Altered platelet-megakaryocyte endocytosis and trafficking of albumin and fibrinogen in RUNX1 haplodeficiency

Endocytosis is a major mechanism by which cells, including platelets and megakaryocytes (MK), take up diverse molecules to traffic them to distinct intracellular membrane compartments.^1-5 Platelet α-granules have numerous proteins; some, such as platelet factor (PF4) and von Willebrand factor (VWF), are synthesized by MK, whereas others, such as fibrinogen, albumin, immunoglobulin G (IgG), and factor V, are not synthesized by human MK but are incorporated into granules by endocytosis and endosomal trafficking.¹ In MK, after endocytosis, proteins sequentially traffic from early endosomes to multivesicular bodies to late endosomes, and finally to α-granules or to recycling endosomes for extrusion from the cell or to lysosomes for degradation.⁶ Our understanding of mechanisms regulating endocytosis, trafficking, and granule formation in MK and platelets is incomplete.

Transcription factor RUNX1 is a major regulator of definitive hematopoiesis, megakaryopoiesis, and platelet production.^7-9 Human RUNX1 haplodeficiency (RHD) owing to heterozygous germ line mutations is associated with familial thrombocytopenia, platelet dysfunction, α- and dense-granule deficiencies, and a predisposition to myeloid malignancies—a constellation termed familial platelet disorder with associated myeloid malignancy (FPDMM).^10-14 Platelet granules deficiencies are a hallmark of FPDMM,^11,14-16 but the underlying mechanisms are poorly understood.

In prior studies in a patient with FPDMM, we described the presence of α- and dense-granule deficiencies, and impaired activation-induced aggregation, secretion, protein phosphorylation (eg, pleckstrin and myosin light chain [MLC]), and αIIbβ3 activation.^11,16-19 On platelet expression profiling many genes were found to be downregulated,¹⁸ and several were shown to be direct RUNX1 targets,^15,20-22 including PF4²¹ and MYL9.²⁰ Our recent studies revealed that RAB31²³ and RAB1B²⁴—2 small GTPases involved in endosomal trafficking (both RUNX1 targets) are downregulated and associated with defective trafficking of VWF, mannose 6-phosphate receptor and epidermal growth factor receptor. An important finding in our patient was that platelets had decreased albumin, fibrinogen, and IgG¹⁹—proteins not synthesized by MK but incorporated by endocytosis.¹ To understand the mechanisms, we pursued the hypothesis that endocytosis and trafficking are impaired in platelets or MK in RUNX1-deficiency. Our studies provide the first evidence that FPDMM platelets have defective endocytosis of albumin, fibrinogen, and IgG. In RUNX1-deficient megakaryocytic human erythroleukemia (HEL) cells and MK differentiated from human CD34⁺ cells, uptake of albumin and fibrinogen was increased with impaired trafficking by distinct mechanisms leading to decreased cellular levels. There was upregulation of caveolin-1 (Cav1) and flotillin-1 (Flot1), 2 proteins linked to clathrin-independent endocytosis,^4,25 and of lysosomal marker lysosome associated membrane protein 2 (LAMP2)²⁶ and recycling endosomal marker RAB11,^26,27 both linked to endosomal trafficking. These studies provide the first evidence that MK endocytosis and trafficking are perturbed in RHD.

The patient is a previously described,^1-3 40-year-old White male with FPDMM with a point mutation in RUNX1 (RUNX1 c.352-1 G>T) in intron 3 at the splice acceptor site for exon 4, leading to a frameshift with premature termination in the conserved Runt homology domain. We performed studies in his 17-year-old daughter, who has the same mutation. Healthy control participants were recruited from staff and students at the Lewis Katz School of Medicine at Temple University. This research was approved by institutional human subjects review board, and all participants gave a written informed consent.

The following reagents were obtained: human fibronectin (EMD Millipore, Burlington, MA), poly-L-lysine solution (Sigma Life Science, St. Louis, MO), SPHERO Ultra Rainbow fluorescent particles (Spherotech, Inc, Lake Forest, IL), phorbol 12-myristate 13-acetate (PMA) (Enzo Life Sciences, Farmingdale, NY), human fibrinogen (Haematologic Technologies, Essex Junction, VT), and Pitstop 2 (Cayman Chemical, Ann Arbor, MI). Small interfering RNAs (siRNAs) against RUNX1, CAV1, FLOT1, IFITM3 interferon induced transmembrane protein 3, and CLTC clathrin heavy chain, and control siRNAs, were purchased from Santa Cruz Biotechnology Inc, (Dallas, TX). Supplemental Table 1 presents the antibodies and other reagents used.

