Hexokinase 3 enhances myeloid cell survival via non-glycolytic functions

Abstract

The family of hexokinases (HKs) catalyzes the first step of glycolysis, the ATP-dependent phosphorylation of glucose to glucose-6-phosphate. While HK1 and HK2 are ubiquitously expressed, the less well-studied HK3 is primarily expressed in hematopoietic cells and tissues and is highly upregulated during terminal differentiation of some acute myeloid leukemia (AML) cell line models. Here we show that expression of HK3 is predominantly originating from myeloid cells and that the upregulation of this glycolytic enzyme is not restricted to differentiation of leukemic cells but also occurs during ex vivo myeloid differentiation of healthy CD34⁺ hematopoietic stem and progenitor cells. Within the hematopoietic system, we show that HK3 is predominantly expressed in cells of myeloid origin. CRISPR/Cas9 mediated gene disruption revealed that loss of HK3 has no effect on glycolytic activity in AML cell lines while knocking out HK2 significantly reduced basal glycolysis and glycolytic capacity. Instead, loss of HK3 but not HK2 led to increased sensitivity to ATRA-induced cell death in AML cell lines. We found that HK3 knockout (HK3-null) AML cells showed an accumulation of reactive oxygen species (ROS) as well as DNA damage during ATRA-induced differentiation. RNA sequencing analysis confirmed pathway enrichment for programmed cell death, oxidative stress, and DNA damage response in HK3-null AML cells. These signatures were confirmed in ATAC sequencing, showing that loss of HK3 leads to changes in chromatin configuration and increases the accessibility of genes involved in apoptosis and stress response. Through isoform-specific pulldowns, we furthermore identified a direct interaction between HK3 and the proapoptotic BCL-2 family member BIM, which has previously been shown to shorten myeloid life span. Our findings provide evidence that HK3 is dispensable for glycolytic activity in AML cells while promoting cell survival, possibly through direct interaction with the BH3-only protein BIM during ATRA-induced neutrophil differentiation.

Fig. 1. HK3 expression was predominant in myeloid tissue and markedly induced upon myeloid maturation.

A mRNA expression (Cp values) of HK1–3 in PBMCs and specific isolated/activated subpopulations. 293 T cells were used as negative ctrl. (n = 2, two technical replicates each, three individual donors, mean ± SEM). B HK1–3 transcript expression pattern across the hematopoietic cell lineages (Mac. macrophage lineage, Gran. granulocytic lineage, MDCs myeloid dendritic cells, PDCs plasmacytoid dendritic cells, B B cell lineage, T T cell lineage, NK NK cell lineage). Data accessed on the Haemoshpere Database. C The Human Protein Atlas HK1–3 single-cell gene expression data in various cell types [35]. D N-fold relative expression of HK1–3 mRNA levels during 12 days of in vitro neutrophil (G-CSF) or monocyte/macrophage (M-CSF) differentiation of CD34⁺ HSPCs isolated from human cord blood (n = 2, two technical replicates each, four individual donors, mean ± SEM). E N-fold relative HK1–3 mRNA expression levels in HL60 cells after 6 days of either ATRA-induced neutrophil-like differentiation (n = 3) or Vitamin D₃ (VitD₃)-induced monocytic-like differentiation (n = 2). Data were expressed as mean ± SEM. Each experiment was performed with at least two technical replicates. F Relative transcript abundance as a percentage of total hexokinase expression in HL60 cells treated with either ATRA, VitD3, or DMSO control. Calculated as described here [56]. G, H Relative protein levels of HK2 and HK3 during 4 days of ATRA treatment (G) or 3 days of VitD₃ (H) as measured by luminescent quantification of HiBiT-tag expression (G: n = 3, 4 individual clones for HK3-HiBiT, 2 individual clones for HK2-HiBiT. Mean ± SEM, Statistical analysis: two-way ANOVA. H: n = 2, 1 clone each). Mann–Whitney U: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Fig. 2. HK3 is dispensable for glucose-6-phosphate generation and glycolysis.

A Hexokinase activity assay measuring G-6-P generation in HL60 and NB4 cell line panels at steady state and after 3 days of ATRA treatment (n = 3, 2 technical replicates each, mean ± SEM). B, C Seahorse analysis of HK2 and HK3 KO in HL60 and NB4 cells. Glycolytic rate assay (B) and mitochondrial stress test (C) measuring extracellular acidification rate as a function of glycolytic activity and oxygen consumption as a function of mitochondrial respiration, respectively (n = 3, 4 technical replicates each, mean ± SEM). Mann–Whitney U: *p < 0.05, **p < 0.01, ****p < 0.0001.

Fig. 3. Loss of HK3 significantly increases ATRA-induced cell death.

