Loss of Function of DOCK4 in Myelodysplastic Syndromes Stem Cells is Restored by Inhibitors of DOCK4 Signaling Networks

Dedicator of cytokinesis 4 (DOCK4) is one of the members of the 11 DOCK family proteins, which are conserved across different mammalian species (1). It is a large protein of approximately 225 KDa with multiple signaling/protein–protein interaction domains (2). The gene for DOCK4 protein is located in the q arm of chromosome 7. Recent studies have highlighted the importance of normal levels of DOCK4 expression across multiple tissue types in maintaining cellular homeostasis (3–7). Mutations or reduced expression of DOCK4 can lead to malignancies in prostate, breast, lung, brain, and blood tissues as well as solid tumor metastasis (8–11). Its known functions include regulation of motility via Rac1 GTPases and actin cytoskeleton (12, 13). However, very little is known with respect to the impact of reduced levels of DOCK4 expression within the stem cell compartment. Using healthy blood stem cells and blood stem cells expressing reduced levels of DOCK4, [as seen in the malignant blood disorder myelodysplastic syndromes (MDS)], we identified downstream signaling networks regulated by DOCK4 and functional implications of reduced DOCK4 expression within the blood stem cell compartment.

MDS are clonal stem cell disorders, where DOCK4 expression is reduced because of either deletion of chromosome 7q or mutations or promoter hypermethylation in DOCK4 gene (14). Patients with this disorder experience multi-lineage dysplasia and peripheral cytopenia (15). Our previous studies have shown that reduced levels of DOCK4 lead to dysplastic erythropoiesis and restoring DOCK4 expression in primary MDS erythroblasts improved erythroid differentiation (14, 16). This work focused on the impact of DOCK4 aberrations on postlineage-committed erythroid progenitors and not in hematopoietic stem cells (HSC) even though DOCK4 is highly expressed in early-stage stem cells. Moreover, downstream signaling networks regulated by DOCK4 in prelineage-committed HSCs are not known.

In this article, we used a population of early-stage human primary HSCs, as defined by expression of surface marker proteins CD34 and CD90 to identify downstream signaling networks regulated by DOCK4. Furthermore, we determined the functions of DOCK4 and the consequences of reduced DOCK4 expression in early HSCs. These studies revealed several phosphatases and kinases are regulated by DOCK4. We demonstrate that DOCK4 regulated tyrosine phosphorylation of a large number of signaling proteins resulting in significant increases in global phospho-tyrosine levels. Using mass spectrometry phosphoproteomic approaches, we precisely identified LYN (tyrosine protein kinase-Lyn), PTPN6 [tyrosine-protein phosphatase nonreceptor type 6, also known as Src homology region 2 domain-containing phosphatase-1 (SHP1)], and INPP5D [(phosphatidylinositol 3,4,5-trisphosphate 5-phosphatase 1, also known as Src homology 2 domain containing inositol polyphosphate 5-phosphatase 1 (SHIP1)] as most significantly impacted (hyper tyrosine-phosphorylated) proteins. LYN kinase directly phosphorylated phosphatases SHP1 and SHIP1 at tyrosine sites 536 and 1021, respectively. Low DOCK4 levels led to increased stem cell migration, which was blunted in the presence of pharmacologic inhibitors of LYN or SHP1 or SHIP1. In DOCK4-deficient MDS patient samples [-7/(del)7q], we observed increased numbers of CD34⁺/CD45⁺ cells in circulation. Finally, we demonstrate that pharmacologic inhibition of SHP1 in DOCK4-deficient HSCs from patients with MDS can improve erythroid differentiation. Altogether, this study has identified a new signaling network that can be leveraged to potentially overcome the functional defects that arise because of reduced expression of DOCK4 in MDS.

CD34⁺/CD90⁺ HSCs were purified from mobilized peripheral blood of healthy donors purchased from Key Biologicals, Inc. using a CliniMACS (Miltenyi Biotec, Inc) device (16–19). Purified cells were cultured in Stemspan SFEM II (Stemcell Technologies Inc) supplemented with 50 ng/mL thrombopoietin (TPO), 50 ng/mL stem cell factor (SCF), 50 ng/mL Fms-related tyrosine kinase-3 ligand (FLT3-L), 50 ng/mL IL3, and 50 ng/mL IL6 until used for subsequent experiments. All the cytokines were purchased from R&D Systems. Benzidine-hematoxylin staining of the cytospun HSCs was performed as described previously (20).