Platelet-rich plasma was prepared by centrifugation (200 g, 20 minutes) from whole blood collected in 1 of 10 volume of 3.8% sodium citrate and incubated with carbacyclin (30 nM) for 20 minutes at room temperature (RT). Platelets were pelleted by centrifugation (650 g, 15 minutes, RT), resuspended in HEPES (4-(2-hydroxyethyl) piperazine-1-ethanesulfonic acid)-Tyrode buffer after washing twice by centrifugation (650 g, 10 minutes, RT) and used in experiments.

HEL cells obtained from the American Type Culture Collection (Rockville, MD) were cultured in RPMI 1640 medium (Cellgro, Manassas, VA), supplemented with 10% fetal bovine serum (FBS) (GE Healthcare, Mississauga, ON) and penicillin/streptomycin (100 U/mL per 100 mg/mL, Invitrogen) at 37°C in a humidified 5% CO₂ atmosphere. Cells were treated with 30 nM phorbol myristate acetate to induce megakaryocytic transformation.²⁸ PMA-treated HEL cells were transfected with 100 nM of RUNX1, CAV1, FLOT1, IFITM3, CLTC, or control siRNAs using Lipofectamine reagents (Invitrogen, Carlsbad, CA) and consisted of pools of 3 20 to 25 bp oligonucleotides (Santa Cruz Biotechnology, Dallas, TX). Cells were harvested at 24 to 48 hours after transfection and used for uptake studies.

Primary MK were grown in vitro from human CD34⁺ hematopoietic stem cell progenitors as described and validated for studies on RUNX1-deficiency.^29,30 CD34⁺ cells were infected with shRUNX1 or shNT-lentiviruses,^28,29 and cells expressing mCherry (mCherry⁺) were sorted on day 4 and cultured until days 11 to 12 to obtain RUNX-deficient (shRX) and control (shNT) MK used for studies.

Protein uptake and retention (up to 24 hours) was studied in washed platelets, PMA-treated HEL cells and CD34⁺ cell–derived MKs suspended in buffer using flow cytometry. Platelet suspensions (1 × 10⁶/mL) were incubated with fluorescence conjugated Cy3-albumin (10 μg/mL), Cy3-fibrinogen (50 μg/mL), or Cy3-IgG (20 μg/mL) at 37°C for different times. HEL cells or MK (1.5 × 10⁵) were incubated with fluorescent-conjugated albumin Alexa Fluor 488 (30 μg/mL), fibrinogen Alexa Fluor 647 (10 μg/mL), or IgG-Alexa Fluor 488 (30 μg/mL) at 37°C. Cells were fixed (2% paraformaldehyde) and uptake-retention was evaluated using a BD LSRII flow cytometry (San Jose, CA) and analyzed with FlowJo software, v10.5.3. The LSRII flow cytometer was calibrated using standard beads, SPHERO Ultra Rainbow Fluorescent Particles (Spherotech, Inc) on each experimental day. Where indicated, cells were incubated (15 minutes, RT) before uptake studies with 30 μM Pitstop 2 (Cayman Chemical).⁴ 

Washed platelets (1.5 × 10⁶/mL) were immobilized on coverslips precoated with 2 μg/mL human fibronectin (EMD Millipore, Rockville, MD) for 1 hour at 37°C; unbound platelets were removed by washing with HEPES Tyrode buffer (pH 7.4) and coverslips incubated with albumin Alexa Fluor 488, fibrinogen Alexa Fluor 647, or IgG-Alexa Fluor 488 (Jackson ImmunoResearch Lab, West Grove, PA) at 37°C for indicated times. Coverslips were rinsed with HEPES Tyrode buffer pH 7.4, fixed with 2% PFA for 15 minutes, and assessed using Epifluorescence microscope (EVOS FL Auto imaging). Anti-PF4 antibody was used to mark platelet α-granules.

