO was performed using the gene expression signature of cells treated with parthenolide (left panel). The 4 anti–AML-SC compounds identified by the search were celastrol, 4-hydroxynonenal, 15Δ-prostaglandin J2, and MG-132 (right panel).

Biological characterization of terpenoids identified by gene expression analyses. (A) Bar chart of FACS analysis indicating significant impairment of viability (single asterisk, P < .01; N = 3 patient samples) of primary CD34⁺ CD38⁻ AML cells versus normal human CD34⁺ CD38⁻ bone marrow treated with 2 μM celastrol. Alongside, a representative FACS analysis of viability is shown. The lower left quadrant (annexin V negative/7-AAD negative) represents viable cells, whereas events in the lower right and upper right panels represent dying and dead cells, respectively. Viability is represented as a percentage of untreated controls. (B) Methylcellulose colony assays of primary human cells treated with 2 μM celastrol. Normal erythroid, myeloid, and AML colony forming units are shown relative to untreated controls. Error bars represent the standard error of the mean. Black bars are normal cells. White bars are AML. Significant selectivity (asterisk) is indicated by either the AML versus myeloid or AML versus erythroid comparisons with celastrol treatment (single asterisk, P < .01; N = 3 patient samples). Viability is represented as a percentage of untreated controls. (C) Bar charts indicate the percent engraftment of AML cells in NOD/SCID mice after 18 hours in culture with 2 μM celastrol (Cel) versus untreated control (Unt). White bars represent the cohort of animals injected with untreated cells; black bars represent 2 μM celastrol treatment. Significant loss of engraftment (P < .01, single asterisk; N = 3 patient samples, 3-4 mice/sample) in NOD/SCID mice is observed with celastrol treatment. No asterisk, P = .11 (4 mice). Overall P value for the treatment is P = .0003. (D) Bar chart indicating viability of phenotypically defined AML cells 24 hours after treatment with 20 μM gedunin (N = 3 patient samples). Viability is represented as a percentage of untreated controls. (E) Representative confocal microscopy images of primary CD34⁺ AML cells treated with 2 μM celastrol, 20 μM gedunin, or left untreated. Top panels show overlays labeling for NF-κB p65 (green) and nucleus (red). Bottom panels show overlays for Nrf2 (yellow) and nucleus (blue). (F) Top panel: EMSAs indicating NF-κB binding relative to untreated (UNT) cells with either 2 μM celastrol (CEL) or 20 μM gedunin (GED) 6 hours after treatment of CD34⁺ AML cells. Bottom panel: immunoblots indicating phospho-p65, HO-1, and β-actin levels at 6 hours in primary CD34⁺ AML treated with 2 μM celastrol (CEL) treatment, 20 μM gedunin (GED), or left untreated (UNT).

Biological characterization of nonterpenoids identified by gene expression analyses. (A) Bar chart of FACS analysis indicating significant impairment of viability (single asterisk, P < .01; N = 3 patient samples) of primary human CD34⁺ CD38⁻ AML cells versus normal CD34⁺ CD38⁻ marrow cells treated with 30 μM HNE. Alongside, a representative FACS analysis of viability is shown. The lower left quandrant (annexin V negative/7-AAD negative) represents viable cells, whereas events in the bottom right and top right panels represent dying and dead cells, respectively. Viability is represented as a percentage of untreated controls. (B) Methylcellulose colony assays of primary human cells treated with 30 μM HNE. Normal erythroid, myeloid, and AML colony forming units are shown relative to untreated controls. Error bars represent the standard error of the mean. Black bars are normal cells. White bars are AML. Significant selectivity (double asterisk) is indicated by either the AML versus myeloid or AML versus erythroid comparisons with HNE treatment (P < .001; N = 3 patient samples). Viability is represented as a percentage of untreated controls. (C) Bar charts indicate the percent engraftment of AML cells in NOD/SCID mice after 18 hours in culture with 30 μM HNE versus untreated control (Unt). White bars represent the cohort of animals injected with untreated cells; black bars represent 30 μM HNE treatment. Significant loss of engraftment (P < .001, double asterisk; P < .0001, triple asterisk; N = 3 patient samples, 4 mice/sample) in NOD/SCID mice is observed with HNE treatment. Overall P value for the treatment is P < .0001. (D) 50 μM hemin lacks toxicity to total (white bar), CD34⁺/CD38⁻ (gray bar), and CD34⁺ (shaded bar) primary human AML populations. Viability is represented as a percentage of untreated controls. (E) Representative confocal microscopy images of primary CD34⁺ AML left untreated or treated with 30 μM HNE or 50 μM hemin. Top panels show overlays labeling for NF-κB p65 (green) and nucleus (red). Bottom panels show overlays for Nrf2 (yellow) and nucleus (blue). (F) Top panel: EMSAs of NF-κB binding relative to untreated (UT) cells with either 30 μM HNE or 50 μM hemin at 6 hours after treatment in CD34⁺ AML. Bottom panel: immunoblots of phospho-p65, HO-1, and β-actin levels at 6 hours in primary CD34⁺ AML treated with 30 μM HNE treatment, 50 μM hemin, or left untreated (UT).