In the experiments involving cytokine deprivation and exposure, HSCs were washed twice to get rid of cytokines and cultured in Iscove's modified Dulbecco's medium (IMDM; Lonza) containing 1% (v/v) BSA fraction V (Thermo Fisher Scientific) for 3 hours. Following this, the cells were exposed to cytokines for 15 minutes at a concentration of 250 ng/mL. Specific cytokines and cytokine cocktail used in experiments are described in figure legends appropriately.

TF1 erythroleukemia cells were purchased from ATCC and cultured in RPMI (Gibco) containing 10% (v/v) FBS (Life Technologies), supplemented with 4 ng/mL GM-CSF. HEK293 cells were cultured in DMEM (Gibco) containing 10% (v/v) FBS.

All studies involving human subjects were conducted in accordance with U.S. Common Rule. Peripheral blood or bone marrow aspirates from each patient with MDS were obtained after institutional review board approval and informed written consent. CD34⁺ HSPCs from patients with MDS were purified as described previously using CD34⁺ Selection Kit purchased from Stemcell Technologies, Inc (16, 21).

Control (#D-001810-10, Dharmacon Inc.) or DOCK4 siRNA (Dharmacon Inc.) was nucleofected into CD34⁺ HSCs using CD34⁺ Cells Nucleofection Kit (Lonza) according to the manufacturer's protocol. Briefly, 2.5 × 10⁶ cells were nucleofected with 300 nmol/L siRNA using the U-08 program in the Nucleofector II machine. Following nucleofection, cells were cultured in Stemspan SFEM II supplemented with 50 ng/mL TPO, 50 ng/mL SCF, 50 ng/mL FLT3-L, 50 ng/mL IL3, and 50 ng/mL IL6 until used for subsequent experiments. Similarly, in the experiments involving knockdown of LYN, SHIP1, or SHP1, respective smartpool siRNA (Dharmacon Inc.) were used. In the experiments involving TF1 cells, control or DOCK4 siRNA was nucleofected using Nucleofection Kit T (Lonza) according to the manufacturer's protocol.

HSCs that were briefly cultured for 3 hours were used to knockdown DOCK4 and recultured for 24 hours prior to lysing both DOCK4-knockdown (50%) and DOCK4-intact cells using phosphorylation lysis buffer [50 mmol/L HEPES (pH 7.3), 150 mmol/L sodium chloride, 1 mmol/L EDTA, 1.5 mmol/L magnesium chloride, 100 mmol/L sodium fluoride, 10 mmol/L sodium pyrophosphate, 200 μmol/L sodium orthovanadate, 10% glycerol, 0.5% Triton X-100, and 1 mmol/L phenylmethylsulfonyl fluoride) as described previously (22). Protein concentration in the supernatants was determined by BCA assay. A total of 450 μg of total protein for each of the two samples were reduced and alkylated prior to trypsin digestion. Phosphopeptides from the digests were enriched using TiO2 beads and fractionated them by high pH reverse phase into four fractions each (23). Each fraction was desalted prior to LC/MS analysis. Nano LC/MS-MS analyses were performed with a 75 μm × 10.5 cm PicoChip column packed with 3 μm Reprosil C18 beads with Dionex UltiMate 3000 Rapid Separation nanoLC coupled to a Q Exactive HF Hybrid Quadrupole-Orbitrap Mass Spectrometer (Thermo Fisher Scientific Inc). A 150 μm × 3 cm trap packed with 3 μm beads was installed in-line. Peptides were separated in 120-minute gradient. Data were acquired in data-dependent MS-MS mode with a top-15 method. Dynamic exclusion was set to 20 seconds and charge 1+ ions were excluded. MS1 scans were collected from 300–2,000 m/z with resolving power equal to 60,000. The MS1 automatic gain control (AGC) was set to 3 × 10⁶. Precursors were isolated with a 2.0 m/z isolation width, and the HCD normalized collision energy was set to 30%. The MS2 AGC was set to 1 × 10⁵ with the resolving power set at 30,000. Phosphopeptides in which the phosphorylation sites that can be assigned to a single amino acid with 75% probability or better in at least one sample were filtered. Robust z-scores were computed from the log₂-fold changes to compare knocked down cells versus control cells and defined a z-score of ≥1.5 as upregulated and ≤−1.5 as downregulated.