HEL cells or MK were seeded on poly-L-lysine–coated coverslips and permeabilized with 0.1% Triton X-100 before immunostaining with antibodies as described.¹⁶ The antibodies used are shown in supplemental Table 1. Images were obtained on an EVOS microscope or Leica TCS SP5 confocal microscope, using a 63×/1.40 n.a. oil immersion objective at RT and EVOS or Leica imaging software, respectively. Postacquisition processing and analysis were performed with Adobe Photoshop and ImageJ³¹ and were limited to image cropping and brightness or contrast adjustments applied to all pixels per image simultaneously.

HEL cells (1 × 10⁶/mL) treated with control or RUNX1 siRNA were incubated with FITC (fluorescein isothiocyanate)-mouse antihuman CD41a clone–HIP8 or FITC-mouse antihuman PAC1 antibody (supplemental Table 1) to evaluate surface expression of αΙΙΒβ3 complex in resting state and after activation with adenosine diphosphate (ADP; 50 μM) or thrombin (10 U/mL) (Millipore Sigma, Billerica, MA) using flow cytometry.

Cell lysates collected in M-Per protein extraction reagent (Pierce-Thermo Scientific) with protease inhibitors (Enzo Life Sciences) were subjected to 10% or 12% sodium dodecyl sulfate-polyacrylamide gel electrophoresis, transferred to 0.2 μM nitrocellulose membranes and probed with antibodies. Proteins were detected with IRDye-labeled secondary antibodies using Odyssey Infrared Imaging system (Li-Cor Biosciences).

Results were expressed as mean ± standard error of the mean. Differences were compared using Student t test or 1- and 2-way analysis of variance, using the GraphPad Prism, version 8 (GraphPad Software) and considered significant at P < .05.

The uptake of Cy3-Chrom-albumin, Cy3-Chrom-fibrinogen, and Cy3-Chrom-IgG over 15 to 30 minutes was markedly decreased in platelets from the father and daughter with FPDMM compared with a healthy donor (Figure 1A). Because these proteins are incorporated into α-granules,¹ we assessed PF4 as an α-granule marker. It is synthesized by MK and is a RUNX1 target gene²¹; in prior studies, platelet PF4 protein and transcripts were decreased in the father.¹⁸ PF4 immunofluorescence of cells immobilized on fibronectin was reduced in the father, consistent with α-granule deficiency, but not the daughter (Figure 1B). We studied albumin Alexa Fluor 488 and fibrinogen Alexa Fluor 647 uptake in platelets immobilized on fibronectin using immunofluorescence microscopy (Figure 1C-D). The uptake of albumin was decreased at 1 and 15 minutes in the father and daughter compared with the healthy donor (Figure 1C). At 1, 5, and 15 minutes the father’s platelets had decreased fibrinogen; many platelets entirely lacked fibrinogen, whereas others had substantial uptake (Figure 1D), indicating heterogeneity among platelets. In the daughter the uptake appeared normal (Figure 1D). Overall, the uptake of albumin, fibrinogen, and IgG as analyzed by flow cytometry, was decreased in both patients; in the immobilized platelets, albumin uptake was decreased in both patients; whereas fibrinogen uptake was decreased in father but not in daughter, possibly because of differences in platelet adhesion between the 2 patients.

We assessed endocytosis in RUNX1-deficient HEL cells. In preliminary studies, the uptake of albumin Alexa 488 over 24 hours in PMA-treated HEL cells was concentration- (10-100 μg/mL) and time-dependent over 24 hours. With siRNA RUNX1 KD, albumin uptake over 24 hours was increased compared with control cells (Figure 2A). Cav1 and Flot1 are 2 major proteins involved in endocytosis.^2-5 Cav1 regulates albumin uptake in several cell types.^32,33 On immunoblotting, control HEL cells showed minimal Cav1, which increased markedly on RUNX1 downregulation (Figure 2B). Although Flot1 was detected in control HEL cells, its expression increased further on RUNX1 downregulation (Figure 2B). Notably, siRNA CAV1 downregulation did not appear to decrease albumin uptake compared with control cells; however, it abrogated the increase in albumin uptake resulting from RUNX1 KD (Figure 2A). FLOT1 KD did not alter albumin uptake, and in contrast to the findings with CAV1 downregulation, did not abrogate the increase in albumin uptake noted on RUNX1 downregulation (Figure 2C). Immunoblotting confirmed the expected decreases in the respective proteins on KD of RUNX1, CAV1, and FLOT1 and the increases in Cav1 and Flot1 on RUNX1 downregulation (Figure 2B). CAV1 downregulation did not affect Flot1 expression, and vice versa (not shown). Thus, RUNX1 downregulation increased albumin uptake along with increased Cav1 and Flot1; the increased uptake was dependent on Cav1.

