Krüppel-like factor 4 promotes survival and expansion in acute myeloid leukemia cells

Acute myeloid leukemia (AML) is an aggressive hematological malignancy of the bone marrow that affects mostly elderly adults. Alternative therapies are needed for AML patients because the overall prognosis with current standard of care, high dose chemotherapy and allogeneic transplantation, remains poor due to the emergence of refractory and relapsed disease. Here, we found expression of the transcription factor KLF4 in AML cell lines is not silenced through KLF4 gene methylation nor via proteasomal degradation. The deletion of KLF4 by CRISPR-CAS9 technology reduced cell growth and increased apoptosis in both NB4 and MonoMac-6 cell lines. Chemical induced differentiation of gene edited NB4 and MonoMac6 cells with ATRA and PMA respectively increased apoptosis and altered expression of differentiating markers CD11b and CD14. Transplantation of NB4 and MonoMac-6 cells lacking KLF4 into NSG mice resulted in improved overall survival compared to the transplantation of parental cell lines. Finally, loss-of-KLF4 did not alter sensitivity of leukemic cells to the chemotherapeutic drugs daunorubicin and cytarabine. These results suggest that KLF4 expression supports AML cell growth and survival, and the identification and disruption of KLF4-regulated pathways could represent an adjuvant therapeutic approach to increase response.


INTRODUCTION
Acute myeloid leukemia (AML) is a malignant hematological disease that arises from genetic lesions in hematopoietic cells early in the hierarchy of normal myelopoiesis and a difficult to treat leukemia because of complexity of mutations and chromosomal translocations. Another complicating factor is the dynamic expansion of leukemic clones in the course of disease progression and in response to selective pressure imposed by antineoplastic drugs. Morphologic, cytogenetic, and genome sequence studies have been useful for the classification and treatment stratification of AML patients. The standard of care drugs, such as daunorubicin and cytosine β-D-arabinofuranoside (ara-C), non-selectively eliminate dividing cells, and can result in the outgrowth of aggressive leukemic clones that drive drug resistance and relapse [1,2]. For the development of curative treatments, the identification of genes involved in AML pathogenesis will lay the groundwork for designing novel targeted therapies.
The Krüppel-like factor 4 (KLF4) gene is located on human chromosome 9q31.2 and encodes for a transcription factor with key stemness properties better recognized as one of the factors regulating reprogramming of adult human fibroblasts into induced pluripotent stem cells [3,4]. KLF4 exerts its functions in cell division, survival, and differentiation through various gene regulation mechanisms, including interactions with the histone modifying enzymes p300 [5,6] and histone deacetylases [7,8], maintaining chromatin structures [9,10], and binding to both methylated and non-methylated DNA [11,12]. In hematopoietic cells, KLF4 has important functions in the regulation of hematopoiesis, such as regulation of monocyte development while restraining granulocyte production [13], survival of natural killer cells [14], and restriction of homeostatic and antigen driven T cell