Multiple doses of recombinant active LYN kinase (SignalChem) were incubated along with 250 ng recombinant active SHIP1 (SignalChem) or recombinant SHP1 (SignalChem) in the presence of ATP at a final concentration of 100 μmol/L. The total volume of each in vitro reaction was 25 μL using kinase dilution buffer I (SignalChem) and deionized water. The samples were incubated in a 30°C water bath for 15 minutes. The kinase reaction was stopped by adding 8.3 μL 4× Laemmli Buffer (Bio-Rad) and boiling the samples for 5 minutes. The samples were then analyzed by immunoblotting. In the samples where inhibitor was added, the LYN/Src inhibitor, RK20449 (Selleck Inc.) was added to a final concentration of 500 nmol/L.

LYN kinase activities in the presence of flag-tagged DOCK4 or recombinant DOCK4 C-terminus were measured using the Universal Tyrosine Kinase Activity Assay Kit (Takara) according to the manufacturer's guidelines. In experiments involving flag-tagged DOCK4, we ectopically expressed Flag-tagged full-length DOCK4 in HEK293 cells and immunoprecipitated flag-tagged DOCK4 using anti-Flag coated protein G Dynabeads (Thermo Fisher Scientific) according to the manufacturer's protocol. Following this, increasing amounts of immunoprecipitated flag-tagged DOCK4 as indicated was incubated with 20 ng of active recombinant LYN kinase or active recombinant JAK2 kinase. Similarly, in experiments involving recombinant DOCK4 C-terminus, increasing concentration of DOCK4 C-terminus (Proteintech) as indicated was incubated with 20 ng of active recombinant LYN kinase or active recombinant JAK2 kinase.

Flag-tagged recombinant DOCK4 C-terminus was generated by cloning DOCK4 C-terminus using primers DOCK4 C-terminus forward primer - 5′-GGTGCCATGGGCCACCATCACCACCATCATCACCACCATCACCCTTTGTTGTCTGACAAACACAC and DOCK4 C-terminus reverse primer - 5′-CACCCTCGAGTCACTTGTCGTCATCGTCTTTGTAGTCTAACTGAGAGACCTTGCGG into pET15b plasmid purchased from Novagen. Following cloning, we produced recombinant flag-tagged DOCK4 C-terminus according to the manufacturer's protocol. A total of 100 ng recombinant LYN and 400 ng of flag-tagged recombinant DOCK4 C-terminus was incubated overnight along with anti-IgG- or anti-Flag–coated magnetic beads in a rotating shaker at 4°C. Immunoprecipitation followed by immunoblot analysis was performed as described earlier.

Flag-tagged full-length DOCK4 plasmid (ref. 8; gift from Dr. Linda Van Aelst, Cold Spring Harbor Laboratory) was cotransfected along with GFP-tagged LYN plasmid (ref. 24; gift from Dr. Anna Huttenlocher, University of Wisconsin, Madison, WI) or GFP-tagged SHIP1 plasmid (ref. 25; gift from Dr. Aaron Marshall, University of Manitoba, Winnipeg, Canada) or GFP-tagged SHP1 plasmid (ref. 26; gift from Dr. Stephen Shaw, NCI) using Lipofectamine 2000 Reagent (Thermo Fisher Scientific) according to the manufacturer's protocol. Twenty-four hours later cells were harvested, lysed, and protein concentration was calculated using Lowry–Bradford assay. Protein G Dynabeads (Thermo Fisher Scientific) were coated with indicated antibodies and immunoprecipitation was carried out according to the manufacturer's protocol. GFP-trap beads (ChromoTek Inc) were used to immunoprecipitate GFP-tagged proteins. The eluted samples after immunoprecipitation were analyzed by immunoblotting as described earlier.

Proteins that were hyper-tyrosine phosphorylated greater than 1.5-fold and tyrosine phosphorylated proteins identified only in DOCK4 knockdown samples were subjected to functional annotation analysis available in the Database for Annotation Visualization and Integrated Discovery (DAVID website, http://david.abcc.ncifcrf.gov/) Gene ontology option GOTERM_BP_ALL was selected and a functional annotation chart generated.