We examined other major proteins implicated in endocytosis. Clathrin-mediated endocytosis regulates fibrinogen uptake¹ and is dependent on dynamin-2,^3,5,34 and GTPase ARF6 (ADP ribosylation factor 6).³⁵ No changes were noted in these on RUNX1 downregulation (Figure 2D).

In above experiments, HEL cells were cultured in an albumin-rich medium with FBS. We assessed albumin in washed HEL cells lysates after 24 to 48 hours of siRNA RUNX1 KD. Albumin was increased upon RUNX1 KD (Figure 2E). To assess the fate of endocytosed albumin, siRNA-treated HEL cells were resuspended in buffer without FBS, and levels assessed at 30 minutes, 2 hours and 24 hours in lysates. Albumin was higher in RUNX1-deficient cells at each time (Figure 2F). Under both conditions, the 24 hour albumin was lower than that at baseline, indicating loss over time, particularly in RUNX1-deficient cells.

In preliminary studies, HEL-cell fibrinogen uptake was dose- and time-dependent over 24 hours period. Fibrinogen Alexa Fluor 647 uptake over 24 hours was increased on RUNX1 KD compared with control cells (Figure 3A). CAV1 downregulation neither affected fibrinogen uptake in control cells nor the increased uptake on RUNX1 downregulation (Figure 3A). FLOT1 downregulation did not decrease uptake in control cells, but partially decreased it in RUNX1-deficient cells (Figure 3B). Thus, as with albumin (Figure 2), RUNX1 KD increased fibrinogen uptake but was unaffected by CAV1 KD.

To understand the fate of endocytosed fibrinogen, we assessed fibrinogen using immunoblotting in washed HEL cells after 24-hour incubation in media containing FBS and fibrinogen (50 μg/mL); the levels were lower in RUNX1-deficient cells compared with control cells (Figure 3C). In cells maintained in buffer over subsequent 24 hours, fibrinogen was lower in RUNX1-deficient cells at 30 minutes, 2 hours, and 24 hours (Figure 3D) indicating continued loss.

Fluorescent-conjugated IgG uptake was increased in RUNX1-deficient compared with control cells (supplemental Figure 1).

To understand the trafficking of albumin, we performed immunofluorescence microscopy of HEL cells, stained for Cav1, Flot1, and RAB11—a marker for recycling endosomes,^26,27 and LAMP2—a marker for lysosomes.²⁶ In control cells at 30 minutes, albumin colocalized with Cav1 but not with Flot1 (Figure 4A). Upon KD of CAV1 alone or together with RUNX1, albumin was decreased indicating that CAV1 regulates albumin uptake (Figure 4A). On RUNX1 KD, there was loss of albumin colocalization with Cav1 and a strikingly increased colocalization with Flot1 (Figure 4A). There was no colocalization in the nuclear area (not shown). At 120 minutes, in the control cells, albumin colocalization with Cav1 was not as strong as observed at 30 minutes, and there was some albumin colocalization with Flot1 (Figure 4A). However, on RUNX1 KD, there was little albumin colocalization with Cav1 and a strong colocalization with Flot1 (Figure 4A). With KD of CAV1 alone or of RUNX1 + CAV1 there was no albumin in the cells. At 24 hours, albumin did not strongly colocalize with Cav1 or Flot1 in control or RUNX1-deficient cells (not shown). With respect to RAB11, immunoblots showed an increase on RUNX1 KD (Figure 4B). However, at 30 and 120 minutes (Figure 4B) and at 24 hours (not shown), there was no albumin colocalization with RAB11 in control or RUNX1-deficient cells, suggesting that albumin does not traffic to recycling endosomes. In podocytes and other cells, albumin is degraded in lysosomes.^36,37 LAMP2 levels were increased on RUNX1 KD on immunoblotting (Figure 4C). We observed moderate colocalization of albumin with LAMP2 at 30 minutes in control cells and on RUNX1 KD (Figure 4C), but they did not appear to be different. At 120 minutes, the colocalization of albumin with LAMP2 appeared more prominent with RUNX1 KD than in control cells (Figure 4C). Overall, these findings suggest RHD upregulates Cav1 and Flot1 (Figure 2), with altered albumin trafficking from Cav1 to Flot1 compartment, and increased colocalization with LAMP2 but not RAB11.

