Dok1 and Dok2 Proteins Regulate Cell Cycle in Hematopoietic Stem and Progenitor Cells

Myeloid cells are sensitive to cytokines and growth factors that activate intracellular protein tyrosine kinases (PTK). Constitutive PTK activation results in hematopoietic cell transformation as highlighted in chronic myeloid leukemia with the BCR-ABL fusion gene product showing PTK activity. Among the main substrates identified downstream BCR-ABL, the adaptor proteins, downstream of kinase Dok1 and Dok2 (1, 2), are RasGAP-binding proteins acting as negative regulators of signaling pathways in lymphoid (3, 4) and myeloid cells (1, 2, 5, 6). Dok1 and Dok2 proteins have been shown to inhibit growth factors and cytokine-induced cell proliferation in hematopoiesis (6–11). Mice deficient for both Dok1 and Dok2 double knockout (Dok DKO) develop a myeloproliferative disease similar to human chronic myelomonocytic leukemia (CMML) with a latency of 1 y. We recently identified DOK2 loss-of-function point mutations in CMML patients (12). Dok DKO mice display medullary and extramedullary hyperplasia and increased activation of RAS and PI3K-dependent pathways in cytokine-activated granulomonocytic progenitors (6, 7).

In healthy adult mammals, early hematopoietic precursors are mainly located in the bone marrow (BM). They are part of the Lin⁻Sca-1⁺Kit⁺ (LSK) compartment that can be further divided into hematopoietic stem cells (HSC) and multipotent progenitors (MPP) 1–4 using additional markers such as CD34, CD150, CD48, and CD135 (13–16). HSC represent a very small cell population with a clonal capacity to provide lifelong regeneration of all differentiated hematopoietic cells. To maintain hematopoietic homeostasis, HSC differentiation, self-renewal, and maintenance must be tightly regulated (17–20). Most adult HSC are quiescent, contributing to HSC maintenance in the BM and ensuring lifelong hematopoietic replenishment (19, 21). Furthermore, slow cycle progression is essential to maintain all lifelong pools of self-renewing HSC (22). To avoid exhaustion, HSC must be protected from stress. Indeed, quiescent HSC populations are resistant to 5-fluorouracil (5-FU)–induced myelosuppression (23, 24), suggesting that HSC quiescence is closely linked to the protection of the hematopoietic system from various stresses. Several mechanisms regulate HSC cell cycle and maintain cell quiescence (24–27). Tie2/Angiopoietin-1, Mpl/thrombopoietin (TPO), and Wnt/β-catenin signaling pathways; cell adhesion molecules; and p21 and p57 Cyclin-dependent kinase (CDK) inhibitors are involved in the maintenance of HSC and protect them from cellular stress (25–31). Because Dok proteins are critical negative regulators of proliferation and consequently of cell cycle downstream of growth factor signaling, we wondered whether they could play a role in HSC quiescence.

To test this hypothesis, we analyzed the effects of double Dok1 and Dok2 gene deletion on hematopoietic precursors in young mice. We show imbalanced frequencies and numbers of mature hematopoietic cells and precursors at steady state with an increased proportion of cells in active phases of cell cycle. Interestingly, upon 5-FU–induced myeloablation, Dok DKO HSC return to a quiescent state faster than wild-type (WT) HSC. Serial 5-FU injections of these mice demonstrated that Dok DKO HSC show enhanced resistance to exhaustion compared with WT mice. Taken together, our results reveal a new unexpected function of Dok proteins as switch controllers of cell cycle entry, promoting quiescence at steady state and increasing susceptibility to HSC exhaustion upon myeloablative stress.

Dok-1^−/−Dok-2^−/− DKO mice (Dok DKO) were obtained by interbreeding Dok-1^−/− mice with Dok-2^−/− mice both in 129/Sv background as described previously (2, 7). We generated CD45.1⁺ and CD45.2⁺ mice for chimera and transplantation experiments by crossing a 129/Sv (CD45.2⁺) male and CD45.1⁺ C57BL/6 female to obtain F₁(129/Sv × C57BL/6) CD45.1⁺/CD45.2⁺ recipient mice. WT 129/Sv and CD45.1⁺ C57BL/6 mouse strains were purchased from Charles River Laboratories (L’Arbresle, France). Six- to eight-week-old females were used throughout the study. All experiments were performed in agreement with the French guidelines for animal handling and were approved by the Inserm ethical committee.