Control and DOCK4-knockdown HSCs were immobilized on Alcian blue–coated coverslips and stained for actin with Phalloidin (Thermo Fisher Scientific) as described previously (16). Images were captured using Leica STED-SP5 confocal microscope at 63× magnification. Fiji ImageJ software was used to quantify the promigratory cells using the circularity feature. At least 100 cells from four random fields were analyzed.

HSC migration assays were performed using transwells. Briefly, 150,000 HSCs were suspended in 100 μL of starvation media [IMDM containing 1% (v/v) BSA fract V] and was added to the top chamber of the 5-μm pore transwell insert (24-well plate format transwell, Corning). A total of 0.5 mL of starvation media with various concentrations (0–100 ng/mL) of chemokine stromal-derived factor-1 alpha was added to the bottom chamber. The transwell plates were incubated for 4 hours in a 37°C 5% CO₂. The transwell inserts were carefully removed and the migrated cells in the bottom chamber were resuspended. Samples were obtained from the bottom chamber and stained with Acridine orange and propidium iodide nuclear dyes (AO-PI dye, Nexcelom). The stained samples were enumerated using Nexcelom Auto 2000 Cell Counter (Nexcelom). All the migration experiments were performed in triplicates. In the experiments involving inhibitor treatments [LYN/Src inh. – RK20449 (Selleck Inc.), SHIP1 inh. – 3AC (EMD Millipore Inc.), and SHP1 inh. – TPI-1 (Cayman Inc.)], the cells were exposed to inhibitors for 1 hour in IMDM containing 1% BSA media prior to adding to the top chamber of the transwell. Migration assays with TF1 cells were performed as described previously (27).

A total of 1,500 HSCs from healthy or MDS patients were cultured in methylcellulose (Stemcell Technologies Inc. catalog no. H4434) supplemented with 2 U/mL erythropoietin for 14–16 days in the presence/absence of SHP1 inhibitor, 4 μmol/L TPI-1 (Cayman Inc.). DMSO was used as a vehicle control. After 14–16 days, the colonies were enumerated and imaged. Images of the colonies were captured using Olympus microscope at 10× magnification. Following colony enumeration, cells were collected from the methylcellulose plates and hemoglobin levels were quantitated using Hemoglobin ELISA Kit (Abcam #ab157707) according to the manufacturer's guidelines.

The error bars are computed as mean ± SEM. Student t tests were performed to determine the statistical significance between the samples. In experiments involving dose responses, one-way ANOVA test was performed to determine statistical significance. Experiments were from three biological replicates, and values with P < 0.05 were considered statistically significant.

We first examined steady state expression levels of DOCK4 in HSCs and in lineage-committed progenitors, which showed a relatively high level of expression in HSCs but low levels in committed progenitors (Fig. 1A). We confirmed that cells used in our studies were highly enriched for the HSC phenotype by determining the CD34⁺/CD90⁺ expression and cell morphology (Supplementary Fig. S1A–S1C), which showed 84% cells were double positive and greater that 99% were positive for CD34 early stem/progenitor cell marker. We then knocked down DOCK4 in these cells by siRNA, which enabled us to reduce the levels by 50% consistently in multiple primary samples as determined by qPCR (Fig. 1B). We then determined HSC response to cytokines in cells that are expressing DOCK4 at 50% of their normal levels, by exposing them to a cocktail of stem cell cytokines (TPO, SCF, Flt3L, IL3, and IL6; ref. 28) after a short deprivation of cytokines. We also exposed cells expressing DOCK4 at their normal levels to the same cocktail before harvesting cells and performing immunoblotting against an anti-phospho tyrosine antibody. These experiments revealed increased tyrosine phosphorylation of a large number of proteins in DOCK4 knockdown samples compared with the ones that had normal levels of DOCK4 expression (Fig. 1C and D). This increase was observed regardless of cytokine stimulation suggesting DOCK4 levels alone were sufficient to elicit global phosphorylation response. Levels of protein under each condition were equivalent as reflected by no difference in AKT phosphorylation or GAPDH in DOCK4-knockdown cells and nontargeting control cells (Fig. 1C and E). Collectively, these data demonstrated that reduced levels of DOCK4 increased global tyrosine phosphorylation and suggested that DOCK4 functions as a signaling intermediate downstream of several cytokine receptors in HSCs.

