Focal Adhesion Kinase as a Potential Target in AML and MDS

Abstract

Although overexpression/activation of focal adhesion kinase (FAK) is widely known in solid tumors to control cell growth, survival, invasion, metastasis, gene expression, and stem cell self-renewal, its expression and function in myeloid leukemia are not well investigated. Using reverse-phase protein arrays in large cohorts of newly diagnosed acute myeloid leukemia (AML) and myeloid dysplastic syndrome (MDS) samples, we found that high FAK expression was associated with unfavorable cytogenetics (P = 2 × 10⁻⁴) and relapse (P = 0.02) in AML. FAK expression was significantly lower in patients with FLT3-ITD (P = 0.0024) or RAS (P = 0.05) mutations and strongly correlated with p-SRC and integrinβ3 levels. FAK protein levels were significantly higher in CD34⁺ (P = 5.42 × 10⁻²⁰) and CD34⁺ CD38⁻ MDS (P = 7.62 × 10⁻⁹) cells compared to normal CD34⁺ cells. MDS patients with higher FAK in CD34⁺ cells tended to have better OS (P = 0.05). FAK expression was significantly higher in MDS patients who later transformed to compared with not transformed to AML and in AML patients who transformed from MDS compared with those with de novo AML. Co-culture with mesenchymal stromal cells (MSCs) increased FAK expression in AML cells. Inhibition of FAK decreased MSC-mediated adhesion/migration and viability of AML cells and prolonged survival in an AML xenograft murine model. Our results suggest that FAK regulates leukemia-stromal interactions and supports leukemia cell survival; hence FAK is a potential therapeutic target in myeloid leukemia.

Introduction

Focal adhesion kinase (FAK) coordinates a signaling network that orchestrates a diverse range of cellular processes through both kinase dependent and independent mechanisms (1). Once activated through integrin (ITG) ligation (2) or growth factor receptor-mediated (3) signals within the tumor microenvironment, FAK cooperates with SRC leading to SRC phosphorylation and subsequent FAK/SRC phosphorylation at multiple sites which relays the external signal into cells by activating multiple cell proliferating/survival pathways such as PI3K/AKT and MAPK and by regulating the expression of various genes (4). Most recently, FAK was found to be able to generate a tumor suppressive microenvironment by regulating chemokine transcription and promoting Treg recruitment and retention in squamous cell carcinoma (5) and inhibition of FAK renders pancreatic cancers responsive to checkpoint immunotherapy (6).

FAK is overexpressed and/or constitutively activated in many solid tumors, and both increased FAK expression and activity are associated with poor clinical outcomes (7–9) suggesting that FAK is a potentially critical target for cancer therapy. Indeed, FAK inhibition has been investigated extensively in solid tumors (10, 11). A recent study in a patient-derived xenograft model of mesothelioma showed that FAK inhibitor VS-4718 preferentially eliminated the cancer stem cells that were enriched following treatment with chemotherapeutic agents (11). VS-4718 treatment in combination with dasatinib prolonged survival in a model of B-ALL (12). Several FAK inhibitors, such as VS-4718, have entered clinical trials (NCT01849744, NCT02651727) in solid tumors. However, the proposed trial in AML (NCT02215629) was withdrawn largely due to lack of pre-clinical studies of the compound in acute leukemia.

The critical role of the bone marrow (BM) microenvironment in leukemia progression and drug resistance has only been recognized and investigated in the last decade (13). Although it is extensively studied in solid tumors, limited work was done with regard to the expression and function of FAK in myeloid leukemia. Recher et al. demonstrated that FAK is expressed in approximately 40% of AML patient samples and that high expression of FAK in AML was associated with enhanced blast migration, increased cellularity, and poor prognosis (14). Subsequently, Tavernier-Tardy et al. showed that FAK expression negatively associated with overall survival (OS) in AML, and patients overexpressing 2 to 3 factors of FAK, CXCR4 and VLA4 had a significantly shorter OS (15). Given the importance of the BM microenvironment in leukemogenesis, progression, and drug resistance, it is not surprising that leukemic cell adhesion molecules play critical roles in regulating the interactions with mesenchymal stromal cells (MSCs), an important BM component affecting patient outcome. Although FAK as a therapeutic strategy has not been explored in AML, other agents potentially impacting the BM microenvironment have been investigated. Pharmacological disruption of the CXCR4-CXCL12 interaction has demonstrated pre-clinically and in clinical trials that it is capable of mobilizing leukemia cells from the protective BM microenvironment and sensitizes to chemotherapy (16–19). Targeting VLA4/VCAM-1 signaling has also shown the potential of overcoming stroma-mediated chemoresistance in BM-resident leukemia cells (20), indicating that disruption of adhesion pathways in leukemic cells may have a number of clinical utilities.

