Requirement for LIM kinases in acute myeloid leukemia

Abstract

Acute myeloid leukemia (AML) is an aggressive disease for which only few targeted therapies are available. Using high-throughput RNA interference (RNAi) screening in AML cell lines, we identified LIM kinase 1 (LIMK1) as a potential novel target for AML treatment. High LIMK1 expression was significantly correlated with shorter survival of AML patients and coincided with FLT3 mutations, KMT2A rearrangements, and elevated HOX gene expression. RNAi- and CRISPR-Cas9-mediated suppression as well as pharmacologic inhibition of LIMK1 and its close homolog LIMK2 reduced colony formation and decreased proliferation due to slowed cell-cycle progression of KMT2A-rearranged AML cell lines and patient-derived xenograft (PDX) samples. This was accompanied by morphologic changes indicative of myeloid differentiation. Transcriptome analysis showed upregulation of several tumor suppressor genes as well as downregulation of HOXA9 targets and mitosis-associated genes in response to LIMK1 suppression, providing a potential mechanistic basis for the anti-leukemic phenotype. Finally, we observed a reciprocal regulation between LIM kinases (LIMK) and CDK6, a kinase known to be involved in the differentiation block of KMT2A-rearranged AML, and addition of the CDK6 inhibitor palbociclib further enhanced the anti-proliferative effect of LIMK inhibition. Together, these data suggest that LIMK are promising targets for AML therapy.

Figure 1. High LIMK1 expression in AML is associated with patient survival and genetic subgroups.

(A) Results of RNAi screens in HL-60, U937, and OCI-AML3 cell lines. The Venn diagram shows the number of candidate essential genes scoring with at least three independent shRNAs in each cell line. The heatmaps display the number of scoring shRNAs (blue) and mRNA expression using RNA-seq data from the Cancer Cell Line Encyclopedia (red; downloaded from www.depmap.org, version 19Q1) of twelve common candidate genes. TPM, transcripts per million. (B) Survival of AML patients from the TCGA study (n=132) with LIMK1 mRNA expression above the median (LIMK1 ^high) and below the median (LIMK1 ^low). Statistical significance was assessed by log-rank test. (C) LIMK1 mRNA expression according to cytogenetic subtype from the Leukemia MILE study (MLL = KMT2A). Data were downloaded from www.bloodspot.eu. Black bars indicate the median. Statistical significance was assessed by one-way ANOVA with Dunnett‘s correction for multiple comparisons (each condition vs. healthy bone marrow). The non-indicated comparisons were not significant (P > 0.05). **** P < 0.0001. (D) Multivariable regression analysis with backward variable selection to assess associations between recurrent AML driver mutations and LIMK1 expression. Coef, coefficient; S.E., standard error. (E) Enrichment of gene expression signatures associated with rearranged KMT2A (MLL = KMT2A, right panel) and mutant NPM1 (left panel) in TCGA AML patients with high LIMK1 expression. The analysis was performed with the 10% highest (n=13) and 10% lowest (n=13) LIMK1-expressing TCGA AML patients.

Figure 2. Genetic suppression of LIMK1/2 inhibits proliferation and colony formation of KMT2A-rearranged AML cell lines.

(A) Competition assay of three KMT2A-rearranged AML cell lines transduced with two LIMK1- targeting shRNAs or NTC co-expressing GFP. GFP-positive cells were monitored by flow cytometry for 17 days starting three days post-transduction. Statistical analysis was performed on log-transformed data for the final day of the experiment by unpaired t-tests. * P < 0.05, *** P < 0.001, **** P < 0.0001. (B) Colony formation in methylcellulose after nine days of KMT2A-rearranged AML cell lines transduced with two LIMK1-targeting shRNAs or NTC. Statistical analysis was performed by unpaired t-tests. * P < 0.05, *** P < 0.001. ns, not significant (P > 0.05). (C) Proliferation determined by manual cell counting of THP-1 LIMK1 knockout and control (sgmCherry) clones. Each LIMK1 knockout clone was generated with a different sgRNA. Error bars indicate the standard deviation of three individual clones. Lines indicate the fitted model for exponential growth. Statistical significance for the corresponding proliferation constants was assessed by an unpaired t-test. * P < 0.05, *** P < 0.001. (D) Colony formation in methylcellulose of THP-1 LIMK1 knockout and control (sgmCherry) clones after ten days. Each LIMK1 knockout clone was generated with a different sgRNA. Statistical significance was assessed by an unpaired t-test. * P < 0.05, *** P < 0.001. (E) May-Grünwald-Giemsa-stained cytospin preparations of KMT2A-rearranged AML cell lines eight days after transduction with two LIMK1-targeting shRNAs or NTC. Original magnification, x 40. Insets show 1.5-fold magnified details of the corresponding photographs. (F) Competition assay of THP-1 cells transduced with two LIMK2-targeting shRNAs or NTC co-expressing GFP. GFP-positive cells were monitored for 17 days starting three days post-transduction. Statistical analysis was performed on log-transformed data for the final day of the experiment by unpaired t-tests. **** P < 0.0001. (G) Colony formation in methylcellulose of THP-1 cells transduced with LIMK2-targeting shRNAs or NTC after nine days. Statistical analysis was performed by unpaired t-tests. * P < 0.05. (H) Colony formation in methylcellulose after eleven days of two KMT2A-rearranged PDX samples transduced with LIMK1- or LIMK2-targeting shRNAs or NTC. For PDX AML-388, two technical replicates are shown. Statistical analysis was performed by unpaired t-tests. * P < 0.05, ** P < 0.01.