Research Paper
Oncotarget 256 www.oncotarget.com proliferation [15,16]. The role of KLF4 in hematological malignancies has not been clearly defined because of its dual role as tumor suppressor and an oncogene depending on the leukemic cell context [17]. Supporting a suppressive role in leukemia, KLF4 inhibits T-cell acute lymphoblastic leukemia (T-ALL) and KLF4 expression is silenced in pediatric T-ALL by CpG promoter methylation, which was associated with aberrant expression of MAP2K7, a kinase pathway that accelerates disease progression [18,19]. Similarly, low expression levels found in B-cell non-Hodgkin and Hodgkin lymphomas, multiple myeloma, CDX2-driven AML, and FOXC1 positive AML hematopoietic stem/progenitor cells suggest a leukemic suppressive function [20][21][22][23][24]. However, other findings highlight a potential pro-leukemic capacity of KLF4. For example, high KLF4 levels have been associated with poor prognosis in pediatric Burkitt lymphoma [25] and KLF4 was found essential for the survival of leukemic stem cells via repression of the DYRK2 gene that encodes a kinase involved in the regulation of p53 activity and c-Myc stability [26]. Particularly in AML, there are contrasting evidences in the role of KLF4. The lack of mutations and genetic alterations suggest that KLF4 may be required for disease progression [27] whereas studies using the Cancer Genome Atlas have shown that KLF4 expression is not uniform among AML patients, but is generally lower than normal blood cells [7,28] and associated with leukemia stage [22,28,29]. Although these findings have fueled a presumption of tumor-suppressive function for KLF4 in AML, a model of ZMYM2-FGFR1 leukemia suggests KLF4 may be involved in AML leukemogenesis [30]. Altogether, these observations highlight a need to better define the role of KLF4 in leukemia and in particular AML.
Here, we report that KLF4 promotes survival and expansion of AML cells. KLF4 is expressed in a panel of AML cell lines, which correlates with absence of KLF4 gene methylation and post-translational proteasomal degradation. Further, KLF4 gene deletion via CRISPR-Cas9 technology in NB4 and MM6 cell lines resulted in impaired cell growth and survival and significant longer latency when transplanted into NSG mice. Additionally, targeting of KLF4 regulated pathways as adjuvant therapy is possible because KLF4 deletion did not affect sensitivity to daunorubicin and ara-C. These findings suggest targeting of KLF4 or KLF4-regulated pathways as a novel alternative to control leukemia burden in AML and further mechanistic studies to identify associated potential targets are warranted.

Regulation of KLF4 expression in AML cell lines
In order to compare CpG DNA methylation in the human KLF4 gene across diverse hematologic malignancies, we analyzed publicly available reduced representation bisulfite sequencing (RRBS) data through the cancer cell line encyclopedia (CCLE). This analysis demonstrated that the KLF4 gene is hypomethylated in most AML cell lines ( Figure 1A). In contrast, T-cell acute lymphoblastic leukemia (T-ALL) showed elevated gene methylation as previously described by our group [19]. Further analysis of KLF4 transcript levels and DNA methylation revealed no significant correlation in AML while the levels of KLF4 correlated with gene methylation in other types of cancer ( Figure 1B). This analysis suggests that KLF4 expression is not silenced epigenetically by DNA methylation at least in AML cell lines.
To further evaluate whether DNA methylation is involved in the regulation of KLF4 expression, we treated a panel of AML (NB4, THP-1, MonoMac-6, SKM-1), CML (K562), and EBV-transformed lymphoblastoid (LCL) cell lines with the hypomethylating agent 5-Azacytidine (5-aza). Neither KLF4 transcripts measured by qPCR nor protein levels detected by immunoblots demonstrated induction of KLF4 levels upon treatment with 250 and 500 nM 5-aza (Figure 2A and 2B). To examine post-translational regulation of KLF4, treatment of AML cell lines with the proteasome inhibitor MG-132 (10 μM) revealed KLF4 is not actively proteolyzed by the proteasome ( Figure 2C). Collectively, this data is consistent with the analysis of human AML CCLE data and supports our assessment that KLF4 is not being actively repressed by AML cells. To evaluate the importance of KLF4 expression in AML, we selected the NB4 and MonoMac-6 (MM6) cell lines for further study, which represent the range of KLF4 expression found in our panel as well as distinct AML subtypes.

