During megakaryopoiesis, megakaryocytes (MKs) undergo cellular morphological changes with strong modification of membrane composition and lipid signaling. Here, we adopt a lipid-centric multiomics approach to create a quantitative map of the MK lipidome during maturation and proplatelet formation. Data reveal that MK differentiation is driven by an increased fatty acyl import and de novo lipid synthesis, resulting in an anionic membrane phenotype. Pharmacological perturbation of fatty acid import and phospholipid synthesis blocked membrane remodeling and directly reduced MK polyploidization and proplatelet formation, resulting in thrombocytopenia. The anionic lipid shift during megakaryopoiesis was paralleled by lipid-dependent relocalization of the scaffold protein CKIP-1 and recruitment of the kinase CK2α to the plasma membrane, which seems to be essential for sufficient platelet biogenesis. Overall, this study provides a framework to understand how the MK lipidome is altered during maturation and the effect of MK membrane lipid remodeling on MK kinase signaling involved in thrombopoiesis.
Deriving from pluripotent hematopoietic stem cells in the bone marrow, megakaryocytes (MKs) are responsible for the production of platelets, thus being essential for hemostasis and vascular integrity. Since MKs produce thousands of platelets, irregularities in MK differentiation (megakaryopoiesis) affect platelet generation or function and can result in clinically significant disorders. Thrombocytopenia or impaired platelet function might lead to disrupted primary hemostasis with increased risk of bleeding. In contrast, elevated platelet counts (thrombocythemia) or excessive platelet activation increases the risk for thrombotic events and ischemic diseases1.
After the hematopoietic stem cell was discovered more than 150 years ago2,3, the description of megakaryopoiesis at the omics scale is still in its infancy, although the cellular processes underlying megakaryopoiesis are now well defined4. MKs become polyploid during their maturation owing to endomitotic processes and reach diameters of up to 100 µm. However, they only occur with a frequency of 0.2% when compared with other nucleated blood cells, making them difficult to study. The entire process of megakaryopoiesis is accompanied by a substantial membrane reorganization, including shaping a lobulated nuclear envelope, packing granules, generating the lipid-rich demarcation membrane system5, and MK polarization toward the protrusion of proplatelets into the sinusoids of the bone marrow6. Altogether, these processes make it necessary to adapt membrane properties constantly. In particular, sphingolipid metabolism and signaling7,8,9 are indicated in the elongation of proplatelet extensions and the shedding process of platelets. Recently, there is growing scientific evidence that de novo lipogenesis may modulate MK maturation and platelet production10. It has been reported that maturating MKs incorporate fatty acids (FAs) released by adipocytes closely located to MKs in the bone marrow to facilitate thrombopoiesis11. Consequently, the FA transfer from adipocytes to MKs has important clinical implications in obesity-related cardiovascular thrombotic complications.
However, despite its high clinical importance, the MK lipidome composition is still ill defined, and the lipid-dependent processes during MK maturation and platelet biogenesis are largely unknown.
The existence of more than 350 lipids in platelets12 with the capability to influence membrane geometry and platelet signaling13 demands a systematic large-scale modulation potential of the lipid metabolism in MKs. Over the last 10 years, mass spectrometry (MS) has evolved into the state-of-the-art technology for lipid analysis. The improvements in sensitivity, speed and resolution, coupled with developments in systems biology14, ease of access to lipid databases15 and search engines, and the availability of lipid standards for accurate quantification, have made it possible to explore various aspects of lipid function and regulation16. Present-day lipidomics tools provide access to understand lipids’ complexity, homeostatic regulation, and role in differentiation, thus linking lipids to diseases and cellular impairments such as platelet dysfunction. Therefore, it is astonishing that neither a quantitative lipid inventory nor a map of the lipid metabolism of MKs is currently available.
Additionally, information gained from multiomics is more valuable when extracted from multiple layers of evidence of one biological sample. This accounts for missing values and points to new molecular mechanisms and interactions. Although lipidomics and proteomics have been successfully applied to investigate different blood cells, the potential of multiomics has yet to be fully explored. Here, we established a multiomics extraction strategy and quantitative MS workflow to determine lipid metabolism and its modulating effect on megakaryopoiesis and proplatelet formation. Using this unique approach, we were able to define regulatory metabolic mechanisms shaping the MK lipidome during MK maturation. These mechanisms directly influence the processes that are critically involved in thrombopoiesis, and their inhibition results in profound thrombocytopenia.
Dynamic lipid metabolism modulation in megakaryopoiesis
For multiomics method development, hematopoietic stem cells were isolated from bone marrow of 10–14-week-old male mice and subjected to a 7-day differentiation protocol. The SIMPLEX workflow (Fig. 1a), which enables simultaneous lipid and protein sample preparation, was used to determine their general molecular composition17,18,19. The differentiation efficiency was monitored by immunocytochemistry staining of the MK surface marker GPIb and the nuclear lobulation using a DRAQ5 DNA stain (Fig. 1b). The MS-based global proteomics analysis was integrated with top-down shotgun and targeted lipidomics to establish the multiomics workflow (Fig. 1a). In total, 4,651 proteins, with two or more unique peptides, were identified and relatively quantified by nano-liquid chromatography (nLC) high-resolution MS. Across the time course of differentiation, comparing all days, 3,152 proteins were significantly regulated, with approximately 1,908 proteins displaying continuous upregulation, 1,189 showing downregulation, and about 55 were transiently regulated. During MK maturation, protein regulation mostly occurred between days 1 and 3, with 593 proteins being upregulated and 455 downregulated (Extended Data Fig. 1a). Computing fuzzy c-means clustering of all regulated proteins (P < 0.01) from day 0 to day 7, we identified 39 distinct clusters using a similarity threshold of 85% (Fig. 1c and Extended Data Figs. 2 and 3). Here, 607 proteins showed an overall downward trend, whereas 979 proteins showed an upward trend. In a subsequent pathway enrichment analysis, considering significantly regulated proteins from day 0 to day 7 with a log2(fold change) of ≥2 or ≤−2 (Fig. 1d), pathway hallmarks of megakaryopoiesis were enriched, such as platelet and extracellular matrix receptor activation (Fig. 1e). However, most strikingly, seven lipid-specific pathways were identified under the top 15 most enriched biosynthesis pathways, ranging from steroid biosynthesis over the PPAR signaling pathway to FA elongation (Fig. 1e), pointing to a strong dependency of MK differentiation on lipid metabolism. The data were particularly analyzed for markers to underscore the enrichment analysis and further evaluate the differentiation process and the discovered link to lipid metabolism. MK differentiation markers such as RUNX1, RUNX3 and GATA1 (refs. 20,21,22) and surface markers such as GPIb, CD36, VWF, GPV, GPNMB and integrins23,24,25,26,27 were monitored and shown to be highly regulated (Fig. 1f). Moreover, specific domains of lipid metabolism, such as the FA receptors, transporters, the FA synthetase itself or mitochondrial FA importers are highly upregulated in maturating MKs (Fig. 1g and Extended Data Fig. 1b). Futhermore, the metabolism of complex lipids such as phospholipids, sphingolipids and sterols is elevated, whereas other enzymes derived from oxylipin metabolism were downregulated or unregulated as indicated by their corresponding metabolizing enzymes (Fig. 1g and Extended Data Fig. 1c). Using the latest MS technologies, we assembled a lipid metabolic network of more than 300 proteins involved in lipid transport, synthesis and degradation that shows substantial regulation during MK maturation (Extended Data Fig. 1d). This strong metabolic rewiring at the protein level raised the questions as to what extent this is reflected in the lipidome and if this multiomics-derived information can be used to generate novel hypotheses.