Utilizing improved proteomic technologies, we examined the expression of FAK in a large cohort of newly diagnosed AML (n = 511) patient samples by reverse-phase protein array (RPPA) and correlated its expression with patient clinical characteristics. As approximately 30% of MDS patients will progress to AML and these patients with secondary AML carry poor prognosis, we also examined the expression of FAK and its clinical correlation in a large cohort of newly diagnosed MDS (n = 133) patient samples. We determined a role of FAK in AML cell lines in vitro and in vivo AML models by knocking down FAK with shRNA and/or using a FAK inhibitor VS-4718. These experiments offer a compelling rationale for the use of FAK inhibitors clinically in the AML setting.

Materials and Methods

Protein determination by RPPA in AML and MDS patient populations

FAK expression in large cohorts of AML and MDS patient samples and normal controls was determined by RPPA as previously described (21, 22). The AML patient population was the same as previously published (23, 24). Briefly, peripheral blood (PB) and BM specimens were collected from 511 newly diagnosed AML patients evaluated at The University of Texas M. D. Anderson Cancer Center (MDACC) from September 1999 to July 2010. A paired relapse sample was available for 47 patients. Of the 511 AML patients, 415 were treated at MDACC and were evaluable for outcome. For the MDS population, BM or PB specimens were collected from 133 newly diagnosed MDS patients who were evaluated at MDACC between 1999 and 2007. The 133 newly diagnosed MDS patients include 7 refractory anemia (RA), 8 refractory cytopenia with multilineage dysplasia (RCMD), 3 refractory cytopenia with multilineage dysplasia and ring sideroblasts (RCMD-RS), 67 refractory anemia with excess blast (RAEB), 3 refractory anemia with ringed sideroblasts (RARS), 1 myelodysplastic syndrome unclassified (MDS-U), 31 chronic myelomonocytic leukemia (CMML), and 13 others. The median age at diagnosis was 68.3 years (range 26.5–89.4). By international prognostic scoring system (IPSS), 13 low risk, 48 intermediate-1, 37 intermediate-2 and 18 high risk, 17 unknown. Cytogenetics of the population includes 72 diploid, 31 isolated 5q/7q, 4 isolated 20q, and 26 others. CD34⁺ cells (n = 191, 116 diagnosis and 75 sequential including 73 relapse) and CD34⁺CD38⁻ cells (n = 88, 47 diagnosis and 41 sequential including 40 relapse) were separated as previously reported (21). Sample collection and RPPA analysis were done according to the protocols approved by the MDACC Investigational Review Board (IRB). As controls, FAK expression level was also determined in CD34⁺ cells from 16, PB mononuclear cells from 9, and CD133⁺ cells from 5 normal donors (Table 1). FAK antibody was purchased from Cell Signaling Technology (Cat #3285; Danvers, MA).

Table 1.

FAK protein expression in MDS patient samples by RPPA.

	mean	SD	median	min	max	>norm (%)	<norm (%)	=norm (%)	Obs. Num
All MDS	−0.002	0.857	−0.078	−3.523	4.631	90.7	3.2	6.1	279
All CD34+	0.055	0.892	−0.052	−3.523	4.631	91.6	2.1	6.3	191
New CD34+	0.078	0.929	−0.089	−2.236	4.631	91.4	1.7	6.9	116
Rel CD34+	0.026	0.847	−0.018	−3.523	2.328	91.8	2.7	5.5	73
All SC	−0.125	0.767	−0.112	−3.099	1.793	88.6	5.7	5.7	88
New SC	−0.116	0.895	−0.057	−3.099	1.793	85.1	8.5	6.4	47
Rel SC	−0.136	0.607	−0.148	−2.307	1.292	92.5	2.5	5.0	40
CD34+ Control	−1.067	0.209	−0.998	−1.413	−0.734	6.2	6.2	87.5	16
CD133+	−0.739	0.293	−0.737	−1.001	−0.268	60.0	0.0	40.0	5
PBMC	0.497	0.404	0.461	−0.022	1.239	100.0	0.0	0.0	9