KLF4 gene deletion impairs cell growth and promotes differentiation and apoptosis
To address the significance of KLF4 expression in AML, we deleted the KLF4 gene in the MM6 cell line, which represents MLL-rearranged leukemia, and the NB4 cell line which represent PML-RARα PML using CRISPR-Cas9 technology. MM6 cells expressed relatively higher levels of KLF4, transcript and protein, than NB4 cells ( Figure 3B, Supplementary Figure 1A and 1B). Multiplex sgRNA designed to target the second exon of the KLF4 gene were nucleofected into cells and single-cell clones were generated via cell dilution ( Figure 3A). KLF4 knockout clones (NB4 KO and MM6 KO ) were confirmed by DNA sequencing and immunoblots (Supplementary Figure 2A and 2B). Analysis of cell growth demonstrated that KLF4 loss diminished expansion in both MM6 and NB4 cell lines ( Figure 3B). This was caused by increase in apoptosis in NB4 cells and MM6 cells (up 4-fold) ( Figure 3C and 3E). Cell cycle analysis revealed increased sub-G1, consistent with increased apoptosis, and small reduction in G0/G1 phase of the cell cycle ( Figure 3D Oncotarget 257 www.oncotarget.com and 3F). Of note, multiple clones for each cell line were analyzed. These data suggest that KLF4 supports growth and survival in two AML cell lines encompassing different driver translocations.
Because KLF4 is known to play a role in myeloid differentiation [13,28,[31][32][33][34], we next sought to evaluate whether KLF4 regulates both differentiation and apoptosis. We incubated NB4 and NB4 KO cells with alltrans retinoic acid (ATRA) at a concentration of 3 μM and evaluated differentiation via flow cytometric detection of cell surface expression of CD11b. NB4 KO cells showed increased expression of CD11b after 48 hours incubation with ATRA with a moderate increase in apoptosis ( Figure  4A and 4B). As MM6 represents a monocytic leukemia subtype, we treated MM6 and MM6 KO cells with 3 nM phorbol 12-myristate 13-acetate (PMA) and evaluated expression of CD14. MM6 KO displayed lower CD14 expression than vehicle control (15-25% in control) and the difference became more significant after PMA treatment, which also correlated with increased apoptosis ( Figure 4C and 4D). Taken together, these results support a model in which apoptosis induced by lack of KLF4 correlates with cell differentiation depending on the cell line under study.

KLF4 expression shortens latency of in vivo AML xenografts
After observing the effects of KLF4 deletion on cell growth and survival in AML cell lines, we sought to evaluate the importance of KLF4 in in vivo leukemic expansion. We monitored overall survival of NSG mice transplanted with NB4, NB4 KO , MM6, and MM6 KO cells. Consistent with in vitro data, mice receiving NB4 KO and MM6 KO cells had significantly prolonged survival compared to corresponding parental cell line ( Figure 5A and 5B). As reported previously, NB4 cells form tumors in the ovary when transplanted into female NSG mice [35,36] and histologic analysis of NB4 KO tumors revealed large necrotic regions surrounded by apoptotic cells ( Figure 5A). Quantification of immunohistochemistry showed no significant differences in Ki67 positive cells whereas the percentage of caspase 3 positive cells increased in NB4 KO compared to NB4 (not shown). NB4 KO and MM6 KO cells cultured in methylcellulose medium showed smaller and more compact colonies with reduced colony-forming capacity ( Figure 5C and 5D). While NB4 cells did not upregulate CD11b expression in methylcellulose culture, MM6 cells demonstrated induction of CD14 expression. This induction was impaired in MM6 KO cells ( Figure  5C and 5D), consistent with response to treatment with PMA ( Figure 4C and 4D). Collectively, these data suggest pharmacological inhibition of KLF4 downstream targets may represent an alternative to control leukemia burden.

KLF4 does not regulate sensitivity to drugs used in AML therapy
To evaluate whether targeting of KLF4 might interfere with standard-of-care chemotherapies, we studied Oncotarget 258 www.oncotarget.com the role of KLF4 expression in response to daunorubicin and ara-C drugs used in AML therapy. Of note, NB4 cells treated with daunorubicin displayed loss of KLF4 protein, while all other cell line-drug combinations were unchanged ( Figure 6A-6D). When we treated parental and KLF4 knockout cell lines with increasing doses of daunorubicin or ara-C, we found the absence of KLF4 did not significantly alter response to these compounds, with the exception of NB4 cells treated with ara-C ( Figure 6A-6D). To evaluate pharmacological upregulation of KLF4, we tested APTO-253, a compound previously reported to induce cytotoxicity through KLF4 [37]. However, deletion of KLF4 in NB4 cells did not alter APTO-253-induced cytotoxicity (Supplementary Figure 3A-3D), suggesting a KLF4 independent mechanism, likely by loss of MYC [38]. Finally, we treated gene edited cell lines with the BCL2 inhibitor ABT-199 (Venetoclax) to test whether KLF4 deletion might impact induction of apoptosis. NB4 KO and MM6 KO cells showed a dose-dependent cytotoxicity comparable to controls (Supplementary Figure 4A and 4B). Taken together, inhibition of KLF4 should not significantly affect the course of treatment with standard-of-care chemotherapy and warrants further studies to elucidate 'druggable' targets regulated by KLF4.