Figure 3. Transcriptome analyses identify potential effectors of LIMK1.

(A) MA plots showing deregulated genes in THP-1 cells eight days after transduction with two independent LIMK1-targeting shRNAs in biological duplicates followed by RNA-seq. Colors indicate the significance level (red: P < 0.05, grey: P > 0.05). LIMK1, as well as selected genes known to play a role in tumorigenesis and oncogenic signaling, are highlighted. One gene (SMPDL3B; x = 1.37, y = -5.28, P = 0.028) is not included in the right panel to allow increased resolution of the plot. nrc, normalized read count. (B) Survival of TCGA AML patients with high (expression above median) and low (expression below median) EGR1 mRNA levels (n=132). Statistical significance was assessed by log-rank test. (C) TCGA AML RNA-seq data for LSP1 and PAQR8 depending on LIMK1 expression status. The analysis was performed with the 10% highest (n=13) and 10% lowest (n=13) LIMK1- expressing patients. Statistical significance was assessed by an unpaired t-test. * P < 0.05, ** P < 0.01. TPM, transcripts per million. (D) Survival of TCGA AML patients with high (above median) and low (below median) expression of LSP1 and PAQR8 (n=132). Statistical significance was assessed by log-rank test. (E) Enrichment of HOXA9-related gene sets in the transcriptome of THP-1 cells after LIMK1 knockdown (shown in A). Genes were pre-ranked based on the mean log2(normalized read count shLIMK1/normalized read count NTC) of both shRNAs. KD, knockdown. (F) High expression of HOXA and HOXB genes in LIMK1 ^high (n=13) vs. LIMK1 ^low (n=13) TCGA AML patients. Significance levels (adjusted P-value) are indicated by red or grey color, respectively. Missing HOX genes did not pass the expression cut-off (TPM > 0.41).

Figure 4. CDK6 and LIMK are reciprocally regulated.

(A) Western blot of THP-1 cells transduced with LIMK1- or LIMK2-targeting shRNAs or NTC. Lysates were prepared eight days after transduction. (B) Western blot of THP-1 cells transduced with CDK6-targeting shRNAs or NTC. Lysates were prepared eight days after transduction. The result is representative of two independent experiments. (C) Western blot of THP-1-derived CDK6 knockout clones generated with two different sgRNAs and control clones (sgmCherry). (D) Western blot of a PDX sample five days after transduction with shRNAs targeting LIMK1 or CDK6. (E) Western blots of AML cells treated with the CDK6 inhibitor palbociclib or DMSO for 72 hours (NOMO-1) or 120 hours (MOLM-14 and THP-1). (F) Proliferation determined by manual cell counting (left panel) and colony formation in methylcellulose after ten days (right panel) of THP-1-derived LIMK1 knockout or sgmCherry control clones treated with palbociclib or DMSO. Proliferation data are presented as mean +/- SD of three individual clones, and the connecting lines indicate the fitted model for exponential growth. Statistical significance was assessed for the corresponding proliferation constants by two-way ANOVA with repeated measurements (P-value for palbociclib effect: < 0.0001, P-value for LIMK1 knockout effect: < 0.001, P-value for interaction effect: > 0.05). For colony formation, statistical significance was also analyzed by two-way ANOVA with repeated measurements (P-value for palbociclib effect: < 0.05, P-value for LIMK1 knockout effect: < 0.001, P-value for interaction effect: < 0.05). Each LIMK1 knockout clone was generated with a different sgRNA. DMSO data are the same as in Figures 2C and 2D, respectively. (G) May-Grünwald-Giemsa-stained cytospin preparations of THP-1 cells eight days after transduction with two shRNA targeting LIMK1 or NTC and treatment with palbociclib or DMSO. Original magnification, x 40. Insets show 1.5-fold magnified details of the corresponding photographs. DMSO data are the same as in Figure 2E. (H) Proliferation indices as assessed by CFSE tracking for THP-1 cells treated for four days with DMSO, the CDK6 inhibitor palbociclib, or the LIMK inhibitor LIMKi3 alone, or in combination. Data were normalized to the DMSO sample for each replicate. Statistical significance for each treatment condition was assessed by one-sample t-test on log-transformed data (asterisks below data points). Statistical significance for single vs. combination treatment was assessed by unpaired t-tests on log-transformed data (asterisks above data points). * P < 0.05, ** P < 0.01, *** P < 0.001.