Dynamins 2 and 3 control the migration of human megakaryocytes by regulating CXCR4 surface expression and ITGB1 activity

Abstract

Megakaryocyte (MK) migration from the bone marrow periosteal niche toward the vascular niche is a prerequisite for proplatelet extension and release into the circulation. The mechanism for this highly coordinated process is poorly understood. Here we show that dynasore (DNSR), a small-molecule inhibitor of dynamins (DNMs), or short hairpin RNA knockdown of DNM2 and DNM3 impairs directional migration in a human MK cell line or MKs derived from cultured CD34⁺ cells. Because cell migration requires actin cytoskeletal rearrangements, we measured actin polymerization and the activity of cytoskeleton regulator RhoA and found them to be decreased after inhibition of DNM2 and DNM3. Because SDF-1α is important for hematopoiesis, we studied the expression of its receptor CXCR4 in DNSR-treated cells. CXCR4 expression on the cell surface was increased, at least partially because of slower endocytosis and internalization after SDF-1α treatment. Combined inhibition of DNM2 and DNM3 or forced expression of dominant-negative Dnm2-K44A or GTPase-defective DNM3 diminished β1 integrin (ITGB1) activity. DNSR-treated MKs showed an abnormally clustered staining pattern of Rab11, a marker of recycling endosomes. This suggests decreased recruitment of the recycling pathway in DNSR-treated cells. Altogether, we show that the GTPase activity of DNMs, which governs endocytosis and regulates cell receptor trafficking, exerts control on MK migration toward SDF-1α gradients, such as those originating from the vascular niche. DNMs play a critical role in MKs by triggering membrane-cytoskeleton rearrangements downstream of CXCR4 and integrins.

Graphical abstract

Figure 1.

DNM2 and DNM3 are expressed in human MKs and modify plasma membrane anisotropy. (A) OMX super-resolution microscopy image of DNM2 (red) and DNM3 (green) staining in human MKs; original magnification, 63× objective. Twenty-five or more primary MKs were analyzed by confocal immunofluorescence. Here shown are OMX super-resolution microscopy images for illustration purposes. DNM2 specks are distinct from the DNM3 dotted staining, and both are distributed throughout the cytoplasm. Blue stains for the nucleus. Scale bar represents 2 µm. (B) Western blot (WB) of DNM2 and DNM3 and efficiency of their knockdown in CHRF cells. DNM2 presented as a single band in the megakaryocytic CHRF cells, whereas DNM3 migrated as 2 bands: a band at ∼90 kDa corresponding to the classical transcript, along with a shorter band, corresponding to an alternate DNM3 transcript (∼75 kDa). The fold change of DNM expression in shRNA-induced knockdown, compared with control (CTRL) cells (nontarget CTRL shRNA), is indicated as a percentage of CTRL. (C) WB staining of DNM3 in CHRF cells and MKs. The presence of the alternate (Alt) DNM3 band (arrow) is confirmed in primary human MKs (hMKs) differentiated from cord blood CD34 cells. (D) Plasma membrane fluorescence anisotropy measurement in CHRF cells with DNM2 and DNM3 knockdown. The measurement of plasma membrane fluorescence anisotropy values (r) in response to activation by PMA (1 µM) indicated a trend for decreased membrane fluidity, deformation, and remodeling for cells with single DNM knockdown and showed significant difference from CTRL for cells with double shDNM2 and shDNM3 knockdown (at 30 minutes). The stronger effect with double knockdown suggested that the actions of DNM2 and DNM3 overlap to a certain degree. Error bars indicate standard errors of the mean of ≥3 independent experiments. **P ≤ .01, ***P ≤ .001. MW, molecular weight.

Figure 2.

DNM activity is required for optimal MK migration. (A) Migration patterns of tracked CHRF cells without or with DNM knockdown on FN toward an SDF-1α gradient in the microfluidics-based µ-slide chemotaxis assay. (B) Quantification of directional migration of CHRF cells. ShDNM2 and shDNM3 CHRF cells exhibited a reduced directional migration toward an SDF-1α gradient (P < .005). (C) Schematic representation of the treatment of CD34⁺ cell-derived human primary MKs. Before transwell migration assays, we differentiated human primary MKs from cord blood CD34⁺ cells, which we would treat with shDNM2 or DNSR. (D) Quantification of MK migration in transwells toward a chemoattractant (SDF-1α) gradient, with the effect of DNSR. Migration was decreased by ∼45% in DNSR-treated MKs vs control (CTRL; paired Student t test P < .001). (E) Quantification of MK migration in transwells toward a chemoattractant (SDF-1α) gradient, with the effect of DNM2 knockdown. Migration was decreased by approximately half in shDNM2-treated MKs vs CTRL (paired Student t test P < .01). Error bars indicate standard errors of the mean of ≥3 independent experiments. *P ≤ .05, **P ≤ .01.

Figure 3.

