Knockout of LASP1 in CXCR4 expressing CML cells promotes cell persistence, proliferation and TKI resistance

Abstract Chronic myeloid leukaemia (CML) is a clonal myeloproliferative stem cell disorder characterized by the constitutively active BCR‐ABL tyrosine kinase. The LIM and SH3 domain protein 1 (LASP1) has recently been identified as a novel BCR‐ABL substrate and is associated with proliferation, migration, tumorigenesis and chemoresistance in several cancers. Furthermore, LASP1 was shown to bind to the chemokine receptor 4 (CXCR4), thought to be involved in mechanisms of relapse. In order to identify potential LASP1‐mediated pathways and related factors that may help to further eradicate minimal residual disease (MRD), the effect of LASP1 on processes involved in progression and maintenance of CML was investigated. The present data indicate that not only overexpression of CXCR4, but also knockout of LASP1 contributes to proliferation, reduced apoptosis and migration as well as increased adhesive potential of K562 CML cells. Furthermore, LASP1 depletion in K562 CML cells leads to decreased cytokine release and reduced NK cell‐mediated cytotoxicity towards CML cells. Taken together, these results indicate that in CML, reduced levels of LASP1 alone and in combination with high CXCR4 expression may contribute to TKI resistance.

treatment cannot serve as a reliable clinical test to identify patients who can safely discontinue their TKI treatment without risk of a molecular relapse.
A proposed mechanism for molecular relapse in CML patients involves leukaemic stem cell homing to the bone marrow, mediated through the chemokine receptor 4 (CXCR4)-C-X-C motif chemokine 12 (CXCR12) axis. While BCR-ABL expression has been reported to down-regulate CXCR4 levels, TKI treatment was shown to restore the chemokine receptor's surface expression. Subsequent activation of the receptor by CXCL12 leads to homing of leukaemic stem cells (LSCs) to the bone marrow (BM) microenvironment, which is thought to promote quiescence and survival of the cells. After treatment discontinuation, these surviving CML stem cells may cause a relapse of the disease. 8 A recently identified CXCR4 binding partner and overexpressed substrate of BCR-ABL in CML is LASP1. 9 The protein has initially been identified from a cDNA library of breast cancer metastases. LASP1 is overexpressed in several human tumours and is involved in cell proliferation, migration, tumorigenesis and chemoresistance. 10,11 The protein comprises an N-terminal LIM domain, two nebulin-like actin binding repeats, a linker region with two phosphorylation sites at S146 and Y171 and a C-terminal SH3 domain. 12 Furthermore, LASP1 has been identified as member of a six genes signature being highly predictive for CML disease phases, thereby allowing a more precise prediction of relapse after stem cell transplantation than clinical risk factors alone. 13 Binding of LASP1 to the conserved leucine/isoleucine (LKIL) motif at the carboxy-terminal domain of CXCR4 requires phosphorylation of LASP1 at S146 and stabilizes the chemokine receptor. 14,15 In breast cancer cells, receptor activation then results in the translocation of LASP1 to the nucleus and interference of the protein with the epigenetic machinery. 16 In CML cells, however, the function and effects of LASP1 have not been investigated. The present work now demonstrates that low LASP1 levels, as observed in TKI non-responders and blast crisis patients, affect proliferation, migration and cytokine release in the CML cell line K562 and might contribute to worse patient outcome.

| Microarray and gene set enrichment analyses
Publicly available gene expression data of n = 62 individual CML patients were retrieved from the Gene Expression Omnibus (GEO) and the Array Express platform hosted at the EBI (http://www. ebi.ac.uk/array expre ss/). Accession numbers GSE14671: CD34positive selected BM cells of n = 59 patients in late or newly diagnosed chronic phase 17 and GSE48294: CD34+ selected cells from n = 3 patients in chronic phase cultured in vitro with or without imatinib under normoxia (21% O 2 ) or hypoxia (0.5% O 2 ) for 24 or 96 hours. 18 Data with accession number GSE14671 were generated on Affymetrix HG-U133plus2.0 microarrays and with the accession number GSE48294 on Illumina HT-12 beadchips. All CEL files of the GSE14671 data set were simultaneously normalized with robust multi-array average (RMA) 19 using a custom brain array chip description file (CDF; v20 ENTREZG). 20 For the analysis of the GSE48294 data set, already normalized expression values from the series matrix file were used.