DISCUSSION
Currently, the response of AML patients to standard multi-drug therapy has not significantly improved through the years because of complex genetic heterogeneity and therapy-induced generation of more aggressive leukemic clones that are resistant to drug therapy. The mutational landscape determined by genome sequencing revealed the presence of multiple driver mutations in each patient [39,40]; however, these studies cannot identify genes with pro-leukemic functions. This work identifies that the transcription factor KLF4 contributes to expansion of AML cells and in turn loss-of-KLF4 reduces leukemic cell growth and survival.
Proteins with tumor suppressor function are inactivated in cancer through DNA methylation, gene mutation/deletion, or post-translational proteolysis [41-  43]. COSMIC data analysis failed to identify mutations in the KLF4 gene [27]. Although a 9q31 breakpoint has been described in a pediatric MDS patient [44], the functional contribution of KLF4, if any, present in this case is unknown. Our findings support a different function for KLF4 by promoting the expansion of AML cells. More consistent with a pro-oncogenic role of KLF4 in AML, DNA methylation did not appear to play a significant role in AML cell lines, in contrast to other cancer types [45,46]. Similarly, KLF4 is not proteolyzed by the proteasome and therefore is expressed in a panel of AML cell lines. Some efforts have been made to explore the effect of KLF4 expression in AML; however, these studies have been carried out via ectopic KLF4 expression that aberrantly causes differentiation and cell death, including via interactions with BAALC and HDACs, and induction of microRNAs [7,28,47]. This approach, while in some cases mechanistically informative, leaves a gap in understanding for the function of KLF4 expression in leukemic cells.
KLF4 shows higher expression in monocytic leukemia compared to other AML subtypes, which comprise M4 and M5 AMLs in the FAB classification system and frequently feature MLL-rearrangements [48]. Although KLF4 is known to regulate monocytic development, the function of KLF4 expression in monocytic leukemia has not been thoroughly explored [13]. The MM6 cell line represents a monocytic AML containing the MLL-AF9-rearrangement and high levels of KLF4 expression relative to other tested cell lines. Interestingly, deletion of the KLF4 gene diminished cell growth and survival not only in this MLL-rearranged cell line, but also in NB4, a PML-RARα-rearranged cell line corresponding to the M3 subtype of AML, which express lower KLF4 levels compared to other AML cell lines [49,50]. Gene edited NB4 and MM6 showed reduced clonogenicity and smaller colonies in methylcellulose cultures correlating with the reduction of proliferation and survival seen in vitro and in the xenograft model, suggesting that KLF4 likely plays a general role in AML regardless of the subtype. While it is promising that KLF4 deletion is disruptive in AML cells with different cytogenetics, a more comprehensive study of downstream targets in a larger cohort of patients would be necessary to evaluate whether inhibiting KLF4 activity could represent a treatment modality applicable to most AML patients. It will be important to define also whether KLF4 is upregulated in leukemic clones emerging in refractory or relapsed leukemia because high levels of expression of KLF4 were found in leukemic clones in a relapsed patient with chronic lymphocytic leukemia [51].
Potential application of pharmacological inhibition of KLF4 regulated pathways in AML as adjuvant therapy to increase treatment response in patients will require further studies using patient samples. We tested whether genetic loss of KLF4 altered response to daunorubicin and cytarabine, two conventional chemotherapeutic agents used in standard of care of AML patients [52]. These chemotherapies are not well tolerated in the elderly patient population comprising AML, with median survival for patients on chemotherapy about 13 months for those 65-75 years of age, and only 6 months for 75 and older [53]. Data show that NB4 and MM6 cells responded to both drugs similarly in the presence or absence of KLF4. This is important because potential use of KLF4 inhibition as adjuvant therapy would require no interference with current therapies. Although no additive or synergistic effect of genetic loss of KLF4 was observed, pharmacological inhibition of downstream targets will be more adequate for pharmacological inhibition.
Collectively, genetic loss of KLF4 in AML cell lines provides novel insights into the importance of KLF4 in the maintenance of AML, particularly supporting leukemic cell growth and survival. Further studies are warranted to identify KLF4-regulated pathways amenable for therapeutic targeting with small molecules.