Freshly dissected femurs, tibias, and hips were isolated from mice. BM was flushed with a syringe into RPMI 1640 medium supplemented with 10% FBS. The BM was spun at 1600 rpm by centrifugation at 4°C, and RBCs were lysed in ACK Lysing Buffer (Life Technologies) for 3 min. After centrifugation, cells were suspended in PBS plus 2% FBS, passed through a cell strainer (BD Biosciences), and counted. For serial transplantation, 5 × 10⁶ total BM cells of WT or Dok DKO CD45.2⁺ were transplanted via retro-orbital injection into lethally irradiated (9.5Gy) F₁(129/Sv × C57BL/6) CD45.1⁺/CD45.2⁺ mice. This model was chosen to keep an allelic reporter system and to overcome graft rejection because Dok DKO cells were on 129/Sv background (H-2^bc) different from the MHC background of C57BL/6 mice (H-2^b) (for details see (http://www.imgt.org/IMGTrepertoireMHC/Polymorphism/haplotypes/mouse/MHC/Mu_haplotypes.html#polymorphism). Mouse chimerism in the peripheral blood was monitored by flow cytometry every 4 wk (weeks 4, 8, and 12). Chimerism and hematopoietic phenotype in the BM and spleen were evaluated at 12 wk after transplantation.

The following Abs used for flow cytometry and cell sorting were purchased from eBioscience or BioLegend: biotinylated anti-CD4 (clone RM4-5), CD8 (clone 53-6.7), CD3 (clone 2C11), DX5 (clone DX5), CD11c (clone N418), CD19 (clone 6D5), B220 (clone RA3-6B2), TER119 (clone TER119), CD11b (clone M1/70) and Gr1 (clone RB6-8C5) (Lineage mixture), cKit-allophycocyanin-eFluor780 (clone 2B8), Sca-1-PerCP-Cy5.5 (clone D7), CD34-FITC (clone RAM34), CD16/32-PECy7 (clone 93), CD150-allophycocyanin (clone TC15- 12F12.2), CD48-PECy7 (clone HM48.1), CD48-BV421 (clone HM48.1), CD135-PE (clone A2F10), CD45-allophycocyanin-eFluor780 (clone 30-F11), CD45.1- PE CF594 or -FITC (clone A20), CD45.2-AF700 or -eFluor 450 (clone 104), CD19-AF700, CD19-allophycocyanin-Cy7, CD11b-FITC, CD3-PECy7, CD115-allophycocyanin (clone AFS98), F4/80-PE (clone BM8), Ly6C-BV421 (clone AL-2), Ly6G-BV711 (clone 1A8), Ly6G-BV421, CD71-PE (clone PE C2), and Ter119-BV421 and CD105-PE (clone MJ7/18). Brilliant Violet 510-Streptavidin (BioLegend) and AF594-Streptavidin (BD Biosciences) were used. For CFU assays, granulocyte monocyte progenitor (GMP) cells were obtained after total BM cells sorting according to their surface markers expression (Lin⁻c-Kit⁺Sca-1⁻CD34⁺ and CD16/32⁺). Flow cytometry was performed on a BD LSR II flow cytometer, and cell sorting was executed on a BD FACSAria III or BD FACSAria II SORP. Flow Cytometry data were analyzed with BD FACSDiva version 6.1.2 software (BD Biosciences).

BM cells were incubated in PBS plus 2% FBS and 2.5 mM EDTA with biotinylated Abs against lineage markers, followed by anti-biotin microbeads (Miltenyi Biotec) and then the Lineage⁺ cells were depleted by autoMACS Pro Separator (Miltenyi Biotec).