The protein level is in log2 scale of arbitrary unit. The normal range of the protein level is defined as 90% inter-percentile of protein level in CD34⁺ control samples. All MDS: All MDS samples analyzed; All CD34⁺: all CD34⁺ MDS samples; New CD34⁺: CD34⁺ samples from newly diagnosed MDS patients; Rel CD34⁺: CD34⁺ samples from relapsed MDS patients; All SC: all CD34⁺CD38⁻ MDS samples; New SC: CD34⁺CD38⁻ samples from newly diagnosed MDS patients; Rel SC: CD34⁺CD38⁻ samples from relapsed MDS patients; SD: standard deviation; > norm (%): percentage above normal range; < norm (%): percentage below normal range; = norm (%): percentage equal normal range; Obs Num: total observed cases per subgroup; PBMC: PB mononuclear cells.

Cell culture and treatment

OCI-AML3 was provided in 2004 by Dr. M. Minden (Ontario Cancer Institute, Toronto, ON, Canada). Molm13 and Molm14 were obtained in 2005 and 2014, respectively from the German Collection of Microorganisms and Cell Cultures (Braunschweig, Germany). MV4-11, purchased in 2006, and KG-1, KG-1a, HL-60, and THP1, all purchased in 1998, were obtained from the American Type Culture Collection (ATCC; Manassas, VA). Cell lines were validated by STR DNA fingerprinting using the AmpF_STR Identifier kit according to manufacturer’s instructions (Applied Biosystems, Cat#4322288). The STR profiles were compared to known ATCC fingerprints, and to the Cell Line Integrated Molecular Authentication database (CLIMA) version 0.1.200808 (http://bioinformatics.istge.it/clima/) (25). The STR profiles matched known DNA fingerprints or were identified as unique (OCI-AML3). Authenticated cells are stored under liquid nitrogen and are never kept in culture for more than 4 month. Cell lines were cultured in RPMI-1640 medium supplemented with 10% heat-inactivated fetal calf serum (FCS), 2 mmol/L L-glutamine, 100 U/mL penicillin, and 100 μg/mL streptomycin. Primary leukemic samples were acquired from AML patients with high blast counts, after informed consent following the institution approved protocol. Mononuclear cells were isolated from these samples by density-gradient centrifugation using a Lymphocyte Separation Medium (Corning, Manassas, VA) and cultured in α-MEM medium supplemented with 10% heat-inactivated FCS. Cells were kept at 37°C in a humidified atmosphere of 5% CO₂. Human MSCs isolated from BM samples obtained from healthy subjects as described previously (26, 27) and mouse MSC cell line MS5 were cultured in the same medium as AML patient samples. For co-culture experiments, leukemia cells were added to MSCs (AML cells: MSCs = 4:1 ratio) that were plated the night before and cultured as above. Leukemia cells cultured alone or co-cultured with MSCs were collected at 24 h for measuring FAK expression or treated with FAK inhibitor VS-4718 for further assessment. VS-4718 was purchased from Chemietek, Indianapolis, IN. The chemical structure for VS-4718 is presented in Supplementary Fig. S1.

Adhesion and migration assays

Migration of leukemia cells toward and adhesion to MSCs were determined as previously reported (28). Migration was determined at 6 h and adhesion at 24 h.

Cell viability assay

Viable cell numbers were determined by flow cytometry using counting beads (Life Technologies, Grand Island, NY). Apoptosis was estimated via flow cytometry measurement of phosphatidylserine externalization with annexin V staining (BD Biosciences, San Jose, CA). Cell membrane integrity was simultaneously assessed by 7-aminoactinomycin D (7AAD) exclusion in the annexin V-stained cells. To assess cell numbers and apoptosis in leukemia cells co-cultured with MSCs, CD45⁺ cells were counted and apoptotic cells were defined as annexin V⁺/AAD⁺ CD45⁺ cells.

Generating FAK knockdown cells

FAK was knocked down by lentiviral transduction using gene-specific shRNA transfer vectors (clones TRCN0000001620: shRNA1 targeting residues 3053-3073 and TRCN0000001621: shRNA2 targeting residues 2739–2759 on RefSeq NM_005607.4) (GE Dharmacon, Lafayette, CO). Lentivirus was prepared by cotransfecting HEK293T cells (ATCC) with an equimolar mix of transfer vector and packaging plasmids (psPAX2 and pMD2.G, gifts from Didier Trono, plasmids #12260 and 12259, Addgene; Cambridge, MA) using JetPrime transfection reagent as directed by the manufacturer (Polyplus, Illkirch, France). Fresh lentiviral supernatants were passed through 0.45 micron pore surfactant-free cellulose acetate membranes and then used immediately to infect leukemic cells by incubation overnight at 37°C under 5% CO₂. Infected cells were selected with puromycin (Invivogen, San Diego, CA) starting at 0.5 μg/mL. In parallel, cells were transduced using lentivirus delivering a non-specific control (pLKO.1-TRC control, a gift from David Root, plasmid 10879, Addgene). Knockdown was verified by western blot and real time RT-PCR analyses.

