Dual inhibition of HSF1 and DYRK2 impedes cancer progression

Abstract

Preserving proteostasis is a major survival mechanism for cancer. Dual specificity tyrosine phosphorylation-regulated kinase 2 (DYRK2) is a key oncogenic kinase that directly activates the transcription factor heat-shock factor 1 (HSF1) and the 26S proteasome. Targeting DYRK2 has proven to be a tractable strategy to target cancers sensitive to proteotoxic stress; however, the development of HSF1 inhibitors remains in its infancy. Importantly, multiple other kinases have been shown to redundantly activate HSF1 that promoted ideas to directly target HSF1. The eventual development of direct HSF1 inhibitor KRIBB11 suggests that the transcription factor is indeed a druggable target. The current study establishes that concurrent targeting of HSF1 and DYRK2 can indeed impede cancer by inducing apoptosis faster than individual targetting. Furthermore, targeting the DYRK2-HSF1 axis induces death in proteasome inhibitor-resistant cells and reduces triple-negative breast cancer (TNBC) burden in ectopic and orthotopic xenograft models. Together the data indicate that cotargeting of kinase DYRK2 and its substrate HSF1 could prove to be a beneficial strategy in perturbing neoplastic malignancies.

Keywords: inhibition; myeloma; proteasomes; stress kinases.

Figure 1. HSF1 and DYRK2 expressions moderately overlap in tumour microenvironment cell states and therapy responses in various cancers

Individual RNA expressions of HSF1 and DYRK2 overlayed on the (A) UMAP distribution of renal cell carcinoma sc-RNAseq dataset, (B) tSNE distribution of colon adenocarcinoma sc-RNAseq dataset, (C) UMAP distribution of breast cancer sc-RNAseq dataset, (D) tSNE distribution of astrocytoma sc-RNAseq dataset. (E) The signed -log10 P-value coexpressions of HSF1 and DYRK2 shown in response to indicated drug combination treatment in the respective cancers. (F) The signed -log10 P-value coexpressions of HSF1 and DYRK2 shown in relation to the response toward individual cancer monotherapies. See also Supplementary Figure S1.

Figure 2. Dual loss of HSF1 and DYRK2 induces enhanced apoptosis in TNBC cells

(A) MDA-MB-231 and MDA-MB-468 cells were treated with or without the indicated concentrations of KRIBB11 for 16 h. Cells were lysed and immunoblotting was carried out with the indicated antibodies. (B) MDA-MB-231 parental or DYRK2 knock-out cells were treated with or without the indicated concentration of KRIBB11 for 24 h. Cells were lysed and immunoblotting was carried out with the indicated antibodies. (C) MDA-MB-468 cells were treated with either LDN192960 alone or KRIBB11 alone or the combination of both at the indicated concentrations for 72 h and cell viability was analysed by CellTiter 96® AQueous Non-Radioactive Cell Proliferation Assay kit. Data are represented as relative viability of DMSO-treated control. (The P-value provided is the least significant value comparing the combination vs. single-drug treatments. Two-way ANOVA with multiple comparison: Fisher’s LSD test). (D) MDA-MB-468 cells were treated with either Harmine alone or KRIBB11 alone or the combination of both at the indicated concentrations for 72 h and cell viability was analysed as in (C). (E) MDA-MB-231 parental cells were treated with or without 3 μM KRIBB11 and/or 10 μM harmine for 24 h. Cells were lysed and immunoblotting was carried out with the indicated antibodies.

Figure 3. Dual inhibition of HSF1 and DYRK2 bypasses proteasome-inhibitor resistance