The MK membrane lipidome has specific signatures
For global quantitative lipidomics of MKs during maturation, we used the established workflow (Fig. 1a) integrating shotgun and targeted lipidomics to detect both low-abundant lipid species (for example, ceramides) and major membrane components (for example, glycerophospholipids) simultaneously. All lipid molecules were sequenced and their concentrations were reported using internal standards.
To ensure lipid identification with high confidence, all lipid molecules were structurally characterized by tandem mass spectrometry (MS/MS), enabling the determination of the number of carbon atoms and double bonds for each FA chain (Extended Data Figs. 4 and 5). Knowledge about the FA composition is crucial, as rearrangement and exchange of FAs constantly occur during fundamental cell-fate decisions28,29 or differentiation30. Furthermore, it contributes to the physicochemical features of the membrane, including lateral diffusion and stiffness, but most importantly, it provides the precursor reservoir of many signaling molecules12,13. We structurally characterized and identified 473 lipid species in differentiated MKs (Fig. 2) originating from the main lipid categories glycerophospholipid (GP; 343), glycerolipid (GL; 46), sphingolipid (SP; 76) and sterol (ST; 8), thereby covering 24 different lipid classes.
Quantitative lipid analysis was executed using internal standards that co-ionize with the target analyte. Lipids were normalized based on a fixed number of cells and the protein amount. Assembling of the quantitative results revealed a dynamic range of six orders of magnitude similar to the platelet lipidome (Fig. 2a,b)12. In mature MKs, low-abundant species such as the signaling molecule sphingoid base phosphate (SPBP) 18:1;O2 (5 pmol mg−1) were detected alongside major structural components such as cholesterol (48,523 pmol mg−1) and PC 16:0_18:1 (11,427 pmol mg−1) (Fig. 2b). Most lipid classes displayed a quantitative distribution of over two orders of magnitude. In contrast, lysophospholipid species, which have signaling capabilities, had a narrower range, likely representing a more tightly controlled metabolism (Fig. 2a). Our evidence shows that 60% of the entire lipid mass is accounted for by 15 lipids, and 70% by 29 lipids (Fig. 2c), making them essential building blocks for the membrane integrity of the MK lipidome.
Compared with platelets, in which 15 lipids already cover 70% of the lipid mass12, the MK lipidome seems to be twice as complex, with more lipids contributing to membrane properties. The most abundant lipid classes detected within mature MKs were cholesterol (ST), phosphatidylcholine (PC), phosphatidylethanolamine (PE), phosphatidylinositol (PI), phosphatidylserine (PS) and sphingomyelin (SM) (Fig. 2d).
Interestingly, besides arachidonic acid (FA 20:4), which dominated the top 15 lipids in platelets, here eicosatrienoic acid (FA 20:3) is also found under the most abundant molecules in MKs, indicating slight differences in the FA composition of major lipid classes. These differences underscore the likelihood that not only the demarcation membrane system of MKs determines the platelet lipidome, but also subsequent processes such as aging and interaction with the microenvironment during circulation are involved in shaping platelet membranes. Nevertheless, critical precursors for platelet signaling like PI 18:0_20:4 are already abundantly available in MKs. Comparing fully mature MKs against the lipidome of human31 and mouse12 platelets reveals an equal basis for various lipid classes such as PE, PS, ST, cholesterol ester (SE), PC-ether lipid (PCO), triacylglycerol (TG) and lyso-PE (LPE) (Fig. 2e). However, PC, PI, and PE-ether lipid (PEO), and especially SP classes such as ceramide (Cer) and hexosylceramide (HexCer), display very distinct abundances. Currently, we can only speculate why this is the case. Platelets lose their ability to synthesize SPs de novo32; therefore, higher levels in MKs are likely. Higher Cer levels stabilize Cer-rich platforms, which are needed to preserve multiple signaling processes or to be precursors themselves, steering megakaryopoiesis and thrombopoiesis9,33.
Anionic membrane remodeling in MK maturation
To gain further insight, we analyzed the MK lipidome on days 1, 3 and 7. Generally, the total membrane content was found to be increased within the first 3 days, as also observed in the lipids-to-protein ratio. Overall, 337 lipids were shared across all days, whereas approximately 10 lipids were distinct for specific days (Fig. 3a). The lipidome seems to be rather stable, as 81% of lipids were not regulated during differentiation (Fig. 3b). The 19% of lipids that were regulated (fold change of ≥2 or ≤−2; P ≤ 0.05) are mainly derived from low-abundant lipids. Nevertheless, 15 species belonging to 75% of the membrane lipidome are also altered, indicating a change in membrane properties (Fig. 3c). Therefore, we aimed to elucidate higher organizational rearrangements in membranes (Fig. 3d–h). First, we analyzed the coregulation of 506 lipids at the individual molecular lipid species level using absolute concentrations, revealing that most correlated lipids can be found within, but not across, classes (Fig. 3d). Applying the Pearson correlation computed for any lipid pair, 18 distinct clusters of correlated and anticorrelated lipids were identified during differentiation (Fig. 3e). As expected, GPs and SPs were distributed over all clusters, whereas STs were observed only in clusters C8–C13. However, lipid abundance and the individual alterations of each species are difficult to access from a hierarchical view. Therefore, the lipid–lipid correlation matrix was transformed into a network (Fig. 3f–h). Here, most lipid regulation appears between days 1 and 3, and only minor changes can be observed afterward, indicating that the membrane composition is determined relatively early during megakaryopoiesis. Interestingly, similar trends could be observed in the proteomic results (Extended Data Fig. 1a).
To further dissect the reorganization of the MK lipidome during differentiation, we carried out a quantitative analysis at the lipid category, lipid class and molecular species levels, as well as on the corresponding FA composition using absolute concentrations. By investigating the lipidome-wide class-specific representation (Fig. 4a), the results obtained earlier could be emphasized. More specifically, many lipid classes show regulation early on and are rather unchanged in the late stage of differentiation after day 3. Most significantly regulated classes like diacylglycerol (DG), TG, PS, PCO, PE, phosphatidylglycerol (PG), lyso-PG (LPG), PI, lyso-PI (LPI) and sphingoid base (SPB) follow this trend at the class and individual lipid species levels with PG and its lyso forms that are further upregulated at day 7 (Fig. 4b,c). LPI and PCO show opposing trends and are downregulated. Lipids of high interest were validated by high-resolution targeted LC–MS/MS (Extended Data Figs. 6 and 7). To prove that lipid changes are not mirroring the lipid composition of the media or are induced by apoptosis, we analyzed the fetal bovine serum (FBS), determined apoptotic markers, conducted a cell vitality assay, and proved by surface labeling that the PS amount is not increased in the outer membrane leaflet (Extended Data Fig. 8a–d).