BM progenitors were enriched as previously described and stained with c-Kit-allophycocyanin-eFluor750, CD150-allophycocyanin, Sca-1-PerCP-Cy5.5, CD48-PECy7, and CD16/32-PECy7. The biotin-conjugated lineage combination was revealed with Alexa Fluor 594-streptavidin (BD Biosciences). After extracellular staining, cells were permeabilized, fixed, and stained with anti–Ki67-FITC Ab (BD Biosciences) (32) and DAPI (Sigma-Aldrich) as described previously (33).

Ten cells for each population (HSC, MPP1, MPP2, and MPP3/4) from WT or Dok DKO mice were sorted using the autoclone module on an Aria III SORP sorter (BD Biosciences) into 96-well plates in the RT-STA Master Mix Solution. Cell lysis, cDNA synthesis, and amplification were performed according to Fluidigm protocols. Selected TaqMan Gene Expression assays (original magnification ×20) were pooled and diluted with water 100-fold so that each assay is at a final concentration of 0.2× in the pooled assay mix. Five microliters of CellsDirect 2× Reaction Mix (Invitrogen), 2.5 μl pooled assay mix, 0.2 μl SuperScript III RT/Platinum Taqmax mix, and 2.3 μl water were combined to a final volume of 10 μl in 1 well of a 96-well qPCR plate for the sort. cDNA samples were amplified using the following program (1): 50°C, 10 min (2); 95°C, 2 min (3); 95°C, 15 s (4); 60°C, 4 min; repeat steps (3) and (4) 18 times. Preamplified products were diluted 5-fold before analysis with Universal PCR Master Mix and inventoried TaqMan gene expression assays in 96.96 Dynamic Arrays on a BioMark System (Fluidigm). Nomenclature of the used TaqMan probes and targets is reported in Supplemental Table I. For each gene, the relative expression values were calculated as 40 – cycle threshold.

In survival experiments, Dok DKO mice and control mice were injected i.p. once per week with 5-FU (120 mg/kg). Survival was monitored on a daily basis up to 25 d after the first 5-FU injection. In others experiments, animals received just two 5-FU injections (days 0 and 21, when peripheral blood leukocyte count returns to a similar level than at day 0).

Fresh GMP or BM cells treated with ACK Lysing Buffer (Life technologies) were counted and diluted to a concentration of 5 × 10⁵ cells/ml in IMDM supplemented to 10% FBS (Eurobio), 100 U/ml penicillin, and 100 μg/ml streptomycin (Life Technologies). For comparison of GMP differentiation potential between Dok DKO and WT (Fig. 3), 200 sorted cells were added to 2 ml Methocult GF M3434 (Stemcell Technologies) and seeded at 1 ml/dish. For total BM CFU-GM assays, 2 × 10⁴ cells and Methocult GF M3534 were used. After 7–10 d, the dishes were scored for hematopoietic colonies, and results were expressed as number of colonies per well.

Bones were decalcified during 15 d in PBS plus 270 mM EDTA solution at 37°C and then fixed in 1% paraformaldehyde and prepared to paraffin inclusion using automated tissue processor ASP 300 (Leica). Dehydration, clarification, and infiltration steps were performed by successive absolute ethanol, histolemon, and paraffin baths. Then bones were disposed in paraffin. Sections of 5-μm thickness were performed with HM 340E microtome (Thermo Scientific). Hematoxylin eosin safran staining was performed using automated slide stainer JUNG XL (Leica). Finally, slides were dehydrated by absolute ethanol and histolemon baths and mounted in Pertex medium using glass coverslipper CV5030 (Leica). Slides were scanned with a digital slide scanner NanoZoomer 2.0-HT (Hamamatsu).

Results are expressed as mean values ± SEM. Statistical significance of differences between the results was assessed using a two-tailed unpaired Student t test, performed using GraphPad Prism version 5.03 software (GraphPad). The p values < 0.05 were considered statistically significant.