(A) MM.1S parental and MM.1S BR cells were treated with or without the indicated concentrations of KRIBB11 for 72 h and cell viability was analysed by CellTiter 96® AQueous Non-Radioactive Cell Proliferation Assay kit. Data are represented as relative viability of DMSO-treated control. (B) MM.1S parental cells were treated with either 5 μM LDN192960 alone or 8 μM KRIBB11 alone or the combination of both for 72 h and cell viability was analysed by CellTiter 96® AQueous Non-Radioactive Cell Proliferation Assay kit. Data are represented as relative viability of DMSO-treated control. (C) MM.1S BR cells were treated with either 5 μM LDN192960 alone or 5 μM KRIBB11 alone or the combination and analysed as in (B). (D) KMS18 parental cells were treated with either 3 μM LDN192960 alone or 8 μM KRIBB11 alone or the combination and analysed as in (B). (E) KMS18 T21A cells were treated with either 8 μM LDN192960 alone or 5 μM KRIBB11 alone or the combination and analysed as in (B). (F) AN3-12 parental cells were treated with either 5 μM LDN192960 alone or 5 μM KRIBB11 alone or the combination and analysed as in (B). (G) AN3-12 A20T cells were treated with either 5 μM LDN192960 alone or 3 μM KRIBB11 alone or the combination and analysed as in (B). (H) AN3-12 V31E cells were treated with either 5 μM LDN192960 alone or 3 μM KRIBB11 alone or the combination and analysed as in (B). (I) AN3-12 M45V cells were treated with either 3 μM LDN192960 alone or 3 μM KRIBB11 alone or the combination and analysed as in (B). (J) AN3-12 A49E cells were treated with either 3 μM LDN192960 alone or 3 μM KRIBB11 alone or the combination and analysed as in (B). (K) AN3-12 A49T cells were treated with either 3 μM LDN192960 alone or 5 μM KRIBB11 alone or the combination and analysed as in (B). (L) AN3-12 C63F cells were treated with either 3 μM LDN192960 alone or 3 μM KRIBB11 alone or the combination and analysed as in (B). (M) AN3-12 C63Y cells were treated with either 10 μM LDN192960 alone or 3 μM KRIBB11 alone or the combination and analysed as in (B). (N) AN3-12 S130A cells were treated with either 10 μM LDN192960 alone or 10 μM KRIBB11 alone or the combination and analysed as in (B). (O) AN3-12 G183D cells were treated with either 8 μM LDN192960 alone or 8 μM KRIBB11 alone or the combination and analysed as in (B). See also Supplementary Figure S2 for further establishment of drug sensitivities of AN3-12 PSMB5 mutant cells. The P-value provided is the least significant value comparing the combination vs. single-drug treatments; ns: not significant (two-way ANOVA with multiple comparison: Fisher’s LSD test). Abbreviations: BR, bortezomib resistant; CR, carfilzomib resistant; IR, ixazomib resistant; OR, oprozomib resistant.

Figure 4. High nuclear DYRK2 expression predicts faster TNBC recurrence

Relationship between nuclear DYRK2 levels in tumour cells and time to (A) local recurrence or (B) distal recurrence in patients with indicated subtypes of breast invasive ductal carcinoma. Data represented as Kaplan–Meier curves and P-value were derived from survival curve comparison using Mantel–Cox Log-rank test. (C) Representative photomicrographs of tumours from the tissue microarray that were stained by DYRK2 IHC and scored as having either no (−), low, or high nuclear DYRK2 expression.

Figure 5. Dual depletion of DYRK2 and HSF1 impedes tumour growth in vivo

(A) Immunoblot confirming Crispr/Cas9-mediated H-KO MDA-MB-231 cells. (B) Quantitative PCR analysis to confirm shRNA-mediated DYRK2 knock-down in H-KO MDA-MB-231 cells. (C) Bar graph depicting cell invasion in a Matrigel transwell migration assay using MDA-MB-231 H-KO cells with the indicated shRNA. Data were acquired 18 h after seeding in upper chamber of 8 μm pore size transwells. Cells that invaded the Matrigel were quantified based on DNA content using CyQuant dye and data represented as RFU (relative fluorescence units). Reported P-value is derived by comparing to H-KO SCR cells, two-way ANOVA, mean ± SD from n=2 independent experiments with triplicates in each. (D) Bar graph depicting cell invasion in a Matrigel transwell migration assay using DMSO treated or 5 μM curcumin or 10 μM harmine-treated MDA-MB-231 parental or H-KO cells. Data were acquired as in (C). Reported P-value is derived by comparing to DMSO-treated control cells, two-way ANOVA, mean ± SD, with Fisher’s LSD multiple comparison from n=2 independent experiments with triplicates in each. (E) A total of 300000 MDA-MB-231 cells with or without the indicated genome editing or shRNA load were injected subcutaneously in NSG mice. Tumour volume was measured twice a week (n=8 mice per condition) and growth curves were plotted. ***P<0.001 (compared with parental group, two-way ANOVA, mean ± SD with Tukey’s multiple comparison). (F) Tumours from (A) were resected and tumour weight was measured. ***P<0.001, **P<0.01, *P<0.05 (ordinary one-way ANOVA, mean ± SD with Kruskal–Wallis multiple comparison from n=8 mice each). (G) A total of 300000 MDA-MB-231 HSF1 KO cells with the indicated shRNA load were injected into the mammary-fat pad of J:NU nude mice. Tumour volume was measured twice a week (n=5 mice per condition) and growth curves were plotted. **P<0.01 (compared with parental group, two-way ANOVA, mean ± SD with Tukey’s multiple comparison). (H) Tumours from (C) were resected and tumour weight was measured. *P<0.05 (ordinary one-way ANOVA, mean ± SD with Kruskal–Wallis multiple comparison from n=5 mice each).