Fibrinogen uptake was unaffected by CAV1 KD (Figure 5A), suggesting that it is not Cav1-dependent. In control cells or on RUNX1 KD, we did not observe significant fibrinogen colocalization with Cav1 or Flot1 at 30 or 120 minutes (Figure 5A). Findings with respect to RAB11 were different. At 30 minutes, we observed low to moderate fibrinogen colocalization with RAB11 in control cells, which increased with RUNX1 KD alone or together with CAV1 (Figure 5B). However, colocalization with RAB11 was more prominent at 120 minutes (Figure 5B) and 24 hours (not shown). As noted earlier, RAB11 levels were increased with RUNX1 downregulation (Figure 4B). These findings suggest that RUNX1 KD causes trafficking of fibrinogen toward recycling endosomes. In both control cells and RUNX1-depleted cells, we observed moderate fibrinogen colocalization with LAMP2 at 30 minutes but was not different (Figure 5C).

MK fibrinogen uptake is αIIbβ3 receptor-mediated.^1,35,38,39 Our patient platelets have near normal surface αIIbβ3 expression but impaired agonist-induced activation (PAC binding) and MLC phosphorylation (pMLC),¹⁷ the latter related to decreased expression of RUNX1-regulated gene MYL9.¹⁸ Because fibrinogen uptake was increased on RUNX1 KD (Figure 3), we assessed whether the HEL cells recapitulate these platelet abnormalities or show increased αIIbβ3. αIIbβ3 surface expression in RUNX1-deficient cells was comparable to that in control cells (Figure 6A); on immunoblotting αIIb was decreased with unchanged β3 (Figure 6B). Because surface αIIbβ3 drives fibrinogen uptake (Figure 6A), the impact of decreased cell lysate αIIb would be negligible. In HEL cells with RUNX1 KD, thrombin and ADP-induced PAC1-binding (Figure 6C), total MLC protein and, correspondingly, thrombin-induced pMLC (Figure 6D) were decreased. Thus, RUNX1-deficient HEL cells recapitulate the platelet defects,^17-19 validating the HEL-cell model. When pMLC was expressed as a fraction of the total myosin, it was comparable between RHD HEL cells and control cells (Figure 6D), suggesting that phosphorylation mechanisms (MLC kinase-driven) were preserved. These studies indicate that increased fibrinogen uptake is not due to upregulated αIIbβ3 expression. Because αIIbβ3 activation is impaired in these cells (Figure 6C), it suggests that integrin activation is not a major driver of fibrinogen uptake.

Increased IFITM3 expression has been reported to enhance fibrinogen endocytosis in nonviral sepsis.³⁹ In HEL cells, it was increased on RUNX1 KD (Figure 6E). However, IFITM3 KD did not decrease fibrinogen uptake in control or RUNX1-deficient cells (supplemental Figure 2A). Fibrinogen uptake is clathrin-dependent,^1,39 and it was inhibited by clathrin inhibitor Pitstop 2 (supplemental Figure 2B).

Albumin (Figure 7A) and fibrinogen (Figure 7B) uptake at 24 hours were increased in shRX compared with that in control shNT-MK, along with increased Cav1, Flot1, LAMP2, and IFITM3 (Figure 7C), whereas RAB11 was decreased (Figure 7C). On analysis with immunofluorescence microscopy, it was found that there was little albumin colocalization with Cav1 in shNT or shRX MK. In shRX MK, there was increased albumin colocalization with Flot1 (30 and 120 minutes) and LAMP2 at 30 minutes compared with shNT-MK (Figure 7D). There was minimal fibrinogen colocalization with RAB11 or LAMP2 in shNT-MK (supplemental Figure 3). In shRX MK, there was no increased fibrinogen colocalization with RAB11, as noted in HEL cells (Figure 4B). However, RAB11 was decreased in shRX MK (Figure 7C) in contrast with an increase in RUNX1-deficient HEL cells (Figure 4B). Fibrinogen colocalization with LAMP2 appeared minimally increased at 120 minutes in shRX MK (supplemental Figure 3). Further studies, beyond the scope of these studies, are needed to further delineate alterations in trafficking. The differences from the findings in HEL cells may reflect differences in cell maturity. Overall, shRX MK recapitulate the enhanced protein uptake, increases in Cav1, Flot1, LAMP2, and IFITM3 and increased albumin colocalization with Flot1 and LAMP2.