Western blot analysis

Western blot was carried out as previously described (28). Antibodies against FLT3, p-FLT3, and FAK were purchased from Cell Signaling Technology and p-FAK^y397 from Abcam (Cambridge, MA). β-actin was used as a loading control. Signals were detected using the Odyssey Infrared Imaging System (LI-COR Biosciences, Lincoln, NE) and quantified using the Odyssey software (version 3.0, LI-COR Biosciences).

CyTOF mass cytometry

Mononuclear cells from AML primary patient samples were stained with antibodies for cell surface and intracellular proteins (Table 2) and subjected to CyTOF mass cytometry as previously described (29, 30). Data were exported as FCS for subsequent analysis in SPADE (v3.0, http://pengqiu.gatech.edu/software/SPADE/).

Table 2.

Antibody panel for CyTOF analysis

	Target	Label	Clone	Vendor
1)	CD45	89Y	HI30	DVS-Fluidigm
2)	CD34	148Nd	4H11	eBioscience
3)	p-FAK (Y397)	175Lu	D20B1	Cell Signaling Technology
4)	FAK	141Pr	D2R2E	Cell Signaling Technology
5)	p‐ERK1/2	167Er	D13.14.4E	DVS‐Fluidigm
6)	p-AKT	159Tb	M89-61	BD Biosciences
7)	p-SRC(Tyr416)	164Dy	D49G4	Cell Signaling Technology
8)	p-STAT5(Y694)	150Nd	47	DVS-Fluidigm
9)	p-STAT3(Y705)	158Gd	4/P-STAT3	DVS-Fluidigm
10)	p-FLT3	174Yb	30D4	Cell Signaling Technology
11)	FLT3	162Dy	4G8	BD Biosciences

In vivo xenograft mouse model

Animal experiments were performed in accordance with a protocol approved by the Institutional Animal Care and Use Committee at MDACC. Molm14 cells (0.6 × 10⁶) stably expressing a dual luciferase-GFP reporter (Molm14-GFP/Luc) were injected via the tail vein into NOD/SCID IL2Rγ Null-3/GM/SF (NSGS) mice (Jackson Laboratory, Bar Harbor, ME). Once engraftment was confirmed by the IVIS-200 noninvasive bioluminescence in vivo imaging system (Xenogen, Hopkinton, MA), mice were either untreated or treated with VS-4718 twice a day at 75 mg/kg via oral gavage (n = 10/group) for 16 days. Leukemia burden was monitored by IVIS in vivo imaging, flow cytometric measurement of human CD45 cells (anti-human CD45 antibody, BD Biosciences) in mouse PB, and immunohistochemical staining for human CD45⁺ cells in mouse tissues (stained with anti-human CD45 antibody and visualized by Biotin-free Tyramide Signal Amplification System, both from Dako, Carpinteria, CA). Mouse survival was recorded.

Statistical analyses

Protein expressions determined by RPPA, correlations of FAK expression with other proteins in patient samples and with patient clinical outcomes in AML were analyzed as previously described (23, 24). For RPPA analysis of MDS patient samples, Pearson and Spearman correlation coefficient of the protein level and continuous variables were calculated. A log-rank test was used to evaluate the survival difference among the groups of subjects with different FAK protein level. All in vitro experiments were conducted in triplicate. Correlation coefficient for two sets of values was determined by Pearson (Microsoft Excel 2010). Mouse survival was analyzed using log-rank test. Statistical differences between groups were determined using paired Student’s t-test with a P ≤ 0.05 being considered statistically significant. Results are expressed as mean ± standard errors.