Regarding molecular lipid composition, FA shifts can be observed toward a more unsaturated membrane (Fig. 4d,e). However, this is mainly due to an increased level of single monounsaturated FAs rather than total polyunsaturated FAs (PUFAs). Here, a decrease in arachidonic acid with a balancing increase of FA 20:3 can be observed (Fig. 4e). Nevertheless, the total PUFA levels remain unchanged (Fig. 4f). Interestingly, PUFA lipids are more abundant in platelets than in MKs, supporting the hypothesis that the lipidome of platelets is still being formed after release from MKs (Fig. 4f).
Moreover, we observed an increase in odd long-chain FAs (Fig. 4d) from day 1 to day 3, which is likely due to increased branched-chain FAs and their oxidations and unraveled that lysolipids with signaling capabilities such as LPG, lyso-PC (LPC), lyso-PA (LPA) and SPB are upregulated until the end of MK differentiation. Our data reveal that megakaryopoiesis is likely modulated from different mechanisms such as (1) lipidome rearrangement (membrane charge, for example, PG), (2) modulation of the FA lipid composition (membrane fluidity, FA 18:1) and (3) the production of signaling molecules (DG, LPG and SPB).
Phospholipid synthesis is essential for proplatelet-forming MKs
For validation of our previous data, we inhibited de novo phospholipid biosynthesis at two initiation points (Fig. 5a). In this regard, we added a long-chain acyl-CoA synthetase (ACSL) inhibitor (triacsin C), an inhibitor of glycerol-3-phosphate acyltransferase (GPAT) (FSG67), or vehicle control to the freshly isolated bone marrow cell suspension and collected thrombopoietin (TPO)-stimulated MKs on day 7. Nontreated day-0 MKs were used as baseline control, and changes for all lipid classes were calculated as ratios relative to the control. Treatment with either inhibitor diminished the production of almost all phospholipids, including anionic lipids such as PG, PI and PS (Fig. 5b and Extended Data Fig. 9a). Interestingly, phosphatidic acid (PA) abundances were not altered, indicating a redistribution between lipid classes to preserve PA content. In FSG67-treated MKs, we observed slightly, but not significantly, decreased levels of DG and an increase in PC. Of note, TGs increased 3-fold compared with day-7 control, likely owing to the production of lipid droplets to compensate for high levels of acyl-CoA within the cell. Whereas phospholipid biosynthesis is hampered when inhibitors are used, the production of SM and ST appears to be enhanced. This could be owing either to an excess of serine and palmitoyl-CoA that cannot be incorporated via normal lipogenesis, or to SM and ST acting as functional substituents of other stabilizing membrane lipids. To elucidate the role of phospholipids in MK maturation, we first monitored control and inhibitor-treated MKs and visually examined their ability to form proplatelets. MKs were taken after 3 days of differentiation, and proplatelet formation was observed for 30 h. The number of proplatelet-forming MKs increased by only 9% for FSG67-treated and triacsin C-treated MKs, instead of 20% as observed for control MKs (Fig. 5c,d). Further, polyploidization of inhibitor-treated MKs was markedly impaired as reflected by a significant reduction of polyploidy and a significantly higher percentage of MKs with DNA content of <8 N (Fig. 5e). Both inhibitors resulted in an overall greatly reduced number of proplatelet-forming cells compared with the control, indicating that proper MK development and proplatelet formation requires both de novo lipid synthesis and uptake of exogenous FAs from dietary lipids.
These results could also be translated into in vivo conditions. Treatment of mice with either inhibitor over a period of 7 days resulted in a significant reduction in the MK sinusoid contact, with a significantly higher rate of MK fragmentation within the bone marrow of murine femora and reduced polyploidy (Fig. 5f,g and Extended Data Fig. 9b). These marked effects on thrombopoiesis resulted in significant thrombocytopenia in inhibitor-treated mice compared with mice treated with solvent control (Fig. 5h,i). Notably, in vivo visualization of MKs in the bone marrow of the mouse skull using two-photon intravital microscopy (2P-IVM) unraveled a profound reduction of the ratio of proplatelet-forming MKs in bone marrow of mice treated with FSG67 (6.6 ± 3.7%) or triacsin C (10.6 ± 3.1%) compared with vehicle-treated mice (14.4 ± 5.5%), indicating that both treatments impair proplatelet formation in vivo (Fig. 5h and Extended Data Fig. 9c) and consequently result in thrombocytopenia (Fig. 5i) Moreover, 2P-IVM revealed an increased ratio of MKs with altered morphology and premature ectopic fragmentation (triacsin C, 6.5 ± 2.1%; FSG67, 8.6 ± 3.0%; vehicle, 4.5 ± 2.4%), potentially resulting in an inefficient proplatelet release into the vascular sinusoids.
Anionic membrane regulates platelet biogenesis via CKIP-1/CK2α axis
Finally, we investigated how class rearrangement to a more anionic membrane, through increase of PG, PI and PS, can increase the anionic lipid strength in MKs from 13 mol% to 20 mol% (0.2–1.7 mol%, 7.5–10.8 mol% and 7–8.8 mol%, respectively). Thereby, increased signaling capability could potentially modulate megakaryopoiesis. Therefore, we reanalyzed our proteomics data with an interaction analysis and identified a network of 67 strongly regulated kinases, as well as additional 162 regulated proteins containing at least one of the following lipid-binding domains: pleckstrin homology (PH) or PH-like domains34; C1 and C2 (refs. 35,36); four-point-one, ezrin, radixin, moesin (FERM) domains37; or Src homology domains SH2 or SH3, which were recently identified to also show lipid-binding capabilities38. All have a strong link to anionic lipids37,39,40,41,42. Among those proteins, we identified several kinases that themselves harbor lipid-binding sites such as Janus kinases 1 and 2 (JAK1/JAK2), integrin-linked protein kinase (ILK), Bruton’s tyrosine kinase (BTK), protein kinase C-α (PKC-α) and several Src family kinases such as SRC, FYN and LYN (Extended Data Fig. 10a–c). All eight are linked to megakaryocytic development and platelet activation43,44,45 and interact with or are directly or indirectly activated by lipids46,47,48,49.
Given that anionic phospholipids linked to membrane binding exhibit continuous upregulation, we speculate whether the interplay of lipids, lipid-binding proteins and kinases may act as a potential modulatory axis driving megakaryopoiesis and proplatelet formation. Using the Molecular Complex Detection (MCODE)50 clustering algorithm within the open-source software Cytoscape51, we were able to identify six clusters (cutoff score, 2.0) of densely connected regions in the protein interaction network (Fig. 6a). Whereas clusters 1, 2, 4 and 5 were closely clustered together, clusters 3 and 6 showed clear separation. Interestingly, cluster 6 was formed by the casein kinase 2 catalytic subunits (CK2α/CK2α′), the CK2 regulatory β-subunit (CK2β), and the adapter protein PH domain-containing family O member 1/casein kinase interacting protein-1 (PKHO1/CKIP-1).