Single Dok1- or Dok2-deficient mice did not show obvious phenotypes possibly because of overlapping functions of Dok1 and its closest family member, Dok2 (34). In contrast, double deficiency for Dok1 and Dok2 results in myeloproliferative disorder resembling CMML after a latency of 1 yr (6, 7). Thus, we analyzed hematopoiesis in 6- to 8-wk-old double Dok1 and Dok2 deficient (Dok DKO) mice to better understand the molecular mechanisms controlled by Dok1 and Dok2 in hematopoietic stem and progenitor cells (HSPC). Although absolute numbers of WBC were similar in WT and Dok DKO mice in the peripheral blood, a significant increase of monocytes (CD11b⁺Gr1^loF4/80⁺) and neutrophils (CD11b⁺Gr1^hi) was observed in Dok DKO mice as compared with WT (Fig. 1A). The cellularity of BM and spleen (data not shown) was significantly higher in Dok DKO than WT mice with significant increase in neutrophils and monocytes, respectively, CD11b⁺Ly6G^High/Ly6C^Low and Ly6G^Neg/Ly6C^High in the BM (Fig. 1B) and peripheral blood (data not shown). Thus, Dok DKO mice already display features of hematopoietic disorders at a young age, suggesting a constitutive imbalanced commitment of hematopoietic progenitors. This was confirmed by the analysis of early hematopoietic progenitors as defined by the gating strategy described in Fig. 1C. A significant increase of cell numbers in Dok DKO compared with WT mice was detected in the Lin⁻Kit⁺Sca-1⁻ (LK) and LSK compartments. In different LSK subsets, the absolute number of HSC (CD150⁺,CD34⁻,CD48⁻,CD135⁻) and MPP1 (CD150⁺,CD34⁺,CD48⁻,CD135⁻) remained unchanged, but their frequency was decreased (Fig. 1D, 1E). However, a significant increase in the abundance and frequency of the more differentiated subsets (MPP2 [CD150⁺,CD48⁺,CD135⁻], MPP3 [CD150⁻,CD48⁺,CD135⁻], and MPP4 [CD150⁻,CD48⁺,CD135⁺]) was observed in Dok DKO compared with control mice. In the LK cell compartment, the frequency of GMP was significantly increased at the expense of common myeloid progenitors (CMP), whereas differences in absolute numbers were only significant for GMPs, with an increase because of DKO BM hypercellularity (Fig. 1F). However, the ratio of CD105⁻CD150⁻ and CD105⁻CD150⁺ CMP subtypes (35) remained unchanged (Supplemental Fig. 2A, 2B). Finally, an increase in basophilic and orthochromatophilic erythroblasts (36) was observed in Dok DKO mouse BM (Supplemental Fig. 2C, 2D). Altogether, these data show that Dok proteins are involved in hematopoiesis from early stages of hematopoietic differentiation with downstream consequences on the balanced proportions of CMP, GMP, and erythroid progenitors.

Because Dok1 and Dok2 expression is not restricted to hematopoietic cells, we tested whether hematopoietic defects are cell autonomous. Lethally irradiated recipient mice (F₁(129/Sv × C57BL/6) CD45.1⁺/CD45.2⁺) transplanted with WT or Dok DKO total BM (129/Sv CD45.2⁺) were sacrificed 12 wk after transplantation. The percentage of chimerism was monitored in the peripheral blood of recipient mice, and similar reconstitution potential was observed between WT and Dok DKO BM donors (Fig. 2A). Hematopoietic donor cells (CD45.2⁺) chimerism represents up to 95% of the peripheral blood, spleen, and BM, indicating a strong hematopoiesis reconstitution (Fig. 2A, Supplemental Fig. 2B). Mice transplanted with Dok DKO BM showed the same myeloproliferative features than Dok DKO mice: monocytosis, splenomegaly, BM hypercellularity, increased level of LK (Lin⁻Kit⁻Sca-1⁻ cells), and LSK compartment beginning from MPP2 differentiation step (Fig. 2B, Supplemental Fig. 2C, 2D). These results demonstrate that the hematologic disorders observed in Dok DKO mice are transplantable and cell autonomous. In addition, serial transplantations showed an amplification of the observed phenotype, suggesting that Dok DKO hematopoietic cells present an early commitment defect evolving with time (Fig. 2D, Supplemental Fig. 2C, 2D). Frequencies of LK and LSK cells were similar in WT and mutant mice at steady state but significantly increased after serial transplantations (Fig. 2E). In secondary transplants, CMP are increased compared with the first transplant or steady state (Fig. 2D). Although first and secondary transplanted mice have similar frequency of donor cells in the peripheral blood, tertiary Dok DKO BM–transplanted mice present increased CD45.2⁺ donor cells frequency compared with WT BM transplanted mice (respectively 87.95 ± 1.05 versus 82.71 ± 1.65% of CD45.2⁺ cells) (Fig. 2A, 2C). This suggests that Dok DKO HSC have higher engraftment capacity. Indeed, whereas Dok DKO HSC and MPP1 are less abundant in the LSK cell compartment than the WT cells at steady state, after three rounds of transplantation their proportion in the LSK cell compartment is higher (Fig. 2F, 2G), suggesting that Dok1 and Dok2 play a role as modifiers of HSPC expansion.