Results

Expression of FAK in AML patient samples and its clinical correlations

We determined FAK expression by RPPA in PB or BM samples obtained from a large cohort of newly diagnosed AML (n = 511). High FAK expression was associated with the unfavorable cytogenetic group (P = 2 × 10⁻⁴) (Figure 1A). The lowest FAK expression was seen in patients with favorable cytogenetic group including inv(16) (n = 19), t(8;21) (n = 15), and t(15;17) (n = 20) chromosome translocations (green circles, Figure 1B). Patients with −5, −7, and/or +8 (n = 100) expressed relatively higher FAK (red box, Figure 1B). Localization of FAK on chromosome 8 may explain high FAK levels in samples from AML patients with trisomy of chromosome 8. FAK expression was higher in relapsed compared to paired newly diagnosed samples (n = 47, P = 0.02) (Figure 1C). Interestingly, we found that FAK expression was significantly lower in patients with FLT3-ITD (n = 83, P = 0.0024) or RAS (n = 64, P = 0.05) mutations and tended to be lower in patients with FLT3-D835 mutation (n = 24, P = 0.06) (Figure 1D) suggesting functional compensation of these signaling pathways. Perhaps the hyperleukocytosis characteristic of FLT3-ITD AML is related to the decreased FAK levels observed.

Figure 1.

FAK expression determined by RPPA and patient characteristics in AML. A, FAK levels in samples from newly diagnosed AML patients of various cytogenetic groups. B, FAK levels in samples from newly diagnosed AML patients with various cytogenetics. C, FAK levels in paired newly diagnosed and relapsed AML patient samples (n = 47). D, FAK expression in AML patient samples with or without FLT3 or RAS mutations. E, FAK levels in newly diagnosed AML samples and CD34⁺ cells from normal controls. F, FAK expression and OS. G, FAK expression and remission duration. cyto, cytogenetic; FAV, favorable; INT, intermediate; UNFAV, unfavorable; ND, not determined; NEG, negative; and POS, positive.

Figure 1E shows the distribution of FAK expression in samples with newly diagnosed AML (n = 511) and CD34⁺ cells from normal controls (n = 21). Among the 415 evaluable patients, we did not observe significant differences in OS (P = 0.23) among patients whose FAK expression was lower, equal, or higher compared to that of normal controls although patients with lower FAK tended to do worse short-term and patients with higher FAK did worse long-term (Figure 1F). Interestingly, FAK expression levels significantly impact remission duration of the patients (P = 0.02). Patients with FAK levels the same as normal controls had the longest remission duration, followed by patients with lower FAK levels, the shortest being the patients with FAK levels higher than normal controls (Figure 1G). Thus, by multiple criteria, increased FAK expression is a poor prognostic marker in AML, and associates with unfavorable cytogenetic profiles.

Expression of FAK in MDS patient samples and its clinical correlations

RPPA was performed on 279 BM or PB samples collected from 133 newly diagnosed MDS patients and their follow up (Table 1). Elevated or decreased expression was defined as expression levels above or below 90% confidence interval of CD34⁺ normal specimen mean expression, respectively. Overexpression of FAK was seen in 90.7% of all samples compared with CD34⁺ normal specimens. Furthermore, overexpression of FAK was observed in 91.6% of CD34⁺ samples, including 91.4% in newly diagnosed samples and 91.8% in samples from relapsed patients. Overexpression of FAK was observed in 88.6% of CD34⁺CD38⁻ samples, including 85.1% in newly diagnosed samples and 92.5% in samples from relapsed patients (Table 1). We then compared the distribution of FAK level in the 116 newly diagnosed CD34⁺ samples and 47 newly diagnosed CD34⁺CD38⁻ samples with 16 normal CD34⁺ controls. FAK expression levels were significantly higher in both CD34⁺ (n = 116, P = 5.42 × 10⁻²⁰) and CD34⁺CD38⁻ (n = 47, P = 7.62 × 10⁻⁹) cells from MDS patient samples compared to CD34⁺ cells from normal controls (n = 16) (Figure 2A), suggests its involvement in MDS pathogenesis. Patients with higher FAK expression in CD34⁺ cells tended to have a better OS (P = 0.05) in newly diagnosed MDS (Figure 2B). This result was not found for FAK levels in CD34⁺CD38⁻ cells (not shown).

Figure 2.

FAK expression and patient characteristics in MDS. A, FAK levels determined by RPPA in CD34⁺ and CD34⁺CD38⁻ cells from newly diagnosed MDS patient samples and CD34⁺ cells from normal controls. B, FAK levels in CD34⁺ cells from newly diagnosed MDS patient samples determined by RPPA and patients’ OS. C, comparisons of FAK levels between MDS patients not transformed and later transformed to AML and between de novo and MDS transformed AML patients (Statistica, Version 12).