To identify the proteins that were impacted in their phosphorylation because of reduced DOCK4 expression, we performed phosphoproteomic analysis by mass spectrometry. These experiments uncovered a large number of phosphopeptides belonging to mostly kinases and phosphatases that were hyperphosphorylated in cells that expressed low levels of DOCK4. Among them, SHP1, SHIP1, and LYN exhibited greatest increase in tyrosine phosphorylation (Table 1). We then focused on LYN kinase and phosphatases SHP1 and SHIP1 for further study because these enzymes have been extensively studied during blood cell development (29). We used commercially available phospho-specific antibodies against each of the three proteins to confirm our mass spectrometry results, which showed LYN, SHIP1, and SHP1 proteins were phosphorylated at tyrosine sites 397, 1021, and 536, respectively (Fig. 2A–D). These results were also confirmed in TF-1 erythroleukemia cells, which are at early phase (CD34⁺) of blood cell development (Supplementary Fig. S2A–S2F).

Next, we wanted to determine whether LYN kinase was directly responsible for phosphorylating the two phosphatases, SHIP1 and SHP1 as a result of LYN kinase activation because of reduced expression of DOCK4 (Supplementary Fig. S2G). To determine whether SHIP1 and SHP1 are direct targets of LYN kinase, we designed cell-free in vitro kinase assays, where either recombinant full-length SHIP1 protein or recombinant full-length SHP1 protein was incubated with active form of recombinant LYN kinase in a biochemical assay in the presence of a kinase buffer. A parallel assay using the same substrates but recombinant JAK2 as the kinase enzyme was also performed as a control. The reaction products were then analyzed by immunoblotting using an anti-phospho SHIP1 (Y1021) antibody or an anti-phospho SHP1 (Y536) antibody. The results of these experiments showed that LYN kinase phosphorylated SHIP1 at tyrosine 1021 and SHP1 at tyrosine 536 in a dose-dependent manner, whereas JAK2 kinase did not show such phosphorylation of SHIP1 or SHP1 (Fig. 2E and F; Supplementary Fig. S2H and S2I). Furthermore, the presence of the LYN/Src inhibitor, abrogated phosphorylation of both SHIP1 and SHP1, reenforces the specificity of the in vitro kinase assays (Fig. 2E and F; Supplementary Fig. S2H and S2I). To determine whether SHIP1 and SHP1 are substrates of LYN kinase in HSCs, we inhibited LYN using the LYN/Src inhibitor in cultured HSCs. Immunoblotting analysis showed that SHIP1 (Y1021) and SHP1 (Y536) phosphorylation decreased in a dose-dependent manner when LYN was inhibited (Fig. 2G; Supplementary Fig. S2J–S2L). Taken together these results demonstrated that reduced expression of DOCK4 initiate a sequential phosphorylation events impacting the signaling cascade involving LYN kinase, SHIP1, and SHP1.

Next, we wanted to ascertain whether DOCK4 directly interacts with LYN kinase, as well as SHIP1 and SHP1 phosphatases in addition to whether DOCK4 regulates LYN kinase activity. To test this, we ectopically expressed Flag-tagged full-length DOCK4 and GFP-tagged full-length LYN in HEK293 cells and performed reciprocal coimmunoprecipitation experiments coupled with immunoblot analysis. Immunoprecipitation of Flag-tagged DOCK4 followed by using anti-GFP/LYN antibody for detection, as well as immunoprecipitation with GFP-trap beads and immunoblot analysis with anti-FLAG/DOCK4 antibody showed that DOCK4 directly interacted with LYN (Fig. 2H). Similarly, we ectopically expressed Flag-tagged full-length DOCK4 and GFP-tagged full-length SHIP1 in HEK293 cells and performed reciprocal coimmunoprecipitation experiments coupled with immunoblot analysis. Immunoprecipitation of Flag-tagged DOCK4 followed by using anti-GFP/SHIP1 antibody for detection, as well as immunoprecipitation with GFP-trap beads and immunoblot analysis with anti-FLAG/DOCK4 antibody showed that DOCK4 directly interacted with SHIP1 (Fig. 2I). However, when similar experiments were performed using GFP-tagged full-length SHP1, we did not detect a direct interaction between DOCK4 and SHP1 suggesting that changes in phosphorylation seen in SHP1 is indirect and most likely only through LYN kinase (Fig. 2J).