To address the question whether the increased numbers of myeloid cells in Dok DKO mice (Fig. 1F) was due to increased proliferation or differentiation of GMP into monocytes/macrophages, CFU assays were performed using equal numbers of GMPs isolated from Dok DKO or control mice. Increased numbers of colonies with enlarged appearance were obtained using GMPs isolated from Dok DKO as compared with control (Fig. 3A, 3B). This was likely because GMP isolated from Dok DKO mice were less quiescent as compared with control (data not shown). Characterization of CFU-GM colonies showed increased percentage of Dok DKO cells expressing myeloid differentiation markers such as CD11b, CD115, or CD16/32 (Fig. 3C, 3D) with higher frequencies of cells with a more differentiated aspect (Fig. 3E). These results indicate that GMPs isolated from Dok DKO have an increased proliferation rate and differentiation potential into more mature myeloid cells.

Because the size of the HSC pool is tightly controlled by the balance between dormant and cycling cells, we further investigated the cell cycle status of the LK and LSK cell compartments in Dok DKO mice. We combined cell surface staining to define hematopoietic subsets with intracellular staining with DAPI (DNA staining) and anti-Ki67 Ab (a cell proliferation associated marker). We found less quiescent LSK and LK cells in Dok DKO mice than in WT control mice (Fig. 4A, 4B). In the LK cell compartment, a decrease in the proportion of G₀-phase cells with an accumulation in G₁ phase was detectable for CMP and GMP but not for megakariocytic erythroid progenitor subsets (data not shown). In addition, analysis of LSK subsets revealed that the cell cycle was affected in MPP1, MPP2, and MPP3/4 subsets but not HSC at the steady state (Fig. 4C), suggesting that Dok1 and Dok2 control cell cycle entry at the transition between HSC and MPP1. To get further insights into the mechanism by which loss of Dok1/Dok2 expression promotes cell cycle in MPPs, we compared gene expression of cell cycle regulators such as Rb1 and Rbl1 (negative regulators of cell cycle), cyclins (Ccns), cyclin-dependent kinases (Cdks), cyclin-dependent kinase inhibitors (Cdkns), growth factor receptor (Mpl, Csf2ra, and Kit), and transcription factors (Gata3 and Myb) between cells isolated from Dok DKO or control mice. Surprisingly, we found a number of downregulated genes in the HSC compartment although we failed to detect Dok1 and Dok2 expression in HSC (Fig. 4D). This is likely because Dok1 and Dok2 are only expressed by a fraction of HSC and are thus not detected under our experimental conditions (bulk of 10 sorted cells). In addition, we observed robust upregulation of Cdkn1a and strong downregulation of Csf2ra expression in MPPs isolated from Dok DKO mice as compared to control. This indicates that a dynamic complex interplay between Dok signaling and cell cycle regulation may occur in more mature progenitors, although the results pinpoint to a prominent function of Dok1 and Dok2 in controlling HSC proliferation at the transition to MPPs.