In this context, we identified a robust upregulation of the CKIP-1/CK2 cluster during megakaryopoiesis on day 7 (Fig. 6b). CKIP-1 reflects an adapter protein with a PH domain facilitating recruitment of the CK2α isoform to the plasma membrane via direct interaction resulting in nonenzymatic regulation of CK2α activity52,53,54. CKIP-1 contains a PH domain at the amino terminus and five proline-rich motifs throughout the protein, which mediate multiple cellular protein interactions55. Immunoblotting revealed a strong upregulation of the membrane localization of CKIP-1 in MKs and its coexpressed target CK2α at day 7 (Fig. 6c), an effect that was abolished in MKs treated with ACSL inhibitor triacsin C or the GPAT inhibitor FSG67, respectively. These observations suggest a regulation of the CKIP-1/CK2α interplay at the MK plasma membrane by the ACSL/GPAT lipid metabolic axis during megakaryopoiesis. To elucidate the functional role of the recruited catalytic CK2α subunit for the process of thrombopoiesis, we examined MK localization and morphology in immunostained bone marrow cryosections of intact murine femora from mice with an MK-specific or platelet-specific genetic deletion of CK2α (csnk2α1). The visualization of MK distribution within the entire femora confirmed that MKs in the femora of csnk2α1Pf4∆/Pf4∆ mice displayed less direct sinusoidal contact and conversely an accumulation of MKs in the bone marrow hematopoietic compartment with markedly increased MK fragmentation (Fig. 6d and Extended Data Fig. 10d), pointing to insufficient transendothelial platelet biogenesis. Additionally, investigation of MKs flushed out of bone marrow revealed a significantly reduced ploidy in csnk2α1Pf4∆/Pf4∆ MKs, with a significant reduction of ≥16 N-containing MKs (Fig. 6e), indicating that csnk2α1 deficiency results in the accumulation of immature MKs. To study the effect of genetic deletion of CK2α in MK-dependent thrombopoiesis, we performed in vitro proplatelet formation assays using MKs derived from the bone marrow of csnk2α1Pf4∆/Pf4∆ mice and csnk2α1lox/lox littermates. Accordingly, significantly fewer numbers of csnk2α1Pf4∆/Pf4∆ MKs formed proplatelets (Fig. 6f). Thus, abolished MK maturation and proplatelet formation again contribute to the development of significant macrothrombocytopenia in csnk2α1Pf4∆/Pf4∆ mice when compared with csnk2α1lox/lox mice (Fig. 6g). Altogether, these observations let us hypothesize that the lipid-driven CKIP-1/CK2α axis in MKs is crucial for MK maturation and proplatelet formation (Fig. 6h).
Megakaryopoiesis is a complex process by which hematopoietic stem cells differentiate into MKs, which are eventually capable of releasing platelets into the bloodstream through a process called thrombopoiesis. It is characterized by a progressive increase of cellular dimensions, DNA content with subsequent polyploidization, and, finally, proplatelet formation into the bone marrow sinusoids.
Recently, a few studies tried to shed light on how lipid metabolism can affect megakaryopoiesis and proplatelet formation by mainly investigating enzymes derived from SP metabolism7,9,32. Nevertheless, these studies did not elucidate the chemical nature of the involved lipid species to describe the observed functional effects leading to pathologies such as thrombocytopenia. Therefore, it is still unclear if lipids are the actual cause of the functional effect. To investigate the exact mechanisms of lipidome regulation during megakaryopoiesis, lipids must be analyzed in detail simultaneously with their metabolizing proteins. Several aspects should be considered, for example, time and sensitivity of cell isolation, using detergent-free conditions, and reporting of concentrations to understand the dimensions of membrane rearrangement under the given circumstances. Here, we have used the full technological advancement of MS-based lipidomics to report a quantitative lipidomics map of MK differentiation using a lipid-centered multiomics approach18. On the one hand, we quantified 473 lipid species covering a concentration range of over six orders of magnitude. On the other hand, the expression levels of around 4,651 proteins were determined. Using one sample to cover multiple molecule classes reduces the analytical error and enhances the correlation between different molecules17. Quantitatively, the MK lipidome seems twice as complex as the one derived from platelets12,56. Additionally, PUFAs are less enriched in the MK membrane compared with platelets. Nevertheless, both lipidomes are comparable, which is reflected in the abundance of different lipid classes. The higher complexity of the MK lipidome is most likely based on the presence of more organelles and an advanced lipid metabolism.
Using a multiomics approach to dissect megakaryopoiesis, this study revealed three major findings. First, lipid uptake is highly increased during MK maturation, which is reflected by the increased expression of FA receptors (CD36 and FATP1) and transporters (FABP4/5). Second, FA synthesis and oxidation are elevated in differentiating MKs, indicated by the upregulation of FA synthetase (FASN) and different mitochondrial FA transporters needed for β-oxidation, such as CPT2 and CACP (Extended Data Fig. 1b). Third, a significant remodeling of complex lipid synthesis pathways such as SP, GP and ST occurs, which can be observed at the enzyme and lipid levels. Interestingly, increasing TG levels indicate an elevated lipid droplet formation, most likely needed for β-oxidation. Most lipidome changes occur between days 1 and 3, demonstrating that membrane remodeling is an early process during megakaryopoiesis. However, the most striking finding was the upregulation of anionic membrane lipids, which increased by >7 mol% during differentiation. Of note, this elevation in anionic lipid mass correlated well with the upregulation of DG. As a result of pharmacological inhibition of FA uptake or GP de novo synthesis, no upregulation of anionic lipids in maturating MKs was observed. Moreover, the inhibition of either ACSL or GPAT resulted in impaired MK polyploidization and perturbated thrombopoiesis reflected by a 50% reduction of proplatelet formation and release into the bone marrow sinusoids resulting in a significant thrombocytopenia.