We then determined whether LYN kinase activity is regulated by DOCK4 levels. To accomplish this, we set up an in vitro kinase assay where varying amounts recombinant DOCK4 protein was incubated with recombinant active form of LYN kinase and measured LYN kinase activity using a commercially available ELISA. These experiments showed that increasing amounts of DOCK4 protein decreased LYN kinase activity (Fig. 2K and L; Supplementary Fig. S2M and S2N). However, when recombinant JAK2 was used as a control no modulation of JAK2 activity was observed when increasing amounts of DOCK4 was used in the assay (Fig. 2L; Supplementary Fig. S2M).

To identify the functional implications of increased tyrosine signaling in DOCK4-deficient HSCs, we performed in silico DAVID analysis using the list of proteins that were identified by mass spectrometry to be highly phosphorylated (Table 1). This analysis revealed cell migration as one of the highly enriched biological pathways (Fig. 3A). We tested this prediction experimentally by carrying out transwell migration assays using HSCs expressing normal levels and reduced levels of DOCK4, which showed increased rates of migration of cells expressing reduced levels of DOCK4 (Fig. 3B). Similar results were also observed in TF1 cells following DOCK4 knockdown (Supplementary Fig. S3A). Because F-actin in the cytoskeleton play a key function in cell migration, we examined for changes in the F-actin network in HSCs expressing reduced levels of DOCK4 and compared them with HSCs expressing normal levels of F-actin. These experiments revealed significantly increased numbers of cells displaying promigratory features such as cell spreading and bundled F-actin in the leading edges of the cells (Fig. 3C). In follow-up experiments we quantified the extent of cell spreading in DOCK4-knocked down (50% knockdown) cells and cells expressing normal levels of DOCK4 by computing circularity values using the built-in circularity feature available in the Fiji software. This quantitation revealed that cell spreading was significantly increased when DOCK4 levels were reduced in HSCs as indicated by the decrease in circularity values (Fig. 3D). Similar promigratory features were also observed in TF1 cells when DOCK4 levels were reduced (Supplementary Fig. S3B).

Given that LYN, SHIP1, and SHP1 are downstream of DOCK4, we next determined whether the LYN kinase and its two downstream targets SHIP1 and SHP1 were involved in regulating HSC migration. We reduced the expression levels of LYN, SHIP1, or SHP1 in HSCs by 50% or greater by knocking down these proteins using specific siRNAs (Supplementary Fig. S4A–S4C). Using these cells, we performed in vitro transwell migration assays and compared their migration with cells that expressed LYN, SHIP1, and SHP1 at normal levels. The results of these experiments revealed that reduced expression of LYN or SHIP1 or SHP1 led to a significant decrease in the migration of HSCs compared with the controls (Fig. 4A). We extended these studies and performed a series of experiments where we exposed HSCs to increasing concentrations of inhibitors of LYN/Src kinase, SHIP1, and SHP1 and evaluated their migration response. These studies demonstrated a dose-dependent decrease in HSC migration (Fig. 4B–D). Taken together, increased migration of HSCs observed in cells expressing reduced levels of DOCK4 seemed to be as a result of each of the three signaling molecules identified in this study.

Next, we interrogated whether inhibition of LYN kinase, SHIP1, or SHP1 can reverse the increased migration observed in HSCs expressing reduced levels of DOCK4. We performed in vitro transwell migration assays using HSCs expressing normal and reduced levels of DOCK4 in the presence and absence of pharmacologic inhibitors of LYN/Src kinase, SHIP1, or SHP1. As expected, in the absence of LYN/Src inhibitor, DOCK4-deficient HSCs exhibited significant increase in transwell migration when compared with the controls (Fig. 4E). However, in the presence of LYN/Src inhibitor, the increased migration exhibited by the DOCK4-knocked down cells was significantly blunted and returned to migration levels exhibited by DOCK4 intact HSCs (Fig. 4E). In a similar manner, increased migration exhibited by DOCK4-deficient HSCs was also significantly reduced in the presence of pharmacologic inhibitors of SHIP1 and SHP1 (Fig. 4F and G). Taken together, these studies provide evidence that aberrant migration resulted by reduced DOCK4 levels can be restored to normal levels by pharmacologically targeting its downstream targets LYN or SHIP1 or SHP1.