To alter steady-state hematopoietic homeostasis and to induce peripheral blood replenishment from proliferating progenitors, we treated mice with 5-FU on days 0 and 21 and monitored leukocyte counts in peripheral blood over 42 d. We found that WBC counts were higher in the recovery period in Dok DKO mice as compared to WT mice (9.14 ± 0.80.10³ versus 6.03 ± 0.47.10³/mm³, respectively, at day 14; Fig. 5A). This difference was enhanced during the second round of 5-FU injection. 5-FU stress-induced de novo hematopoiesis significantly increased the myelomonocytic phenotype because Dok DKO mice displayed obvious neutrophilia and monocytosis at day 14 (Fig. 5B). A selective increase of Ki67-positive myeloid cells was observed in Dok DKO leukocytes during the leukocyte proliferative phase (from days 3 to 9) (Fig. 5C), suggesting that loss of Dok protein expression promotes cell cycle entry and proliferation in mature myeloid cells.

To confirm that our previous observations were due to deregulations in cell cycle checkpoints in more immature progenitors, 6- to 8-wk-old mice were treated with a single injection of 5-FU, and the BM was analyzed after 3 or 5 d for cell cycle status and cellularity. We found that BM cellularity in Dok DKO mice was increased as compared with WT mice with a dramatic increase of LK and LSK fractions after 3 d (Fig. 6A). Five days after 5-FU injection, the difference was even more pronounced with BM cellularity in Dok DKO mice being twice as much as compared to WT mice, with increased progenitors with a granulomonocytic fate (Fig. 6B, left and middle panels). This elevated BM cellularity in Dok DKO mice compared with WT littermates has been confirmed by performing immunohistochemistry experiments (Supplemental Fig. 3A). In the LSK cell compartment, the number of MPP1 to MPP4 cells was higher in Dok DKO mice as compared to WT littermates, whereas HSC numbers were not affected (Fig. 6B, right panel). Because HSC proliferation is tightly associated with their differentiation, one possibility is that Dok1 and Dok2 proteins control HSC cell cycle entry, which is revealed by increased numbers of more differentiated progenitors. This was confirmed by analysis of the cell cycle status of HSC and MPP1 at days 3 and 5 after 5-FU injection (Supplemental Fig. 3B). Indeed, in Dok DKO samples, fewer HSC were quiescent (G₀ cell cycle phase) than in WT samples at day 3 (Fig. 6C), but they returned to quiescence at day 5 in greater proportion (Fig. 6D), suggesting that Dok1 and Dok2 regulate cell cycle checkpoints. This directly affects progenitor cell numbers and differentiation because cell cycle entry is known to occur at the transition between HSC and MPP1. This was confirmed by paired analysis of HSC and MPP1 cell cycle status 5 d after 5-FU treatment. Indeed, more Dok DKO than WT HSC exit quiescence (G₀) and enter cell cycle (S-G₂-M) during HSC/MPP1 transition (Fig. 6E). This results in a reduced proportion of DKO HSC in G₁, suggesting that these cells had already proceeded to MPP1 differentiation.

Increased hematopoiesis of Dok DKO mice after myeloablation should result in an enhanced sensitivity to sustained aplasia, but the rapid re-entry of mutant HSC in a quiescent state compared with WT cells may have a protective role on the HSC pool. To address this issue, mice were subjected to weekly 5-FU injections and their survival was monitored over time. WT mice survival was significantly reduced compared with Dok DKO littermates (Fig. 7A). More than 80% of WT mice died compared with only 6% of Dok DKO mice. Peripheral blood monitoring during the experiment showed that there were more WBC (lymphocytes, monocytes, and reticulocytes) in Dok DKO mice as compared with WT mice (Fig. 7B). This was associated to a marked reduction of anemia, which may account for the increased resistance of Dok DKO mice to 5-FU–induced exhaustion. This would be in agreement with increased proportion of erythroid progenitors observed in Dok DKO mice at steady state (Supplemental Fig. 1C, 1D). Altogether, our results reveal an unexpected function for Dok proteins in the control of cell cycle regulation at the transition checkpoint between HSC and MPP1 resulting in an overall higher hematopoietic regenerative capacity of Dok DKO mice in response to myeloablative stress.