| Generation of LASP1 knockout cells
Wild-type cells transfected with control plasmid will be referred to as K562-LASP1↑, whereas cells with the LASP1 knockout plasmids SC-404630-NIC and SC-404630-NIC2 are referred to as F I G U R E 1 Validation of K562 cells with inactivated LASP1 and CXCR4 overexpression. A, Western blot analysis (10% gel) of LASP1 and CXCR4 in K562 cells after CRISPR/Cas9-mediated LASP1 knockout and after lentiviral transduced CXCR4 expression of indicated cell lines. Relative expression of CXCR4 mRNA (B) and of LASP1 mRNA (C) in the generated cell lines was analysed by qRT-PCR as described in the Materials and Methods section. Results represent the mean of three independent experiments ± SD. D, CXCR4 functionality in CXCR4 overexpressing K562 cells was validated by AKT1-S473 phosphorylation after 25 nmol/L (200 ng/mL) CXCL12 stimulation in a time dependent manner by Western blot analysis (10% gel). β-actin served as loading control. E, Flow cytometric analysis of CXCR4 surface expression (50 000 recorded events). CXCR4 cell surface expression in CXCR4 overexpressing cell lines exceeded non-transduced cell lines. CXCL12, C-X-C motif chemokine 12; CXCR4, chemokine receptor 4; DMSO, dimethyl sulfoxide; LASP1, LIM and SH3 domain protein 1; SD, standard deviation K562-LASP1↓-h1 and K562-LASP1↓-h2, respectively. Sustained sensitivity to puromycin was checked.

| Migration assay
For migration assays, Corning ® FluoroBlok™ Cell Culture Inserts with a pore size of 8 μm combined with 24-well cell culture plates were used. Before start of the migration assay, the cells were serumstarved overnight in RPMI with 1% BSA. Cells were then stained with CellTracker™ Red CMPTX dye (Thermo Fisher) at a final concentration of 10 µM for 30 minutes in serum-free RPMI without phenol red. Cells were pelleted and again incubated in serum-free RPMI for 30 minutes without phenol red to remove the surplus dye. Cell migration was displayed as signal compared to the measurement at 4 hours.

| Adhesion assay
Adhesion assays under flow conditions were performed as previously described. 22 Detailed information can be found in Supplemental Materials and Methods.

| Patients and healthy controls
We additionally analysed a cohort of 57 patients (13 female; me-

| Statistics
Unless otherwise indicated, all experiments were performed in three independent experiments in triplicates and results are expressed as means ± standard deviation (SD). Statistical comparison of means was performed by ANOVA using GraphPad Prism 8.0.2 (GraphPad Inc). Wilcoxon-Mann-Whitney test was used to compare patients' cohort with control group. Data were considered to be significant with P < .05 (*P < .05; **P < .01; ***P < .001; ****P < .0001).
Additional Supplemental Materials and Methods can be found in Supplemental Information.

| LASP1 expression is lower in imatinib nonresponders and inversely regulated with gene sets associated with ATP generation, transcription and translation
Microarray analyses of whole BM probes from CML patients in chronic phase, accelerated phase and blast crisis, analysed by Yeung and colleagues, have demonstrated that the protein LASP1 is a member of a six genes signature which is highly predictive for CML disease phases and allows a precise prediction of relapse after stem cell transplantation. 13 While the association between certain expression patterns and disease progression as well as therapy response could be identified in CD34 selected blood and bone marrow samples, this was not possible in unselected probes. 17,24 For that reason, we used microarray data of isolated CD34-positive BM cells in order to further assess the diagnostic value of LASP1 for patient outcome. The clinicopathological data allowed for discrimination between TKI responders (n = 41) and non-responders (n = 18). Within a set of n = 36 samples of CML patients in the late chronic phase and n = 23 samples of newly diagnosed chronic phase patients, 17 LASP1, S100P, SFN and CXCR4 mRNA were found to be expressed significantly lower (P < .05; marked by colour) in patients' samples not responding to imatinib therapy (Table 1). Here, LASP1 was positively correlated with CRKL and STAT3 but negatively correlated with S100A4 and ZO-2.
Gene set enrichment analysis for responders and non-responders yielded 22, respectively, 2 gene sets with a significant positive correlation, and 164, respectively, 17 gene sets, that were negatively correlated with LASP1 (for additional information see Table   S2). In imatinib non-responders, LASP1 was inversely regulated with several gene sets associated with ATP generation as well as transcriptional and translational processes (Table S2). Similar but less pronounced results were obtained for imatinib responders. In these patients, LASP1 gene sets were associated with CXCR4, as well as with immune cell differentiation and function.
Since LASP1 expression is stimulated by hypoxia-inducible factor 1-alpha (HIF1α), 25 we analysed the microarray data for differential LASP1 expression under hypoxic conditions in the BM environment. 26,27 However, we did not find a positive correlation between LASP1 and HIF1α in this data set. Among genes known to be relevant for regulation and function of LASP1, only CXCR2 and DNMT3A appeared to be up-regulated after 96 hours of hypoxia, while SRC was down-regulated (Table 1).