Analysis of data from cancer cell line encyclopedia
KLF4 mRNA expression (RNASeq) and DNA methylation (RRBS) were downloaded for cell lines from the publicly accessible cancer cell line encyclopedia database from Broad Institute. Data were replotted and

Chemical compound treatment
For hypomethylating treatment, 5-aza (Sigma) was prepared in DMSO to a concentration of 100 mM and then diluted to final concentration by diluting in respective medium for each cell line. Cells were plated in 6-well plates (1.25 × 10 5 cells/ml) and 5-aza was replaced every 24 hours for a total of 96 hours treatment. For proteasomal inhibition, 20 mM InSolution™ MG-132 (CalBioChem) was stored at −20°C. Cells were plated in 6-well plates (5.00 × 10 5 /ml) in the presence of 10 μM MG-132 for 4 h.

Quantitative real-time PCR
Total RNA was extracted using the RNeasy Mini kit (Qiagen) from human AML cell lines. cDNA was synthesized from 100-500 ng RNA using

KLF4 gene editing via CRISPR/CAS9
To knockout the KLF4 gene in MM6 and NB4 cells, we used 3 chemically modified synthetic sgRNAs from Synthego with sequences GCCATGTCAGACTCGCCAGG, CGCCGGGCCAGAC GCGAACG, and GAGCGATACTCACGTTATTC and followed the instructions obtained from the manufacturer to form the RNP complex with Cas9 (Cas9 plus sgRNAs). Briefly, the Cas9-sgRNA RNP in a total volume of 12 μl were electroporated using the Neon transfection system (Thermo Fisher Scientific) under the following conditions: 1400V, 10 ms, 3 pulse. Electroporated cells were cultured in growth medium for 3 days and then single cell clones were obtained by diluting cells to 0.5 cells/100 μl and plating in 96-well U-bottom plate for 2 weeks. Genomic DNAs were isolated and then used for PCR amplification using forward primer GTGTTATGTCCTGTCTGCCCAATT and reverse primer GTTTTGGCTTCGTTTCTTCTCTTC, spanning the Cas9-sgRNA cleavage site. PCR amplicons were then used for sequencing analysis using sequencing primer (CTTACCCTCGTTCAGTGGCTCTT) to identify the knockout efficiency using Inference of CRISPR Edits (ICE) which is a free and open source software tool that offers fast and reliable analysis of CRISPR editing data from Synthego. Knockout was additionally verified by immunoblotting and PCR detection of cleaved genomic DNA on 2.0% agarose gel.

Cell death and cell cycle assays
For analysis of cell growth, NB4, NB4 KO , MM6, and MM6 KO cells were plated at 100,000 cells/ml and collected after 24 hours. For apoptosis detection, these cells were stained with Annexin V and 7-AAD according to BD Apoptosis detection kit protocol (BD Biosciences). For cell cycle analysis, cells were lysed in hypotonic DNAstaining solution containing 0.1% sodium citrate, 0.1% Triton-X, 100 μg/ml RNAse A, 50 μg/ml propidium iodide in de-ionized water. All samples were plated in triplicate. Cells were analyzed in a BD FACSCanto instrument and FlowJo software.