Previous reports have shown that Dok1 and Dok2 are involved in myeloid homeostasis and leukemia suppression due to their negative impact on signaling pathways involved in hematopoietic growth and differentiation (8, 11). To investigate the role of these proteins, the double Dok1- and Dok2-deficient mice are a useful animal model because they are overlapping functions of DOK1 with its closest family member, DOK2 (34). DOK1 and DOK2 are docking proteins and are highly related in structure (34, 37).

Loss of Dok1 and Dok2 leads to increased BM cell proliferation (6, 7). Although this suggests a role for the Dok proteins in the maintenance of hematopoietic homeostasis, this issue has never been addressed before. In this study, we demonstrate that Dok1 and Dok2 proteins regulate cell cycle at the transition between HSC and MPP1. After a single 5-FU challenge, a dramatic decrease of quiescent G₀-phase HSC is detected in WT mice between days 3 and 5 (Fig. 6). This drop in the number of quiescent HSC is not detected in Dok DKO mice, indicating that Dok1 and Dok2 are negative regulators of proliferative signals during BM regeneration and may act as brakes counterbalancing microenvironmental cues maintaining HSC quiescence.

At steady state, Dok1 and Dok2 control progenitor cellularity by negative regulation of cell cycle. Indeed, Dok DKO mice showed increased proportions of cells in G₁ and/or S-G₂-M phases of cell cycle in LSK and LK cells (Fig. 4). In human erythroid progenitors, Dok1 phosphorylation has been correlated with a downregulation of CDK inhibitors (8), CDKN1A (p21^cip1) and CDKN1B (p27^kip1), suggesting that Dok1 is involved in negative regulation of molecules involved in cell cycle arrest. Common signaling pathways such as RAS/ERK pathways are involved in cell cycle regulation (38, 39). DOK1 and DOK2 proteins block these signals by recruiting enzymes, RasGAP and SHIP-1, respectively, involved in negative regulation of RAS/ERK and PI3K/AKT signals (34). They have been reported to block RAS/ERK and PI3K/AKT pathways in murine myeloid cells (6, 7, 40). In addition to these posttranslational modifications, here we show that loss of Dok1/Dok2 expression results in upregulation of Cdkn1a (p21^cip1) gene expression (Fig. 4), reported to be involved both in negative or positive role in HSPC pool size (26, 41). Moreover, Dok1/Dok2 deficiency induces a downregulation of Cdkn1c (p57^kip2) gene expression in HSC, known as a master regulator of HSC quiescence (27, 42). Furthermore, Dok1/Dok2 double deletion is also associated to Mpl gene downregulation. TPO/MPL signaling is required for the regulation of HSC quiescence and is associated with upregulation of genes encoding for CDK inhibitors in HSC (43). Altogether, these data suggest a broader effect of Dok1 and Dok2 in HSC cell cycle regulation.