| Generation of a CML precursor cell LASP1 knockout model
The BCR-ABL-positive cell line K562 expresses low to no detectable cell surface protein and mRNA levels of CXCR4 and CXCR7. 28 To investigate a possible role of LASP1 in CXCR4 signalling and to characterize the effect of LASP1 on disease progression, K562 cell lines with low and elevated levels of CXCR4 expression in the presence and absence of LASP1 were generated by stable CRISPR/Cas9-based LASP1 knockout and viral transduction of CXCR4. Western blot analysis revealed complete allelic knockout of LASP1 in K562-LASP1↓-CXCR4↑ and K562-LASP1↓-CXCR4↓ clones ( Figure 1A). In order to exclude possible off-target activity, the CRISPR/Cas9 system used in this study relies on paired nicking which has been shown to reduce off-target activity by 50-to 1000fold. 29 Off-target effects were further minimized by additional pooling of 5 clones of each generated cell line. qRT-PCR reassured

| Knockout of LASP1 enhances viability in the presence of nilotinib
To investigate a possible role of LASP1 in the context of CXCR4mediated cell proliferation in CML, viability of the newly generated TA B L E 1 Expression and correlation of LASP1 and LASP1 binding and interaction partners in imatinib responders vs non-responders Analysis based on GSE14671 Gene FC (responders vs. Note: Bold: Significantly differentially expressed in responders and non-responders.
Bold with italics: Significantly (counter)coregulated with LASP1 expression in responders.
LASP1 interaction partners: Publicly available microarray data sets of isolated CD34-positive BM cells were discriminated in TKI responders and non-responders. 17 The according CEL files were normalized with RMA 19 using a custom brain array CDF. 20 Genes being significantly lower expressed in patients' samples not responding to imatinib therapy are marked in green. Genes being significantly coregulated with LASP1 in responders and non-responders are marked in yellow and orange, respectively. cell lines was tested under several conditions. In order to reflect clinical practice, experiments were conducted with nilotinib.
This TK inhibitor has been shown to achieve faster and deeper molecular response, which is associated with a higher probability for TFR. 30,31 We used three different concentrations (30, 60 or 120 nmol/L) and stimulated with either 12.5 nmol/L CXCL12 alone or in combination with 120 nmol/L nilotinib. As an additional setting, CXCL12 stimulation was carried out after 1 hour pre-incubation with the CXCR4 inhibitor plerixafor ( Figure S1)  Figure 2A). Similar effects were observed when stimulated with CXCL12 ( Figure 2B). In the presence of nilotinib, all cell lines showed high sensitivity to the TKI but the pro-proliferative effect of CXCR4 expression was still observable (K562-LASP1↑-CXCR4↓ and K562-LASP1↑-CXCR4↑, Figure 2C). However, in contrast to basal conditions, LASP1 depletion resulted in enhanced cell growth. The effect was most prominent after 24 hours in the presence of 60 nmol/L nilotinib and was still visible after 48 hours ( Figure 2C; focus on blue bars). After 72 hours, nilotinib-treated cells were dying.

| The combination of CXCR4 overexpression and LASP1 knockout promotes resistance to nilotinib in CML cells
To  Figure 3C) and 60 nmol/L nilotinib ( Figure 3D)]. Plerixafor itself had no effect on cell cycle ( Figure S3).
Taken together, these results suggest a survival advantage of CML cells overexpressing CXCR4 in the presence of low LASP1 levels.