A Relative cell viability as measured by alamarBlue^® reduction capacity on day 2 and 4 of ATRA treatment. Normalized to untreated control (n = 2, 4 technical replicates each, mean ± SEM). B ATRA titration showing dose-dependent induction of cell death in HK3-null HL60 clone after 4 days of treatment (n = 2, 2 technical replicates each, mean ± SEM). C Flow cytometric analysis of viability in HL60 cell lines (DAPI exclusion) during 4 days of ATRA treatment (n = 5, 2 technical replicates each, mean ± SEM). D Flow cytometric analysis of apoptosis induction via AnnexinV staining during ATRA treatment of HL60 cell lines (n = 3, mean ± SEM). E Caspase −3/−7 activity in HL60 cell lines after 48 h of ATRA treatment (n = 2, 3–4 technical replicates each, mean ± SEM). F Flow cytometric analysis of HL60 cell viability (DAPI exclusion) upon 4 days of treatment of indicated cell lines with either DMSO control, ATRA alone, or ATRA in combination with pan-caspase inhibitor Q-VD-OPh at 10 or 20 µM (n = 2, 2 technical replicates each, mean ± SEM). G, H Flow cytometric analysis of viability on day 4 of ATRA treatment (n = 3, mean ± SEM) (G) as well as Caspase-3/−7 activity after 2 days of ATRA treatment (n = 2, 3 technical replicates each, mean ± SEM) (H) in NB4 HK3-null clone stably transduced with either GFP control, a GFP-P2A-HK3wt construct or a GFP-P2A-HK3 D542A mutant. GFP-expressing cells were FACS sorted prior to the experiment. Mann–Whitney U: **p < 0.01, ***p < 0.001, ****p < 0.0001.

Fig. 4. Loss of HK3 increases ROS and was correlated with DNA damage.

A Relative MFI of CM-H2DCFDA staining in HL60 cell lines on day 2 and 4 of ATRA treatment (n = 2, 2 technical replicates each, mean ± SEM). B NBT positive cell area after 4 days of ATRA treatment in HL60 cell lines (n = 2, 2 technical replicates each, mean ± SEM). C Representative pictures of nuclear yH2AX foci in HL60 cell lines after 3 days of ATRA treatment (n = 2). D Quantification of nuclear yH2AX foci measured on an InCell microscope in HL60 cell lines ±3 days of ATRA treatment (n = 2, 500 nuclei/well, ≥2 wells per condition, mean ± SEM). E Western Blot analysis of yH2AX levels in HL60 control and HK3-null cell clones upon 24 h of indicated treatments. NAC samples were preincubated in 3 mM NAC for 2 h prior to adding treatments. F Nuclear fractionation of HL60 cell line ±4 days of ATRA treatment. *unspecific band at ~80 kDa. Mann–Whitney U: *p < 0.05, **p < 0.01, ****p < 0.0001.

Fig. 5. RNAseq and ATACseq analysis reveal enrichment and increased accessibility of cell death and stress pathways in HK3-null AML cells.

A Gene ontology (GO) analysis of upregulated genes (HK3-null vs Cas9 control). B Gene set enrichment analysis (GSEA) represents the normalized enrichment score (NES) of indicated gene sets. C Heatmap of differentially expressed genes in DMSO and ATRA-treated HK3-null and control samples. D Gene ontology (GO) analysis of upregulated genes in ATRA-treated samples (HK3-null vs Cas9 control). E Gene set enrichment analysis (GSEA) represents the normalized enrichment score (NES) of indicated gene sets in HK3-null vs Cas9 control after ATRA treatment. F Heatmap of differentially regulated genes across RNAseq and ATACseq datasets among ATRA-treated samples of HK3-null and Cas9 control cells. G Gene ontology (GO) analysis of genes upregulated in both RNAseq and ATACseq analysis under ATRA treatment (HK3-null vs Cas9 control).

Fig. 6. HK3 interactome reveals interaction with BH3-only protein BIM.

A DESeq2 normalized counts and differential expression analysis of BIM transcript expression in HK3-null cells and control ±ATRA. Data represent the mean of three biological replicates. B Confocal imaging of proximity ligation assay (PLA) using either endogenously HiBiT-tagged HK2 or HK3 HL60 cell lines. Fluorescent foci indicate the colocalization of proteins. Primary antibodies used were mouse anti-HiBiT, combined with rabbit anti-VDAC or rabbit anti-BIM. Control: secondary antibodies only. C PLA quantification. Negative controls: First column, secondary antibodies only, second column HL60 not expressing a HiBiT-tag (n = 2, 2 technical replicates each, mean ± SEM). Anti-VDAC was used as a positive control for HK2 interactions. D Co-IP of HK3 and BIM. IP was performed in HL60 HK2-null cells treated with ATRA for 2 days using an anti-BIM antibody immobilized on Protein A/G magnetic beads. HK3, BCL-2, and BIM immunoblotting of flowthrough (FT) and pulled-down (IP) proteins are shown. BCL-2, a known BIM interacting protein, was used as a positive control. E NOXA western blotting of HL60 control, HK2- and HK3-null cell lines treated for 4 days with DMSO or ATRA. Quantification of NOXA protein expression of two independent experiments is shown below. NOXA expression was normalized to total protein and Cas9 control cell expression.