We next compared FAK levels in MDS patients not transformed to (n = 83) with those later transformed to AML (n = 26) and found significantly higher FAK in the latter group (P = 0.013 or 0.022 by F or Kruskal-Wallis H test, respectively; Figure 2C). We then compared FAK levels between de novo (n = 262) and MDS transformed (n = 127) AML and found significantly higher FAK (P < 10⁻⁶ by either F or Kruskal-Wallis H test) in MDS transformed AML patients (Figure 2C).

FAK signaling in AML

Since FAK is activated by ITG ligation or growth factors from the microenvironment and is known to relay the extracellular signal intracellularly through the FAK/SRC signaling cascade, we compared FAK expression with the expression of SRC and ITGs in the same AML RPPA samples. FAK expression was highly positively correlated with p-SRC^Y416 and ITGβ3 expression in a 3D surface blot (Figure 3A), suggesting a role for ITG/FAK/SRC signaling in AML cells.

Figure 3.

FAK signaling in AML cells. A, 3D surface plot of ITGβ3 against FAK and p-SRC^Y416. B, OCI-AML3 cells were treated with GM-CSF (100 units/mL) for 1, 4, and 24 h and FAK levels were determined by western blot. C, Cells from AML patient samples (n = 7) were cultured alone or co-cultured with mouse MSC cell line MS5 for 24 h. FAK protein level was determined by western blot. D, AML patient samples were treated with VS-4718 (48 h) and signaling proteins in bulk (CD45⁺) and CD34⁺ cells were determined by CyTOF/SPADE analysis; OCI-AML3 and Molm14 cells were treated with VS-4718 and protein levels were measured by western blot. WT, wild type; VS, VS-4718.

To better define growth factor and cell-cell interaction determinants of FAK signaling in AML, we first investigated whether FAK in AML cells is regulated by myeloid growth factors. GM-CSF induced FAK expression in OCI-AML3 cells (Figure 3B). MSCs secrete multiple growth factors/cytokines and are an important component of the BM microenvironment that is critical for the homing and survival of leukemia cells. We next examined whether FAK expression in AML cells is increased by MSC co-culture. We cultured leukemic mononuclear cells obtained from patients with AML with or without a murine MSC cell line (MS5) for 24 h and determined the expression of FAK expression by western blot. In an evaluation of 7 AML patient samples, leukemic cells co-cultured with MSCs expressed significantly more FAK than without (P = 0.016) (Figure 3C), suggesting that the microenvironment modulates leukemia cell function in part through activating FAK signaling.

To determine FAK signaling, we treated primary AML cells from two patients, one with wild type FLT3 and one with FLT3-ITD with FAK inhibitor VS-4718 and measured cell signaling in blasts and CD34⁺ subset by CyTOF mass cytometry, aided by SPADE analysis. We found that VS-4718 decreased p-FAK, FAK, and p-SRC as well as FLT3 downstream targets p-AKT and p-STAT5, but not p-FLT3 and FLT3 in bulk and CD34⁺ cells of both samples (Figure 3D) suggesting that VS-4718 inhibits cell survival signaling through FAK, not FLT3 inhibition. We also treated OCI-AML3 and Molm14 cells with VS-4718 and found that VS-4718 reduced FAK in both AML cell lines and did not affect p-FLT3 and FLT3 levels in Molm14 cells with FLT3-ITD (Figure 3D).

Inhibition of FAK blocks leukemia-stromal interactions

To determine if FAK has a role in leukemia-microenvironment interactions, we conducted adhesion and migration assays. Inhibition of FAK by VS-4718 decreased the adhesion (Figure 4A) and migration (Figure 4B) of OCI-AML3 cells to BM-derived MSCs suggesting that inhibition of FAK blocks leukemia-stroma interactions. Note that reduced adhesion and migration of AML cells to MSCs were not caused by FAK inhibition-induced cell death as at the doses and time points for the assay, no marked decreases in cell viability were observed (Figure 4A and B). Furthermore, inhibition of FAK expression by ShRNA in OCI-AML3 cells decreased the adhesion and migration of these cells in a dose-dependent manner to human BM-derived MSCs (Figure 4C) supporting that FAK mediates the interactions.

Figure 4.