Using relative quantitative proteomics, we uncovered a broad spectrum of proteins whose expression was significantly shifted during the early and late stages of megakaryopoiesis. A recent study compared the proteome and transcriptome of round versus proplatelet-producing MKs by two-dimensional (2D) electrophoresis and polysome profiling to uncover protein changes during megakaryopoiesis57. Using the latest MS technology, we analyzed the proteome at several time points of megakaryopoiesis and proplatelet formation. This enabled us to assess abundances of over 4,400 proteins, compared with 200 proteins in the previously mentioned study. Of the 30 proteins previously identified to be regulated, most also displayed regulation in our study. Due to the increased sensitivity of our approach, more than 3,000 proteins were found to be regulated. Remarkably, we unraveled several proteins and kinases that could potentially be regulated by (anionic) lipids and are significantly regulated during megakaryopoiesis. A signaling pathway that was significantly upregulated in MKs during MK maturation was the CKIP-1/CK2 cluster. The regulatory β-subunit of CK2 has been reported as major regulator of MK maturation and thrombopoiesis58. CKIP-1 is a nonenzymatic and specific regulator of the catalytic CK2α isoform activity52. CKIP-1 binds to the plasma membrane via its PH domain by specific binding of anionic lipids such as PS, PI and PI’s phosphorylated forms53,54. Furthermore, CKIP-1 controls the access of CK2α to specific cellular targets through its ability to selectively recruit CK2α and not CK2α′ to the plasma membrane, again in a PH domain-dependent manner52. Accordingly, we unraveled CKIP-1/CK2α as a potential effector of the lipidome remodeling downstream of ACSL and GPAT. The anionic shift of the MK lipidome during MK maturation culminates in an increase of phospholipids that are able to bind to the PH domain of CKIP-1 with consecutive recruitment of CKIP-1 and CK2α to the plasma membrane of MKs. It has been reported that CKIP-1 is crucially involved in MK differentiation and thrombopoiesis, and a genetic deletion of CKIP-1 (plekho1/ckip1) resulted in defective megakaryopoiesis with reduced MK ploidy and reduced platelet production with significant thrombocytopenia59. Similarly, after MK-specific or platelet-specific genetic deletion of CK2α (csnk2α1), we found significantly reduced MK ploidy and abrogated proplatelet formation with the development of significant macrothrombocytopenia. Remarkably, csnk2α1Pf4∆/Pf4∆ mice displayed a highly comparable pattern of MK distribution within the bone marrow, the same MK characteristics with premature fragmentation, and reduced proplatelet formation with subsequent thrombocytopenia as mice upon treatment with the ACSL or GPAT inhibitor (Fig. 7).
These identified mechanisms in MK maturation and thrombopoiesis are of potential interest to deepen our understanding of how alterations in lipid metabolism in diseases such as obesity or metabolic syndrome, both associated with thrombotic cardiovascular complications, might interfere with platelet production.
In this study, we aimed to establish an MK-specific multiomics workflow to comprehensively analyze MK lipid metabolism. We demonstrated that the MK lipidome remodeling during MK maturation and proplatelet formation involves ACSL-dependent or GPAT-dependent lipid metabolism. As a result from a shift toward anionic membrane properties during MK maturation, the altered MK lipidome may promote specific signaling complexes and kinases, such as CKIP-1/CK2α, that are critically involved in thrombopoiesis. However, further analyses linking anionic membrane remodeling to kinase changes in proplatelet formation are ultimately warranted to tackle the question of how lipids control platelet production and properties.
Materials and standards
Rabbit anti-α-tubulin (Thermo Fisher Scientific, PA5-19489), rat anti-CD42b monoclonal (clone Xia.G7, Emfret, M042-1), rat FITC anti-mouse CD41 (BioLegend, 133904), rabbit anti-CSNK2A1 (Abcam, ab76040), mouse anti-CKIP-1 (Santa Cruz Biotechnology, sc-376355), mouse anti-GAPDH (Thermo Fisher Scientific, MA5-15738), goat anti-rabbit secondary antibody (Life Technologies, A21069), goat anti-rat secondary antibody (Life Technologies, A11006), donkey anti-mouse secondary antibody (LI-COR, 926-32212), Alexa Fluor 594-conjugated anti-CD105 antibody (BioLegend, 120418), Alexa Fluor 546-conjugated anti-CD105 (self-generated, clone MJ7/18, ref. 60), anti-CD42a (GPIX) Alexa Fluor 488 derivative (self-generated, p0p6, ref. 61).
DRAQ5 DNA stain (Thermo Fisher Scientific, 62251), Alexa Fluor 488 phalloidin (Life Technologies, A12379), antibody diluent (Zytomed, ZUC025-100), Roti-Load (Roth, K929.1), bovine serum albumin (BSA) (PanReac AppliChem, A1391,0500), triacsin C (Cayman Chemical, 10007448), FSG67 (Focus Biomolecules, 10-4577), mounting medium (Invitrogen, P36961), poly-L-lysine (Sigma-Aldrich, P8920-100ML, 0.1%), paraformaldehyde (PFA) (Otto Fischar GmbH & Co. KG, 27246), Cell Lysis Buffer (Cell Signaling Technology, 9803S), Protease/Phosphatase Inhibitor Cocktail (Cell Signaling Technology, 5872S), FcR Blocking Reagent mouse (Miltenyi Biotec, 130-092-575), PureLink RNase A (Invitrogen, 12091-021), propidium iodide (Invitrogen, P1304MP), EZ-Link Sulfo-NHS-Biotin (SNB) (Thermo Fisher Scientific, 11851185), EZ-Link NHS-Biotin (NB) (Thermo Fisher Scientific, 10381394), L-lysine (Sigma-Aldrich, L5501), triethylamine (Sigma-Aldrich, 90335), medetomidine (Pfizer), midazolam (Roche), fentanyl (Janssen-Cilag), recombinant TPO (ImmunoTools, 12343615).
Chemicals specific for lipid analysis: formic acid (BioSolve, 6914143), ULC/MS-grade methanol (BioSolve, 13684102), ULC/MS-grade water (BioSolve, 23214102), ULC/MS-grade acetonitrile (ACN) (BioSolve, 1204102), methyl tert-butyl ether (MTBE) (VWR, 34875-1L), ammonium acetate (Merck, 73594-25G-F), ammonium formate (Sigma-Aldrich, 70221-25G-F), HPLC-grade phosphoric acid (Sigma-Aldrich, 79617-250ML, 85–90%), chloroform (Sigma-Aldrich, 650498-1L), isopropanol (IPA) (Sigma-Aldrich, 1010402500).
Chemicals specific for protein analysis: formic acid (VWR, 84865-180P), ULC/MS-grade methanol (VWR, 83638320), ULC/MS-grade water (Honeywell, 14263-2L), ULC/MS-grade ACN (Honeywell, 34967-2.5L), urea (Merck, 1084871000), triethylammonium bicarbonate (TEAB) (Sigma-Aldrich, 18597-100ML), sodium dodecyl sulfate (SDS) (GERBU, 1212), dithiotreitol (DTT) (Sigma-Aldrich, APOSBIMB1015-25G), iodoacetamide (IAA) (Sigma-Aldrich, I6125-25G), Trypsin/Lys-C Mix Mass Spec Grade (Promega, V5073), trifluoroacetic acid (TFA) (Sigma-Aldrich, T6508-100ML).
Standard peptide [Glu1]-Fribrinopeptide B (sequence EGVNDNEEGFFSAR, Sigma-Aldrich, F3261), standard peptide M48 (sequence TTPAVLDSDGSYFLYSK, PSL), standard peptide HK0 (sequence VLETKSLYVR, PSL), standard peptide HK1 (sequence VLETK(ε-AC)SLYVR, PSL).