Serial transplantation experiments show a progressive decrease of WT HSC compared with Dok DKO HSC (Fig. 2), suggesting that serial transplantations generate a lower rate of HSC exhaustion in Dok DKO context. Dok1/Dok2 deficiency is associated with Gata3 gene downregulation at HSC stage (Fig. 4D). Similar to Dok DKO HSC (Fig. 2F), an expansion of Gata3-deficient HSC is reported after serial transplantations showing that Gata3 is associated with increased HSC self-renewal (44). These results are supported by the differential sensitivity of Dok DKO mice as compared with control after serial 5-FU injections (Fig. 7). Indeed, 5-FU myeloablation induces more all LSK subsets (including HSC) in G₁ and S-G₂-M phases of cell cycle (Supplemental Fig. 3B). Thus, in agreement with the finding that genes involved in cell cycle regulation are mostly affected in HSC compartment of Dok DKO mice, it seems that Dok1 and Dok2 negatively regulate cell cycle at the transition between HSC and MPPs. Moreover, the cell cycle status of HSC and MPP1 progenitors in G₀ and S-G₂-M state suggests that Dok DKO HSC proceeds faster to MPP1 differentiation (Fig. 6E), suggesting that cell proliferation and differentiation may represent two sides of the same coin. The percentage of G₁-phase HSC and MPP1 cells is higher in Dok DKO than in WT mice. Because transition from HSC to MPP1 is increased in Dok DKO mice compared with WT cells, these results suggest that a process is engaged to bring back Dok DKO HSC to a quiescent state. These results reinforce the idea of a faster reversion of cycling HSC into quiescent state in response to 5-FU treatment in the absence of Dok1 and Dok2. Dok double deficiency confers protection to the HSC compartment against hematopoietic stress, preventing exhaustion. A similar phenomenon has recently been described in response to chronic exposure to type I IFNs (IFN-I) (45). Indeed, IFN-I exposure induced transient HSC proliferation, followed by return to a quiescent state, and did not exhaust the HSC pool (46). In this study, our results suggest that, as is the case upon IFN-I exposure, Dok1 and Dok2 may be involved in the control of HSC quiescence and in the balance between proliferation and differentiation.

Finally, the study of Dok DKO steady state hematopoiesis allowed us to describe early signs of myeloproliferative disorder. These alterations are already present in healthy young mice and affect the LSK cell compartment. Dok DKO HSCs exhibit a decreased expression of Gata3 gene (Fig. 4D), known to be involved in lymphoid lineage specification. Recently, Gata3 downregulation was associated with myeloid biased HSC (44), correlating with biased myeloproliferation detected in Dok DKO mice. Hematopoietic organs (spleen, BM and peripheral blood) of Dok DKO mice display monocytosis and neutrophilia. Loss of Dok genes also promotes cell autonomous expansion of MPP3, MPP4 and GMP subsets (Fig. 1). Therefore, this study reveals that a preleukemic state already exists at steady state in the Dok1 and Dok2 double-deficient mice. Moreover, serial transplantations and 5-FU induced myeloablation experiments showed enhancement of CMML-like features showed at steady state (Fig. 2). 5-FU–induced myeloablation enhances Dok DKO pre-CMML phenotype by exacerbating the number of LSK cells (including HSC) in G₁ and S-G₂-M phases of cell cycle (Supplemental Fig. 3B).

Thus, Dok1 and Dok2 negatively regulate the progression of hematopoietic progenitors from HSC to committed progenitors but also control reversion to HSC quiescence. This is in agreement with a previous report showing that cell transformation is associated to cell cycle increase and upregulation of Dok1 phosphorylation described in a chronic myeloid leukemia leukemic stem cell line (47). Altogether, these data suggest a dual role of Dok1 and Dok2 proteins in hematopoietic progenitor cell cycle regulation. At steady state, they negatively regulate cell cycle and numbers of progenitors. Paradoxically, upon hematopoietic stress, Dok proteins oppose the quiescence of HSC (8, 9, 48, 49). Dok1 and Dok2 dysregulations have been associated with several oncogenic processes (1, 6, 7, 12, 50–56). Therefore, the impact of chemotherapeutic agents on HSCs may have to be considered to avoid resistance of leukemic stem cells and subsequent relapse.

We thank V. Ferrier-Depraetere (Institut Paoli-Calmettes) for thoughtful reading of the manuscript. We thank M.-L. Thibult and F. Mallet for assistance with the use of the cytometry and cell sorting facility; P. Gibier, J.-C. Orsoni, and F. Gianardi for animal facilities; E. Agavnian for work on the Institut Paoli-Calmettes/Centre de Recherche en Cancérologie de Marseille experimental pathology platform; C. Fauriat, C. Acquaviva, S. Sarrazin, F. Brunet, D. Birnbaum, E. Duprez, and P. Dubreuil for helpful discussions; and F. Bardin, S. Morin, and A. Goubard for helping to conduct the mouse experiments.

The authors have no financial conflicts of interest.