Inhibition of FAK decreases AML-MSC interactions. A, OCI-AML3 cells added to human BM-derived MSCs plated the night before and treated with VS-4718. At 24 h, total cells (floating and attached) were counted and attached over total cells were calculated. B, Migration of OCI-AML3 towards MSCs was determined 6 h after cells were treated with VS-4718 using trans-well plates. Viability of both floating and adherent or migrated and non-migrated leukemia cells was determined by measuring annexin V/7AAD positivity in CD45⁺ cells. C, Adhesion and migration of FAK knockdown and vector control OCI-AML3 cells to MSCs.

Inhibition of FAK decreases viability and induces apoptosis of leukemia cells

To determine the role of FAK in leukemia cell growth and survival, we first determined the expression of FAK and p-FAK^y397 in AML cell lines (Figure 5A). We then treated these cell lines with FAK inhibitor VS-4718. VS-4718 potently decreased viable cell numbers in all the leukemia cell lines tested (IC₅₀ range from 90 nM to < 2 μM) (Figure 5B). Six out of eight cell lines tested had IC₅₀ at nM levels and VS-4718 was active even under MSC co-culture conditions with most of these cell lines (Figure 5B, top panel). IC₅₀ did not correlate with the levels of p-FAK (R = 0.26, P = 0.54). However, when AML cells were co-cultured with MSCs, IC₅₀ was significantly correlated with the level of p-FAK (R = 0.73, P = 0.04) (Figure 5B, bottom right panel) suggesting a role of FAK in mediating BM microenvironment signals to leukemia cells. VS-4718 induced cell death in leukemia cell lines with variable potencies, even in cells co-cultured with MSCs (Figure 5C).

Figure 5.

VS-4718 decreases cell viability and induces apoptosis in leukemia cell lines. A, Expression of p-FAK and total FAK in various leukemia cell lines. B, VS-4718 decreases viable cell counts of various leukemia cell lines cultured alone or co-cultured with human BM-derived MSCs (the top panel); the lower panel shows correlation of p-FAK expression and IC₅₀ of VS-4718 in these cells. C, VS-4718 induces apoptosis of various leukemia cell lines cultured alone or co-cultured with MSCs. Cells were treated with VS-4718 for 48 h.

Inhibition of FAK by VS-4718 exerts antileukemia activity in vivo in human AML xenografted NSGS mice

To explore the role of FAK in an in vivo model of AML, NSGS mice were injected with Molm14-GFP/Luc cells and treated with 75 mg/kg VS-4718 twice a day via oral gavage for 16 days following the experimental scheme showing in Figure 6A. VS-4718 treated mice had decreased leukemia burden by in vivo imaging, lower human CD45 positivity in PB by flow cytometry analysis, and less tissue infiltration of leukemia cells by immunohistochemistry staining of human CD45⁺ cells (Figure 6B). VS-4718 treated mice survived significantly longer than the untreated controls (medium survival 27 versus 20 days, P = 0.0003) (Figure 6C). No weight loss or other treatment related toxicities were observed. One mouse in the treatment group died of causes unrelated to the experiment.

Figure 6.

Inhibition of FAK by VS-4718 exerts antileukemia activity in vivo. NSGS mice injected with Molm14-GFP/Luc cells were treated with VS-4718. A, Experimental scheme. B, IVIS in vivo image, flow cytometric measurement of human CD45 cells in mouse PB, and immunohistochemistry staining of human CD45 cells in various tissues (20X) of the mice treated with VS-4718 or untreated controls. C, Mouse survival curves.

Discussion

We demonstrate in this study that FAK is expressed in AML patient samples and that high expression is associated with unfavorable cytogenetics. FAK, activated in AML cells by the BM microenvironment, promotes leukemia/stroma interaction and supports the survival of leukemia cells. FAK inhibition decreases viability of leukemia cells in vitro and prolongs mouse survival in a human AML xenograft model.

There were significantly higher FAK levels in relapsed versus paired newly diagnosed AML samples, suggesting that high FAK expression contributes to drug acquired or intrinsic resistance in AML. FAK was reported by others to predict poor prognosis in patients with AML (14, 15). Although we observed a significant difference in remission duration in AML patients with different FAK levels (the shortest in patients with FAK levels higher than normal controls), we did not find significant differences in OS among these patients based on their FAK levels. When the analyses were performed in subsets of patients with intermediate cytogenetics, intermediate cytogenetics with or without FLT3-ITD mutations, or unfavorable cytogenetics, we did not find a FAK level impact OS either. These findings may in part be due to the highly heterogeneous genetic and epigenetic backgrounds of AML patients and to the fact that leukemia cell signaling is regulated and compensated by multiple interconnected signaling pathways. For example, we found that FAK expression was significantly lower in patients with FLT3-ITD or RAS mutations suggesting that FAK signaling and FLT3/RAS signaling may compensate each other. Although FAK plays multiple roles to support cancer cell survival, it by itself may not be sufficient to predict AML patient OS. Reports from Tavernier-Tardy et al. (15) demonstrated that AML patients with combined overexpression of 2 or 3 adhesion proteins including CXCR4, VLA4, and FAK had a significantly shorter OS. In that study, flow cytometry was used to determine functional cell surface CXCR4 and VLA4 levels. RPPA analysis cannot detect cell localization and CXCR4 and VLA4 were not included in our RPPA panels.