Because anemia is central to morbidity and mortality of patients with MDS, we investigated whether one or more of the inhibitors of downstream effectors of DOCK4 are capable of improving erythroid differentiation with minimal toxicity to HSCs (Supplementary Fig. S5A). We setup hematopoietic colony assays using HSCs from patients with MDS in the presence and absence of inhibitors of LYN/SRC kinase (RK20449), SHIP1 (3AC) and SHP1 (TPI-1) under conditions to promote erythroid colony formations. The results of these experiments revealed that LYN/SRC kinase inhibitors suppressed formation of erythroid colonies, whereas the SHIP1 inhibitor, 3AC, showed no change in colony numbers in patient with -7/(del)7q MDS HSCs (data not shown). However, MDS patient samples that were exposed to the SHP1 inhibitor exhibited a 50% increase in erythroid colonies, as well as up to 5-fold increase in hemoglobin content without suppressing overall colony numbers (Fig. 5A–D). In agreement with these data morphology of the erythroid colonies was larger and brighter red in intensity compared with HSCs from patients with -7/(del)7q MDS that were not exposed to the inhibitor. In addition, enumeration of differential colonies showed a shift from myeloid to erythroid under the culture conditions that was used in these experiments. To test whether pharmacologic inhibition of SHP1 under conditions where DOCK4 expression was at 50% will result in improved erythroid differentiation, we performed methyl cellulose colony assays using HSCs that have been treated with DOCK4 siRNAs to reduce DOCK4 expression to haploinsufficient levels in the presence or absence of SHP1 inhibitor. These experiments revealed that exposure of cells expressing reduced levels of DOCK4 to SHP1 inhibitor significantly increased the erythroid colonies, whereas cells expressing normal levels of DOCK4 showed no increase in colony numbers after exposure to the same inhibitor in comparison with the vehicle-treated controls (Fig. 5E). In addition, SHP1 inhibitor increased the hemoglobin levels and size of the erythroid colonies in the experimental arm of the study compared with the control arm (Fig. 5F and G). SHP1 inhibition did not have any impact on differentiation of HSCs expressing normal levels of DOCK4 (Fig. 5E–G; Supplementary Fig. S5B–S5E).

Because increased migration of HSCs within the bone marrow can lead to increased HSPC mobilization, we examined peripheral blood samples from patients that are haplo-insufficient for DOCK4 [-7/(del)7q] expression. Flow cytometry analysis was performed to determine the percentages of CD34⁺ subpopulation within the CD45⁺ population using peripheral blood samples from healthy individuals, patients with non -7q MDS, and patients with -7/(del)7q MDS. We found that compared with non -7q MDS samples, percent CD34⁺ cells in -7/(del)7q MDS samples were approximately 8.8-fold higher (P = 0.05; Supplementary Fig. S6A and S6B).

In this study, we identify key signaling pathways regulated by the adaptor protein DOCK4 (Fig. 5E). As a classical adaptor protein, DOCK4 lacks catalytic activity but provide multiple docking sites for other signaling elements and regulate their catalytic activities or stabilities via protein–protein interaction (30, 31). In our previous work we demonstrated that in differentiating primary human erythroblasts DOCK4 activates one of its downstream targets, RAC1 GTPase, which in turn promotes formation of the actin skeletal network required for terminal differentiation (16). Now we show in HSCs that reduced levels of DOCK4 results in global increase in tyrosine phosphorylation both with and without exposure to hematopoietic cytokines. Among the increased phosphorylated proteins LYN kinase, SHP1, and SHIP1 were prominent. Although activation of phosphatases, SHP1 and SHIP1 leads to dephosphorylation of their targets, overall reduced expression of DOCK4 resulted in phosphorylation of a large number of proteins due to activation of kinases. As result we observed a net gain in global phosphorylation when DOCK4 levels were low. Overall, the mechanism of action of DOCK4 is to act as a negative regulator of protein phosphorylation. LYN kinase, SHP1 and SHIP1 identified in this study are examples of downstream targets of DOCK4.