Mouse SPLASH LIPIDOMIX Mass Spec Standard (Avanti Polar Lipids, 330710X-1EA) consisting of PC 15:0-18:1(d7), PE 15:0-18:1(d7), PS 15:0-18:1(d7), PG 15:0-18:1(d7) (as internal standard for PG and CL), PI 15:0-18:1(d7), PA 15:0-18:1(d7), LPC 18:1(d7), LPE 18:1(d7) (as internal standard for all lysophospholipids except LPC), SE 18:1(d7), PC-ether (PCO-a) 18:0-18:1(d9), PE-ether (PEO-a) 18:0-18:1(d9), DG 15:0-18:1(d7), TG 15:0-18:1(d7)-15:0 and SM d18:1-18:1(d9); Ceramide/Sphingoid Internal Standard Mixture II (Avanti Polar Lipids LM6005-1EA) consisting of sphingosine (SPB) d17:1, sphinganine (SPB) d17:0, sphingosine-1-P (SPBP) d17:1, sphinganine-1-P (SPBP) d17:0, SM d18:1/12:0, Cer d18:1/12:0, glucosylceramide (GlcCer) d18:1/c12:0 (as internal standard for HexCer), lactosylceramide (LacCer) d18:1/12:0 (as internal standard for dihexosylceramide (Hex2Cer)) and ceramide-1-P (CerP) d18:1/12:0; cholesterol-d7 (Avanti Polar Lipids, 700041P); lysosphingomyelin (LSM)-d7 (Avanti Polar Lipids, 860639P); PS 14:0-14:0 (Avanti Polar Lipids, 840033P) (self-generated biotinylated internal standard for biotin-PS).
Csnk2a1lox/lox mice were generated elsewhere62. For MK-specific or platelet-specific deletion of CK2α, csnk2a1lox/lox mice were crossed with Pf4-Cre transgenic mice (The Jackson Laboratory, 008535) and studied at the age of 12–14 weeks. All animal experiments were performed according to Directive 2010/63/EU of the European Parliament on the protection of animals used for scientific purposes and were approved by local authorities (Regierungspräsidium Tübingen) following the ARRIVE guidelines (protocols M01/20G and M03/19M).
For in vivo treatment studies, 6-week-old C57BL6/J mice, were treated daily intraperitoneally with either 0.285 mg per kg (body weight) triacsin C, 5 mg per kg (body weight) FSG67 or dimethylsulfoxide (DMSO) over a period of 7 days. Concentrations were adapted according to refs. 63,64.
Bone marrow isolation and MK differentiation
For the bone marrow isolation, a centrifugation protocol previously published by ref. 65 was used. Briefly, 10–14-week-old, male C57BL/6J mice (The Jackson Laboratory) were anesthetized using isoflurane and killed by cervical dislocation following the institutional guidelines and the German law for the welfare of animals. Both femora were dissected and cleaned, cut open at the knee side, and placed with the cut side facing down in a 0.5-ml Eppendorf tube with a pre-pierced hole in the bottom. The tube was placed into a 1.5-ml tube, pre-filled with 100 µl of DMEM (supplemented with 1% penicillin/streptomycin and 10% FBS) and centrifuged for 1 min at 2,600 × g at room temperature (21°C (69.8°F)). Next, 1 ml of supplemented medium was added, and bone marrow cells were resuspended, then filtered through a pluriStrainer Mini (70 µm), and the strainer was rinsed with 1 ml of medium. Afterward, cells were centrifuged for 5 min at 300 × g at room temperature, and the supernatant was removed.
To induce MK differentiation, the freshly isolated bone marrow cells (pool of five individual animals) were cultivated in 10-cm cell culture dishes containing supplemented DMEM, and differentiation was initiated by adding (1%) recombinant TPO. Cells were cultivated at 37 °C, 5% CO2 for different periods of time. On days 1, 3 and 7, cells were collected (1,000 r.p.m., 5 min) and resuspended in 950 µl of PBS. The cell suspension was carefully pipetted on a two-phase BSA gradient (bottom, 1.5 ml 3% BSA in PBS; top, 1.5 ml 1.5% BSA in PBS) to separate cells by weight. After 40 min, the supernatant was removed, and the cell pellet was washed three times with 500 µl of PBS. Cells were counted in a Neubauer chamber and adjusted to 200,000 cells per tube. Cell pellets were shock-frozen in liquid nitrogen and stored at −80 °C for later multiomics analysis.
For immunofluorescence microscopy of MKs, cells were isolated and purified as described above and cultivated for 1, 3 and 7 days. After isolation via BSA gradient, 5,000 MKs were seeded on chamber slides pre-coated with 0.1% poly-L-lysine for 60 min at 37 °C and further incubated for 1 h. Cells were fixed for 15 min with 4% PFA at room temperature, washed three times for 3 min each with PBS, 10 min with PBS and 0.1% Triton X-100, and again three times for 3 min each with PBS. Cells were further incubated with 1% BSA in PBS to block the unspecific binding of antibodies. Cells were stained with either the primary antibodies Alexa Fluor 488 phalloidin (1:200 in antibody diluent) and α-tubulin (1:400 in antibody diluent) or CD42b (1:100 in antibody diluent), or CD42d (1:300 in antibody diluent). After overnight incubation at 4 °C and washing three times for 3 min each with PBS, secondary antibodies (anti-rabbit Alexa Fluor 568, 1:300 in PBS; anti-rat Alexa Fluor 488, 1:300 in PBS) were applied for 2 h at room temperature. After three washes for 3 min each with PBS, nuclei were stained for 15 min with DRAQ5 stain (1:1,000), washed again with PBS, and mounted using a mounting medium. An LSM510 confocal laser scanning microscope (Zeiss) and ZEN Blue software (Zeiss) were used for the analysis.
Protein and lipid extraction
Samples, consisting of 200,000 cells per tube, were used for lipid and protein extraction following the SIMPLEX protocol previously described by ref. 18. In brief, 225 µl of methanol and the internal standard mixture were added to all samples, and cell pellets were homogenized through 2–5 min of ultrasonication. Two blanks used as quality controls were processed in parallel, one with and the other without internal standards. Next, 750 µl of MTBE were added, and samples were incubated for 1 h at 950 r.p.m. at 4 °C. To induce phase separation, 188 µl of water (HPLC-grade) were added, and samples were incubated on ice for 5 min. After a 10-min centrifugation step at 10,000 × g at 4 °C, the upper organic phase (containing GPs, GLs, SPs and STs) was carefully removed and dried under a gentle nitrogen flow. The dried organic phase was reconstituted in 100 µl of IPA/methanol/CHCl3 (4:2:1, v/v/v) containing 7.5 mM ammonium acetate for lipid analysis. To complete protein precipitation, 527 µl of methanol were added to the lower aqueous phase, and samples were stored for 2 h at −20 °C, following centrifugation for 30 min at 12,000 × g at 4 °C. The protein pellet was dried and further subjected to protein analysis.