The ITG/FAK/SRC signaling cascade is a well-established pathway that translates environmental signaling and activates multiple intracellular signaling pathways to support cell growth and survival. ITGβ3 was identified by in vivo RNAi screening to be essential for leukemia cells but not for normal hematopoietic stem/progenitor cells (31). We observed that AML samples expressing high ITGβ3 simultaneously expressed high levels of FAK and p-SRC^Y416. The data suggest a functional role of ITG/FAK/SRC signaling in AML cells. We demonstrate that FAK in AML cells is induced by cytokines and MSCs and that FAK regulates AML-MSC interactions. Interestingly, it was previously reported that FAK in AML cells modulates the function of MSCs (32), suggesting that FAK can facilitate the bidirectional crosstalk between leukemic cells and the BM microenvironment.

Although VS-4718 reduced cell viability in all tested AML cell lines, Molm14, Molm13, and MV4-11 all harboring the FLT3-ITD mutation were the most sensitive. It was previously shown that in addition to FAK, VS-4718 has activities against other kinases, particularly FLT3 by in vitro kinase profiling (33). We do not exclude that VS-4718 can inhibit FLT3 and other kinases in a biochemical assay. However, at the doses used, VS-4718 was able to decrease FAK but not p-FLT3 or FLT3 in AML cell lines or patient samples by either western blot or CyTOF analysis. Interestingly, VS-4718 decreased FLT3 downstream signaling proteins p-AKT and p-STAT5, independent of FLT3 mutation status. FLT3 mutated cells may depend more than wild type cells on these signaling proteins for growth and survival.

Inhibition of FAK by VS-4718 for only 16 days significantly prolongs survival of leukemia engrafted mice which may possibly be extended by prolonged inhibition. Although inhibition of FAK significantly prolonged survival of leukemia engrafted mice and greatly suppressed cell growth in all the cell lines tested (IC₅₀ < 2 μM) and was also effective under MSC co-culture at a clinically achievable dose (about 3 μM) in vitro, it was less potent in inducing cell death suggesting combined strategies are needed for enhancing cytotoxic efficacy which is currently under investigation. In addition, a recent study demonstrated that FAK splice variants are overexpressed in stem/progenitor cells of AML patients with poor prognosis and maintain primitive AML cells suggesting that targeting FAK has the potential to eliminate AML stem/progenitor cells (34). Inhibition of FAK in combination with agents that potently eliminate bulk leukemia cells may be more effective in the eradication of various leukemic subpopulations supported by a recent study in Ph+ B-ALL showing synergism of combined inhibition of FAK and Bcr-Abl tyrosine kinase (12).

Although FAK expression tended to positively impact survival in MDS, overexpression of FAK in CD34⁺ cells from MDS patients may also suggest that FAK signaling is involved in the pathogenesis of the disease. It was reported that the 5′-flanking region of FAK promoter contains several potential transcription factor binding sites including NFκB binding sites (35). Inhibition of NFκB decreased and activation of NFκB by TNF-α induced FAK transcription. The presence of abnormal levels of cytokines and chemokines such as TNF-α in MDS patients have been extensively documented (36). Elevated NFκB signaling has been correlated with the progression of MDS (37). It is possible that aberrantly high levels of chemokines and cytokines and upregulated NFκB signaling induce overexpression of FAK in MDS, but the association with disease severity and progression has yet to be determined. Interesting, we found FAK expression was significantly higher in MDS patients who later transformed to compared with not transformed to AML and in AML patients who transformed from MDS compared with those with de novo AML suggesting functional importance of FAK expression in AML and MDS patients.

Collectively, our data suggest that FAK regulates leukemia-stromal interactions and supports leukemia cell survival, hence a potential therapeutic target in myeloid leukemia. Combination strategies may improve the efficacy of FAK inhibition in AML which is currently under investigation.