Chemical induced differentiation
10 mM PMA (Sigma) and 100 mM ATRA (Sigma) were dissolved in DMSO and stored at −20°C. Compounds were diluted to 3 nM (PMA) and 3 μM (ATRA) in respective media and cells were suspended at 100,000 cells/ml in 6-well non-tissue culture plates. Cell differentiation was detected by flow cytometry, staining with PE anti-human CD14 (BioLegend) and APC antimouse/human CD11b (BioLegend) from cells sampled from culture every 24 hours. All sample data was collected on BD FACSCanto instrument and analyzed using FlowJo software.

Sensitivity to chemotherapeutic drugs
10 mM stocks of daunorubicin hydrochloride and cytosine β-D-arabinofuranoside (Sigma) were dissolved in de-ionized water and stored at −20°C. Compounds were diluted to the specified final concentration in their respective media. The volume of vehicle was always lower than 0.1% of the total volume, and the same dilution of vehicle in culture medium was used as the vehicle control. Cells were plated in 96-well plates (1.5 × 10 4 cells in 100 μl per well) with increasing doses of drug (0-32 μM) for 48 hours. Cell viability was measured using CellTiter-Glo Luminescent Cell Viability Assay (Promega). The halfmaximal inhibitory concentration (IC 50 ) was calculated by nonlinear regression analysis using GraphPad software.
Oncotarget 264 www.oncotarget.com AML cell line xenografts NB4 and MM6 cells were washed with PBS and resuspended in medical grade physiologic saline solution. Cells (5 million) were injected into 10-20 week old NOD scid gamma (NSG) mice via tail vein. Mice were monitored in peripheral blood by cytometric detection of human CD45 to determine leukemic cell engraftment and disease progression, and mice were euthanized upon signs of moribundity.

Human colony-forming cells (CFC) assay
From culture, NB4 and NB4 KLF4KO cells (300 cells) and MonoMac-6 and MonoMac-6 KLF4KO cells (1000 cells) were serially plated in methylcellulose medium (MethoCult H4434, Stemcell Technologies) every 7 days. After each passage, colonies were counted, imaged, and replated in fresh methylcellulose medium. MonoMac-6 colonies were counted, imaged, and re-plated on day 14. Cells were also isolated after plating and analyzed by flow cytometry for expression of CD11b and CD14.

Statistical analysis
Experiments were performed without blinding and with no exclusion of samples. Linear regression was performed with GraphPad Prism software. An unpaired 2-tailed Student t test was used for statistical analysis. The survival of leukemic mice was visualized using Kaplan-Meier curves, and statistical significance was calculated using the log-rank test (GraphPad Prism). P values were determined using GraphPad software. Results with a P value < .05 were considered statistically significant.

Author contributions
A.L. designed and performed the experiments, perform statistical analysis, interpreted the data, and wrote manuscript. SB, AC, and VP contributed to immunoblot analysis. MP generated NSG mice used in this study. HDL conceived and directed the project as the principal investigator, contributed to preparation of manuscript, and funded the research.

ACKNOWLEDGMENTS AND FUNDING
The authors acknowledge Angela Major for histopathological analysis and Karen Prince for the preparation of figures. This work was supported by the National Cancer Institute to H.D.L. (RO1 CA207086), the ASH Bridge award, the Baylor College of Medicine Comprehensive Cancer Training Program from the Cancer Prevention and Research Institute of Texas to A.H.L. (RP160283), the Cytometry and Cell Sorting Core at Baylor College of Medicine (P30 AI036211, P30 CA125123, and S10 RR024574) and the Flow Cytometry Core at Texas Children's Cancer and Hematology Center (S10 OD020066).The results discussed here are in part based upon data generated by the TCGA Research Network: https://www.cancer.gov/tcga.

CONFLICTS OF INTEREST
Authors have no conflicts of interest to declare.