Proteomics sample preparation
Protein samples were diluted 1:2 in lysis buffer (8 M urea, 50 mM TEAB, 5% SDS), then heated at 90 °C for 5 min, and protein concentrations were determined using a Pierce BCA Protein Assay Kit (Thermo Scientific). For enzymatic digestion, 20 µg of protein were used, and ProtiFi S-Trap technology was applied66. In short, solubilized proteins were reduced and carbamidomethylated by adding 64 mM DTT and 48 mM IAA, respectively. Before loading the samples onto S-Trap mini cartridges (ProtiFi), trapping buffer (90% (v/v) methanol, 0.1 M TEAB) was added. Subsequently, samples were thoroughly washed and then digested using Trypsin/Lys-C Mix for 2 h at 37 °C. Finally, peptides were eluted, dried, and stored at −20 °C until LC–MS analysis.
LC–MS/MS analysis was performed as described previously67,68,69. In brief, reconstitution of dried peptide samples was achieved by adding 5 µl of 30% formic acid containing four synthetic standard peptides. Afterward, samples were diluted with 40 µl of loading solvent (97.9% H2O, 2% ACN, 0.05% TFA), of which 5 µl were injected into the Dionex UltiMate 3000 nano high-performance liquid chromatography (HPLC) system (Thermo Fisher Scientific). A pre-column (2 cm × 75 µm, PepMap 100 C18, Thermo Fisher Scientific) run at a flow rate of 10 µl min−1 using mobile phase A (99.9% H2O, 0.1% formic acid) was used to pre-concentrate peptides before chromatographic separation. Peptides were then separated on an analytical column (25 cm × 75 µm, 25 cm, Aurora Series emitter column, IonOpticks) by applying a flow rate of 300 nl min−1 and using a gradient of 8–40% mobile phase B (79.9% ACN, 20% H2O, 0.1% formic acid) over 155 min, resulting in a total LC run time of 195 min including washing and equilibration steps. A timsTOF Pro mass spectrometer (Bruker) with a captive spray ion source run at 1,700 V was used for MS analysis. The timsTOF Pro was operated in parallel accumulation–serial fragmentation (PASEF) mode, and moderate MS data reduction was applied. Further parameters included a scan range (m/z) from 100 to 1,700 to record MS and MS/MS spectra and a 1/k0 scan range of 0.60–1.60 V.s/cm2 resulting in a ramp time of 100 ms to achieve trapped ion mobility separation. All experiments were performed with ten PASEF MS/MS scans per cycle, leading to a total cycle time of 1.16 s. Furthermore, the collision energy was ramped as a function of increasing ion mobility from 20 eV to 59 eV, and the quadrupole isolation width was set to 2 Th for m/z < 700 and 3 Th for m/z > 700. All samples were analyzed in technical duplicates.
Label-free proteomics data analysis
The publicly available software package MaxQuant (v184.108.40.206) running the Andromeda search engine was used for protein identification and label-free quantification (LFQ)70. Therefore, raw data were searched against the Swiss-Prot database Mus musculus (v220621 with 17,519 entries). Search parameters included an allowed peptide tolerance of 20 ppm, a maximum of two missed cleavages, carbamidomethylation on cysteines as fixed modification, methionine oxidation, and N-terminal protein acetylation as variable modification. A minimum of two peptides per protein, of which at least one has to be unique for each protein, was set as a search criterium for positive identifications. In addition, the ‘match between runs’ option was applied, using a 0.7-min match time window, a match ion mobility window of 0.05, a 20-min alignment time window, and an alignment ion mobility of 1. A false discovery rate (FDR) of ≤0.01 was set for all peptide and protein identification. LC–MS data evaluation and statistical analysis were accomplished using the Perseus software (v220.127.116.11)71. Identified proteins were first filtered for reversed sequences and common contaminants and annotated according to differentiation time points. Before statistical analysis, LFQ intensity values were log2-transformed, means of technical duplicates were calculated, and proteins were additionally filtered for their number of independent identifications (a minimum of three identifications in at least one group). Two-sided t-tests and statistics for volcano plots were performed by applying an FDR cutoff of 0.05 and an S0 of 0.1, whereby S0 controls the relative importance of the t-test P value and difference between the means. Figure visualization was done using OriginPro (v2021), RStudio (v1.4.1106) and Instant Clue (v0.10.10)72.
Protein network and cluster analysis
For the generation of protein networks, we divided our proteomics data into two groups: kinases and lipid-binding proteins. All proteins that were significantly regulated on either day were used for network analysis using STRING. Networks generated were loaded in Cytoscape (v3.9.1). For protein clustering, the MCODE application inside the Cytoscape interface was used with the following conditions: network scoring including loops with a degree cutoff of 2, cluster finding with fluffing using a node density cutoff of 0.8 and node score cutoff of 0.24 with K-core of 2 and max depth of 100.
A Q Exactive HF (Thermo Fisher Scientific) coupled to a TriVersa NanoMate ion source (Advion Biosciences) was used for direct infusion experiments. A total of 12 µl of the sample were delivered over 14 min with a backpressure of 0.95 psi. After 6 min, polarity switching from +1.25 kV to −1.25 kV was applied to acquire mass spectra in both positive and negative ion modes in one measurement. Full MS spectra covering the mass range of 350–1200 m/z in both positive and negative modes were acquired with a resolution of 240,000, an AGC target of 1e6, and a maximum IT of 105 ms. MS1 acquisition was followed by data-independent acquisition of precursor masses at an interval of 1,001 Da. The precursor isolation window was 1 Da, and normalized collision energy (nCE) was 21% and 26% for positive and negative modes, respectively. MS2 spectra were acquired with a resolution of 60,000, an AGC target of 1e5, and a maximum IT of 105 ms.
Analysis of SP and ST was performed as previously described by ref. 73. Inclusion lists for targeted measurements were generated using LipidCreator (v1.2.0). For the reversed-phase LC, the UltiMate 3000 system was equipped with an Ascentis Express C18 main column (150 mm × 2.1 mm, 2.7 µm, Supelco) and fitted with a guard cartridge (50 mm × 2.1 mm, 2.7 µm, Supelco) in a column oven set to 60 °C. Solvent A was ACN/H2O (3:2, v/v), solvent B was IPA/ACN (9:1, v/v), and both contained 0.1% formic acid, 10 mM ammonium formate and 5 µM phosphoric acid. The separation was carried out with a flow rate of 0.5 ml min−1 with the following 25-min-long gradient: initial, 30% B; 0–2 min, hold 30% B; 2–3 min, 30–56.1% B; 3–4 min, 56.1–58.3% B; 4–5.5 min, 58.3–60.2% B; 5.5–7 min, 60.2–60.6% B; 7–8.5 min, 60.6–62.3% B; 8.5–10 min, 62.3–64% B; 10–11.5 min, 64–64.5% B; 11.5–13 min, 64.5–66.2% B; 13–14.5 min, 66.2–66.9% B; 14.5–15 min, 66.9–100% B; 15.0–19.0 min, hold 100% B; 19 min, 5% B; 19–22 min, hold 5% B; 22 min, 30% B; 22–25 min, hold 30% B. The injector needle was automatically washed with 30% B and 0.1% phosphoric acid, and a volume of 5 µl per sample was injected.