Compared with 3 healthy donors, Flot1, LAMP2, and RAB11 were not increased in FPDMM platelets (Figure 7E). Cav1 was undetectable in all donors, as reported.³⁴ IFITM3 was undetectable in the patient and 1 healthy donor. On platelet transcript profiling of our patient, these genes were not found among those increased.¹⁸ 

Endocytosis is the mechanism by which proteins, not synthesized by MK, are incorporated into α-granules.¹ Platelet albumin, fibrinogen, and IgG were observed to be decreased in our patient¹⁹ (and Figure 1). We provide the first evidence that platelet endocytosis is defective in our patients (Figure 1), constituting 1 explanation. In addition, we provide evidence that RUNX1-deficient HEL cells and primary MK have increased endocytosis but defective trafficking of albumin and fibrinogen, by distinct mechanisms, leading to decreased levels. Multiple pathways drive endocytosis, mediated by clathrin, Cav1, and Flot1, and others.^2-4,40 Cav1 and Flot1, but not clathrin, were increased in RUNX1-deficient HEL cells (Figure 2B) and MK (Figure 7C). CAV1 KD nearly eliminated intracellular albumin (Figure 4) and the increase in albumin uptake with RUNX1 KD (Figure 2), indicating that Cav1 regulates albumin uptake, as shown in renal podocytes.^32,CAV1 KD did not reduce intracellular fibrinogen (Figure 5), and CAV1 KD did not abrogate the increased fibrinogen uptake on RUNX1 KD (Figure 3), indicating that distinct mechanisms mediate albumin and fibrinogen endocytosis. Flot1 may have a role in fibrinogen handling, suggested by partial inhibition noted on FLOT1 KD in RUNX1-deficient cells (Figure 3B). MK from induced-pluripotent stem cells from a patient with FPDMM had increased uptake of factor V,⁴¹ a protein not synthesized by human MK. Thus, endocytosis of multiple proteins is increased in RHD. Interestingly, in NBEAL2-null mouse (gray platelet syndrome) MK fibrinogen uptake was normal but associated with defective retention,⁴² indicating differences from RHD.

MK fibrinogen uptake is αIIbβ3 receptor-mediated^1,35,39 and clathrin- and ARF6- dependent.^1,35 In RUNX1-deficient HEL cells, there was no increase in surface αIIbβ3 (Figure 6), clathrin, or ARF6 (Figure 2D). Studies in sepsis advance IFITM3 as a regulator of platelet or MK fibrinogen endocytosis.³⁹ IFITM3 was upregulated in RUNX1-deficient HEL cells and primary MK (Figures 6E and 7C), but IFITM3 KD did not inhibit fibrinogen uptake (supplemental Figure 2), suggesting that in these cells it may not be IFITM3-driven. IFITM3 effect in sepsis³⁹ may require other interferon-induced genes as well.^43,44 

Our studies suggest that RUNX1 deficiency alters intracellular trafficking of albumin and fibrinogen and in a differential manner. On RUNX1 KD, there was a decrease in albumin colocalization with Cav1 and an increase with Flot1 (Figure 4). Cav1 and Flot1 are distinct clathrin-independent mechanisms of endocytosis,^2,45 and they do not colocalize.² The cargo of Flot1-positive endosomes has been shown to be delivered to lysosomes^46,47 with Flot1 present on cytosolic face of lysosomes.^48,49 Lysosomal marker LAMP2 was increased on RUNX1 downregulation, and albumin colocalized with LAMP2 (Figures 4 and 7). Albumin has been reported to traffic to lysosomes in which it is degraded.^32,37 Thus, albumin mistrafficking with lysosomal degradation is a mechanism in RHD-MK with decreasing levels over time and in platelet progeny.