Increased cell migration and morphologic changes we observed when DOCK4 levels were low are consistent with functions that have been ascribed to LYN kinase based on previously published work (32–36). Previous studies have also shown that increased HSC migration was consistent with increase in HSC mobilization (37, 38), which we also observed in -7/(del)7q MDS patient blood samples. Our current work seemed to suggest that increased HSC mobilization is specifically associated with -7/(del)7q MDS because MDS samples with other chromosomal abnormalities did not exhibit increased HSC mobilization. Furthermore, our current findings showing increased phosphorylation/activation of SHIP1 and SHP1 is also consistent with previous findings showing both these phosphatases are substrates for LYN kinase (39, 40). In fact, Lyn-deficient mice exhibit similar phenotypic characteristics to mice lacking Ship1 and Shp1 (29). On the basis of these data we show DOCK4, LYN kinase, SHIP1, and SHP1 are all part of the same signaling cascade and because deficiency of DOCK4 in HSCs impacts all three enzymes one could target this pathway to reverse functional deficiency observed in these HSCs.

In fact when we evaluated for terminal differentiation of MDS patient–derived CD34⁺ cells under culture conditions that was permissive for erythroid differentiation, cells that were exposed to the SHP1 inhibitor (TPI-1) showed prodifferentiation characteristics (increased BFU-Es and hemoglobinization). These results were in agreement with previous studies using healthy cells, which had demonstrated that SHP1 phosphatase is a negative regulator of erythroid differentiation (41–43). Therefore, by blocking SHP1 activity one can potentially reverse anemia in patients with -7/(del)7q MDS. Recent work by Kundu and colleagues (44) have demonstrated that TPI-1 and its analogues are nontoxic when administered to mice and is effective in reducing the melanoma tumor burden in mice (44). In another study it was shown that mice lacking Shp1 do not respond to TGF-β (45). Because previous work by us and others have shown that in MDS TGF-β signaling is overactive and inhibiting this pathway can restore hematopoiesis in MDS (46–48), our current work identifying SHP1 inhibition leading to terminal erythroid differentiation provides a unique opportunity to develop inhibitors of SHP1 that might be effective in treating patients with MDS.

Kinase inhibitors such as midostaurin are effective in patients with high-risk Flt3-mutated MDS/acute myelogenous leukemia, whereas proerythroid differentiation agents such as luspatercept and erythropoietin are effective in patients with low-risk MDS. Our findings in this study highlight the potential use of SHP1 inhibitor as a proerythroid differentiation agent in patients with intermediate/high risk MDS with chromosome 7 deletions. Future investigation of SHP1 inhibitor as single agent and in combination with luspatercept or 5-azacytidine/decitabine as a treatment strategy in -7/del(7q) MDS is warranted. Taken together, this study has uncovered a signaling network regulated by DOCK4 that can be targeted to reverse the aberrant phenotypes arising because of reduced expression of DOCK4.

A. Verma holds ownership interest (including patents) in and is a consultant/advisory board member for Stelexis. No potential conflicts of interest were disclosed by the other authors.

Conception and design: S. Sundaravel, A. Verma, A. Wickrema

Development of methodology: S. Sundaravel, W.-L. Kuo, T.D. Bhagat

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S. Sundaravel, W.-L. Kuo, J.J. Jeong, G.S. Choudhary, S. Gordon-Mitchell, H. Liu, T.D. Bhagat, K.L. McGraw, S. Gurbuxani, A.F. List, A. Verma

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S. Sundaravel, W.-L. Kuo, J.J. Jeong, G.S. Choudhary, T.D. Bhagat, A. Verma, A. Wickrema

Writing, review, and/or revision of the manuscript: S. Sundaravel, S. Gurbuxani, A. Verma, A. Wickrema

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): T.D. Bhagat

Study supervision: A. Wickrema

This work was supported, in part, by NIH R01 HL16336 (to A. Verma and A. Wickrema), Leukemia and Lymphoma Society translational research program (to A. Verma and A. Wickrema) and NCI F99/K00 CA223044 predoctoral to postdoctoral fellow transition award (to S. Sundaravel). We thank Dr. Linda Van Aelst (Cold Spring Harbor Laboratory) for generously providing the flag-tagged DOCK4 construct and Dr. Aaron Marshall (University of Manitoba, Winnipeg, Canada) for generously providing the GFP-tagged SHIP1 construct. Proteomics services were performed by the Northwestern Proteomics Core Facility, generously supported by NCI CCSG P30 CA060553 awarded to the Robert H. Lurie Comprehensive Cancer Center and the National Resource for Translational and Developmental Proteomics supported by P41 GM108569.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.