| LASP1 knockout impairs migration but has no significant effect on adhesion of CML cells
The importance of CXCR4 for migration and adhesion in CML cells has already been demonstrated. 28 So far, LASP1 involvement in migratory processes has only been demonstrated for solid tumours. 11 Therefore, we aimed to characterize the role of CXCR4 and LASP1 in CML cell migration using the different K562 cell lines ( Figure S4).
A schematic drawing and assay conditions are provided in Figure 4A.  Figure 4C) and CXCL12 ( Figure 4D). Pappenheim staining of cytospin-fixed migrated cells was used to confirm cell migration visually ( Figure 4E).
Adhesion of the K562-generated cell lines was tested by counting pre-labelled K562 cells on a HUVEC monolayer under flow conditions ( Figure 5A). In the presence of CXCL12, CXCR4 expression led to a significant increase in adhesion. The process seemed to be augmented by the knockout of LASP1; however, this effect was not statistically significant ( Figure 5B).

| CXCR4 overexpression and LASP1 knockout attenuate natural killer cell-mediated cytotoxicity
Microarray analyses suggested an impact of LASP1 on immune cell differentiation and function (Table S2). Therefore, we ana-  Figure 6A,C,E). As LASP1 is known to affect transcriptional activity 16,32 and to enhance secretory processes, 32-34 we used qRT-PCR ( Figure 6B,D,F) to differentiate between reduced cytokine release solely due to lowered transcriptional expression or a direct effect of LASP1 on the vesicular secretory process. Our results demonstrate that, despite higher mRNA levels in the LASP1-knockout cells ( Figure 6F), MCP-1 release is reduced ( Figure 6E), while the secretion of IL-6 and IL-8 followed their expression levels ( Figure 6B,D).
To establish a possible link between CXCR4 and LASP1 expres- observed for the highest effector to target cell ratio 2:1 ( Figure 6G and Figure S7). These findings support our hypothesis that CXCR4 F I G U R E 4 Importance of LASP1 and CXCR4 for migratory potential. (A) Schematic of a migration chamber and assay conditions: Fluorescent signal from actively migrating cells was detected from below without shine-through artefacts. Relative migration towards gravitation (B), 10% FCS (C) and 12.5 nmol/L (100 ng/mL) CXCL12 (D). In the presence of CXCR4, knockout of LASP1 resulted in reduced migration. (E) Cytospin and Pappenheim staining of migrated cells. Cytospins were prepared as previously described. 9 Results represent the mean of three independent experiments in duplicates ± SD. BSA, bovine serum albumin; CXCL12, C-X-C