DNMs regulate actin cytoskeleton polymerization, RhoA activation, and surface CXCR4 expression in human MKs. (A) Distribution of F-actin in MKs of vehicle-treated control (CTRL) and DNSR-treated MKs plated on FN and stained with phalloidin (red) and DAPI (blue), without and with DNSR treatment. Twenty-five or more primary MKs were analyzed by confocal immunofluorescence. Here shown are OMX super-resolution microscopy images for illustration purposes. DNSR-treated MKs showed more unevenly distributed, disorganized, and clumped F-actin (arrow) compared with CTRL cells; original magnification, 63× objective. Scale bar represents 2 µm. (B-C) F-actin polymerization in MKs without and with DNSR treatment. Flow cytometry representative histograms and quantification in 3 experiments using fluorophore-conjugated phalloidin staining in permeabilized and fixed MKs. Actin polymerization was decreased by ∼25% in DNSR-treated primary MKs (paired Student t test P ≤ .01). (D-E) F-actin polymerization in CHRF cells without and with DNM knockdown. Flow cytometry representative histograms and quantification in 4 experiments using fluorophore-conjugated phalloidin staining in permeabilized and fixed CHRF cells. CHRF with shDNM2 or double knockdown for DNM2 and DNM3 showed an ∼25% decrease in actin polymerization when compared with CTRL cells (P < .05). A similar trend was observed for CHRF cells with single shDNM3 knockdown. (F) RhoA activation quantification by G-LISA in MKs on FN. RhoA activation was reduced by half in DNSR-treated MKs (paired Student t test P < .05). (G-H) Surface CXCR4 expression in MKs. Flow cytometry representative histograms and quantification in 3 experiments. By flow cytometry, surface CXCR4 was slightly increased in DNSR-treated MKs (paired Student t test P < .05). Error bars indicate standard errors of the mean of ≥3 independent experiments. *P ≤ .05, **P ≤ .01. MFI, mean fluorescence intensity; OD, optical density.

Figure 4.

DNM inhibition affects ITGB1 activity and Rab11 cell distribution in human MKs. (A-B) Quantification of active ITGB1 at the surface of MKs. Representative flow cytometry histograms (A) and flow cytometry quantification (B) in 3 experiments measuring active ITGB1 at the surface of MKs after 2 hours on FN. ITGB1 activation was decreased by 17% in DNSR-treated MKs, when compared with control (CTRL) cells (paired Student t test P = .001). (C) Rab11 staining distribution in MKs on FN. MKs were stained with an antibody directed against Rab11 (green), a marker of recycling endosomes, and against activated ITGB1 (red). The nucleus is stained in blue. In the MKs that had not spread, Rab11 staining was more centrally clustered (arrow) in DNSR-treated cells, suggesting a decreased recycling process when compared with CTRL cells. Twenty-five or more primary MKs were analyzed. Images taken by Nikon A1R+ confocal microscope; original magnification, 60× Plan-Apochromat oil immersion lens. Scale bars represent 5 µm. (D-E) Quantification of active ITGB1 at the surface of CHRF cells transduced with nontargeting control or with shRNAs against DNM2 and DNM3. Representative flow cytometry histograms and flow cytometry quantification in 3 experiments in CHRF cells, activated with PMA (1 µM) and stained with an antibody specific for active ITGB1. Active surface ITGB1 in CHRF cells with shDNM2 and shDNM3 knockdown was reduced down to levels of 80% of CTRL cells. (F-G) Quantification of active ITGB1 at the surface of MEG-01 cells transduced with empty vector or with wild-type rat Dnm2 or human DNM3 or with DNM mutants. Flow cytometry representative histograms and quantification in 4 experiments in MEG-01 cells, activated with PMA (1 µM) and stained with an antibody specific for active ITGB1. In MEG-01 cells transduced with dominant-negative Dnm2-K44A or with truncated GTPase-less DNM3, active ITGB1 was reduced when compared with CTRL cells. Error bars indicate standard errors of the mean of ≥3 independent experiments. *P ≤ .05.

Figure 5.

ITGA2 and ITGAV surface expression is affected by DNM activity in human MKs. (A-B) Quantification of ITGAV (part of a receptor to FN) at the surface of CHRF cells. Flow cytometry representative histograms and quantification in 4 experiments measuring ITGAV at the surface of CHRF cells. ITGAV staining shows a decrease in surface expression in CHRF with DNM2 and DNM3 knockdown when compared with control (CTRL; by 22%; paired Student t test P < .05). (C-D) Quantification of ITGAV at the surface of MKs. Flow cytometry representative histograms and quantification in 4 experiments measuring ITGAV at the surface of MKs. (E-F) ITGAV staining also showed a consistent slight decrease in primary MKs treated with DNSR (by 10%) when compared with control (paired Student t test P < .05). Quantification of ITGA2 at the surface of MKs. Flow cytometry representative histograms and quantification in 4 experiments measuring ITGA2 at the surface of MKs. ITGA2 staining (part of a receptor to collagen) showed a decrease in surface expression in primary MKs treated with DNSR (by 26% when compared with CTRL; paired Student t test P < .05). Error bars indicate standard errors of the mean of ≥3 independent experiments. *P ≤ .05, **P ≤ .01.

Figure 6.

DNM activity is required for optimal PPF and adequate demarcation membrane system (DMS) development. (A-B) PPF representative fields on microscopy imaging and proplatelet (PPT) quantification. PPT stalks are shown with arrow. DNSR-treated MKs had reduced PPF by almost half (paired Student t test P ≤ .01). Images taken by a Nikon Eclipse TS100 (×40 objective) and INFINITY capture camera and software (Luminera), analyzed with INFINITY Analyze software (Luminera). Scale bar represents 100 µm. (C) GpIb (CD42; red) staining of MK plasma membrane and intracellular DMS. Clumps, clustering, and uneven distribution of the DMS were observed within DNSR-treated MKs when compared with control (CTRL) MKs (arrows; supplemental Videos 3 and 4). The CD41 staining (green) showed the same trend of poor distribution within DNSR-treated MKs when compared with CTRL MKs. The nucleus is stained in blue. Images taken by Nikon A1R+ confocal microscope under a 60× Plan-Apochromat oil immersion lens. Scale bar represents 5 µm. (D-E) Representative flow cytometry histograms and flow cytometry quantification of CD41 at the surface of MKs. DNSR-treated MKs showed increased surface expression of CD41 (by 25%; paired Student t test P < .05). Error bars indicate standard errors of the mean of ≥3 independent experiments. *P ≤ .05, **P ≤ .01.