In MK, after endocytosis, fibrinogen sequentially traffics from early endosomes to multivesicular bodies to late endosomes, and then to α-granules or recycling endosomes.⁴² We show that fibrinogen trafficking is perturbed in RUNX1-deficient cells. RAB11 was upregulated (Figure 4B) and fibrinogen colocalized with RAB11 (Figure 5B), involved in protein recycling at plasma membrane,^27,50 and cell fibrinogen was decreased (Figure 3C). We postulate that there is enhanced fibrinogen recycling in RHD-MK, as described in NBEAL2^−/− MK,⁶ although lysosomal degradation is not excluded. Thus, the released platelets in RHD have lower fibrinogen plus impaired uptake.

Our studies in primary MK support remarkably well the findings in HEL cells with respect to the changes with RHD-increased uptake of albumin and fibrinogen, upregulation of several involved proteins and some aspects of mistrafficking (Figure 7). Additional studies are needed in MK to delineate these mechanisms.

The overall model (see visual abstract) that emerges is that RUNX1-deficient MK have enhanced endocytosis of albumin and fibrinogen but is associated with defective trafficking and loss of protein via lysosomal degradation or recycling, leading to decreased levels in MK, despite the enhanced endocytosis, and subsequently in platelet progeny. In addition, patient platelets had impaired endocytosis. Studies in a larger number of patients are needed to establish this and the involved proteins or genes. The impaired trafficking is likely related to aberrant granule formation and driven by dysregulation of numerous RUNX1-regulated genes and secondary somatic gene alterations.^18,29,51 We have shown that RUNX1-regulated GTPases, RAB1B,²⁴ and RAB31,²³ are decreased in RHD platelets and MK with altered endosomal trafficking of VWF, mannose 6-phosphate receptor, epidermal growth factor receptor, and strikingly enlarged early-endosomes.²³ These likely contribute to the protein mistrafficking.

IFITMs are important effectors in innate immunity and cancer biology,^43,44 and RUNX1 negatively regulates neutrophil cytokine production.^52,53 IFITM3 upregulation in RUNX1-deficient MK (Figures 6 and 7) is novel and maybe relevant to autoimmune manifestations^12,54 and leukemic predisposition of FPDMM. RUNX1 may negatively regulate IFITM3; platelet IFITM3 messenger RNA is elevated in FPDMM.⁵¹ IFITM3 is implicated in the pathogenesis of malignancies,^43,44,55,56 and acute myeloid leukemia patients with high expression have worse prognosis.⁵⁷ 

Our studies advance the complex mechanistic basis of the FPDMM abnormalities in α-granules and their cargoes. These include dysregulated endocytosis and vesicular trafficking, poorly understood in platelet and MK but shared cellular processes potentially relevant to FPDMM manifestations beyond defective hemostasis.

The authors are grateful to the patients who have supported our studies over many years. The authors thank David E. Ambrose and Amir Yarmahmoodi for discussions and assistance in the flow cytometry studies.

This study was supported by research fundings from National Institutes of Health National Heart, Lung, and Blood Institute (grants R01 HL137376 and R01 HL109568 to A.K.R.; grants R01 HL137207 and HL159006 to L.E.G.; and grant R35 HL150698 to M.P.), the American Heart Association Transformational Project Award (grant 20TPA35490278) to L.E.G., and a RUNX1 Research Program/Alex’s Lemonade Foundation grant to M.P.

Contribution: F.D.C.-C., G.M., and L.G. performed the research, analyzed the data, and contributed to manuscript writing; J.W. performed the research and analyzed the data specifically for the immunofluorescence studies; A.M.A. performed the research; K.L. and M.P. generously provided the megakaryocytes differentiated from the human CD34⁺ cells with RUNX1 knockdown and their expertise; A.K.R. and L.E.G. conceived and designed, performed the research, interpreted data, and wrote the manuscript; and all authors contributed to the manuscript, read, and approved the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: A. Koneti Rao, Sol Sherry Thrombosis Research Center and Section of Hematology, Lewis Katz School of Medicine at Temple University, 3420 N. Broad St, 204 MRB, Philadelphia, PA 19140; email: koneti@temple.edu.