| D ISCUSS I ON
Over the past years, increasing knowledge about the interactions between leukaemic cells and the BM has provided insights into mechanisms of cell survival and CML resistance towards therapeutic agents. Although BCR-ABL expressing cells have functional defects in the CXCR4 signalling axis, imatinib treatment increases CXCR4 expression, and CXCL12 activation improves BM homing and survival of CML stem cells. 8,35 Based on recently published data demonstrating a binding of LASP1 and CXCR4 in breast cancer cells, we investigated the role of LASP1 in the light of this interaction for CML. 15 To do so, we generated cell lines with inactivated LASP1 and/or CXCR4 overexpression using a CRISPR/Cas9-based LASP1 knockout system and a lentivirus-mediated CXCR4 overexpression in K562 cells.
Our data indicate that down-regulation of LASP1 promotes survival advantages for CML cells, especially when treated with TKI.
Data above demonstrate that abrogation of LASP1 expression levels promotes resistance towards TKI treatment, reduces migration, increases adhesive behaviour and contributes to impaired recognition by the immune system. Thereby, our results provide the first cell-based confirmation of the bioinformatics data by Yeung and colleagues, who predicted that a reduced LASP1 concentration might be unfavourable during CML progression, as they observed lower LASP1 mRNA levels in blast crisis patients. 13 Looking at the bigger picture of LASP1 involvement in malignancy, our results are in contrast to data on solid tumours, in which LASP1 overexpression contributes to cancer aggressiveness, 10 thus implementing major biochemical differences of LASP1 action in united cell structures and haematological cells.
In certain solid tumours, hypoxic conditions lead to an up-regulation of LASP1 12 by binding of HIF1α to a hypoxia response element in the LASP1 promotor region. 25 Although the environment in the BM was shown to be hypoxic, 27 we were not able to identify a positive correlation between LASP1 and HIF1α in CML within the microarray data sets.
In solid tumours, localization of LASP1 is not restricted to the cytoplasm, as the protein can also be found within the nucleus. Nuclear localization significantly correlates with poor outcome in breast cancer 36 and hepatocellular carcinoma. 37 Mechanistically, phosphorylation of LASP1 at S146 allows an interaction with CXCR4. 15  translocation into the nucleus through the interaction with zona occludens 2 (ZO-2) 12 and binding to nuclear Snail, ubiquitin-like with PHD and ring finger domains 1 (UHRF1) and histone methyltransferase G9a. 15 We have not been able to detect LASP1 within the nucleus of CML cells, yet, 9 most likely because these cells harbour no  Expression and activation of CXCR4 in K562 cells had no impact on adhesions. 38 The important and well-known effect of CXCR4 on adhesion is not due to zyxin expression as CXCR4 up-regulation is not affecting zyxin levels ( Figure S8).
Cytokine release plays an important role in the regulation of the immune response. Our data revealed an influence of LASP1 Our data not only indicate a role of LASP1 for CML progression, but, with respect to the microarray data, also suggest an impact of the protein in leukaemic stem cell behaviour. 28,42 Even though the K562 cell line is not really suited for drawing conclusions regarding leukaemic stem cells, some findings are similar to cell behaviour observed in LSCs: Stem cell quiescence is closely related to apoptotic resistance as these cells are not fully eliminated by cell cycle-specific drugs or TKIs. 26,43 Promoting cell cycle re-entry of quiescent leukaemic stem cells results in better outcome of CML treatment, but also affects long-term capacity for survival and self-renewal of hematopoietic stem cells. 26 Our data from the K562 model shows that despite nilotinib treatment, knockout of LASP1 results in higher amount of cells entering G2 phase ( Figure 3D).
In cells overexpressing CXCR4, LASP1 depletion even shows improved survival rates ( Figure 3B). It is therefore possible that TKI resistance in LASP1 depleted K562 and drug resistance of quiescent LSCs with low LASP1 expression share the same mechanism.
In addition, a tendency of LASP1-depleted K562 cells to increased adhesion is observed ( Figure 5B). Whether this contributes to cell adhesion-mediated drug resistance (CAM-DR) has to be further elucidated. 26 F I G U R E 7 Negative effect of impaired LASP1 levels worsen patients' outcome: a case presentation. A 66-year-old woman presented 5 months after being diagnosed to suffer from CML in chronic phase in 2011 (157.4 Gpt/L white blood cells) at the university hospital in lymphatic blast crises with low LASP1 levels (blue line), despite initiated imatinib therapy. Cytarabin, vincristine and hydroxycarbamide (indicated as ) were administered unsuccessfully. As cytogenetics finally revealed the presence of Y253H, E255K and T315I mutations, ponatinib (dotted line) was initiated. Subsequently, the administered pre-phase chemotherapy consisted of methotrexate, dexamethasone and cyclophosphamide (indicated as ) according to German Multicenter Study Group for Adult Acute Lymphoblastic Leukemia. 44 Induction therapy had to be interrupted due to clinical deterioration. Finally, the patient underwent conditioning therapy with treosulfan, fludarabine and antithymocyte globulin 45 followed by allogenic peripheral blood stem cell transplantation of 4.1 × 10 6 CD34+ cells/kg bodyweight (indicated as ▼) from an unrelated male, human leucocyte antigen allele matched 10/10 donor. After peripheral blood stem cell transplantation, BCR-ABL levels declined, while LASP1 levels increased. Due to a renewed blast crisis (and concomitant lowered LASP1 levels), the patient died 83 d after SCT (indicated as ) despite further doses of cyclophosphamide (indicated as ). The provenience and preparation of blood samples have been described before. 9 CML, chronic myeloid leukaemia; CT, chemotherapy; LASP1, LIM and SH3 domain protein 1; PBSCT, peripheral blood stem cell transplantation; TKI, tyrosine kinase inhibitor In order to properly investigate LASP1 effects in LSCs, further work has to be done including ex vivo studies of sorted patient stem cells and animal models.

ACK N OWLED G EM ENTS
The excellent technical assistance of Christine Gräfe, Yvonne Schlenker, Kristian Kamp and Mike Fischer is gratefully acknowledged. Thanks to Katrin Hoffmann for statistical assistance. and JJF analysed and compiled the data and wrote the manuscript.

E TH I C A L A PPROVA L
The study has been approved by the institutional ethics committee, and patients provided written informed consent in accordance with the Declaration of Helsinki.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.