The LC was coupled to a QTRAP 6500+ (Applied Biosystems Sciex) with an electrospray ion source (Turbo V ion source). MS scans were acquired in positive ion mode with the following source settings: curtain gas, 30 arbitrary units; temperature, 250 °C; ion source gas I, 40 arbitrary units; ion source gas II, 65 arbitrary units; collision gas, medium; ion spray voltage, +5,500 V; declustering potential, +100 V/−100 V; entrance potential, +10 V; exit potential, +13 V. For the scheduled multiple reaction monitoring, Q1 and Q3 were set to unit resolution. The detection window was set to 2 min, and the cycle time was set to 0.5 s. Data were acquired with Analyst (v1.7.2, Applied Biosystems Sciex).
Lipid identification and quantification
All spectra from shotgun experiments were converted to centroid mode using msConvert (v3.0.20186-dd907d757) and analyzed using LipidXplorer (v18.104.22.168) under the following settings: MS1, mass tolerance of 5 ppm with an intensity threshold of 1e5; MS2, mass tolerance of 10 ppm with an intensity threshold of 5e4. For lipid identification, molecular fragmentation query language queries, based on the previous work from refs. 74,75, were compiled to match precursor and fragment ions to recognize lipid species. The detection and quantification of GLs (DG and TG) were used in positive ion mode. GPs (cardiolipin (CL), LPA, LPI, LPG, LPC, LPE, PA, PG, PC, PCO, PE, PEO, PI and PS) were identified and quantified in negative ion mode. All signal intensities were normalized to the corresponding deuterated internal standard (Mouse SPLASH LIPIDOMIX Mass Spec Standard). Protein concentrations, determined by the Pierce BCA Protein Assay Kit, were used to quantify all lipid species. TGs and CLs were quantified based on precursor intensities (Supplementary Table 1).
SPs (Cer, HexCer, Hex2Cer, Sulfo-HexCer (SHexCer), LSM, SPBP, SPB and SM) and STs (ST and SE) were identified and quantified by LC–MS analysis (Supplementary Table 2). Integration of peaks from targeted measurements was performed using Skyline (v22.214.171.124). Lipid species abundance was calculated using peak areas and quantified to the respective internal standard (Ceramide/Sphingoid Internal Standard Mixture II; cholesterol-d7) and protein amount described above.
Generation of biotinylated PS standards
A biotinylated PS internal standard was generated for quantification of biotin-labeled PS species within the biological sample. The generation of biotinylated standards was performed according to the protocol of ref. 76. In brief, 1 mg of PS 14:0_14:0 standard (Avanti Polar Lipids) was dissolved in 330 µl of chloroform/methanol (2:1, v/v), and 6 mg of NB were added. After vortexing, 3.3 µl of triethylamine (Sigma) were added and incubated for 30 min at room temperature. Excess NB was sedimented by centrifugation for 5 min at 500 × g at room temperature. The supernatant was transferred into a new glass vial. The sediment was washed once with 330 µl of 2:1 CHCl3/methanol, vortexed and centrifuged, and the supernatant was combined from the previous step. After drying under nitrogen flow, the biotinylated standard was resuspended in 600 µl of methanol for HPLC purification. An Agilent 1200 Series LC system with a Discovery C18 column (250 mm × 4.6 mm, 5 µm) was used with the following conditions: temperature, 22 °C; flow rate, 1 ml min−1; gradient elution profile, 50% B (A, water + 5 mM ammonium acetate; B, methanol + 5 mM ammonium acetate) to 100% B over 15 min, held at 100% B for 20 min, then re-equilibrated to 50% B. Ultraviolet absorbance was measured at 205 nm. Six times, 100 µl were injected and all fractions were manually collected, combined and dried using a Genevac and resuspended in 200 µl of methanol. The standard was transferred into a clean, pre-weighted glass vial, dried and weighted again. The standard concentration was adjusted to 100 ng µl−1 in methanol and stored under nitrogen gas at −80 °C. The derivatization of the standard was validated by direct injection on an amaZon speed ion ETD trap instrument.
Surface labeling of externalized PS
Biotinylation of cell surface-exposed PS was performed based on the protocol from ref. 76. A cell-impermeable reagent (SNB) was used to label PS on the outer leaflet, and a cell-permeable reagent (NB) was used for quantification of total PS content. In brief, MK cell suspensions containing 200,000 cells per sample were equally divided in two tubes (100,000 cells each) and treated with either 43 µl of 22 mM SNB in PBS or 20 µl of 20 mM NB in DMSO for 10 min at room temperature. To SNB-treated cells, 72 µl of 250 mM L-lysine in PBS were added and incubated for another 10 min at room temperature to quench excess SNB. To NB samples, 95 µl of LC-grade water were added to reach the final extraction volume of 315 µl. Samples were transferred into 5-ml polypropylene Eppendorf tubes and subjected to the SIMPLEX protocol as described in the section ‘Protein and lipid extraction’ using a tripled amount of all solvents. For normalization of lipid intensities, 10 µl of a self-generated biotinylated PS standard (biotin-PS 14:0_14:0, 10 ng µl−1) and 5 µl of Mouse SPLASH mix (Avanti Polar Lipids) were added prior to extraction. Dried lipid extracts were resuspended in 25 µl of butanol solvent (1-butanol:IPA:H2O, 8:23:69, v/v/v + 5 mM phosphoric acid), and lipids were identified using targeted LC–MS/MS.
Reversed-phase LC–MS/MS of biotinylated PS
Lipid extracts were separated by reversed-phase HPLC according to ref. 77 with minor adaptions. For separation, an Ascentis Express C18 column (150 mm × 2.1 mm, 2.7 μm, Supelco) fitted with a guard cartridge (50 mm × 2.1 mm, 2.7 μm, Supelco) was used. Mobile phase A was ACN/H2O (60:40, v/v), mobile phase B was IPA/ACN (90/10, v/v), and both contained 10 mM ammonium formate and 0.1% formic acid. The temperatures of the autosampler and the column oven were set to 10 °C and 60 °C, respectively. Separation was carried out with a flow rate of 0.5 ml min−1 with the following 35-min-long gradient: initial, 30% B; 0.0–3.0 min, hold 30% B; 3.0–15.0 min, ramp to 75% B; 15.0–17.0 min, ramp to 100% B; 17.0–30.0 min, to 5% B; 30.1–35.0 min, to 30% B. The injector needle was automatically washed with 30% B, and a volume of 5 µl per sample were injected.
The LC was coupled to the Q Exactive HF instrument, and data were acquired in negative ion mode. The following electrospray ionization (ESI) source parameters were applied: spray voltage, 3.8 kV; capillary temperature, 270 °C; sheath gas flow rate, 50; auxiliary gas flow rate, 15; auxiliary gas heater temperature, 380 °C; S-lens RF level, 60. Full MS spectra from 500 to 1,200 m/z were acquired in negative mode with a resolution of 60,000, an AGC target of 106, and a maximum IT of 50 ms. For MS/MS, a resolution of 30,000, an AGC target of 105, a maximum IT of 115 ms and an nCE of 24 were applied.
Identification and quantification of biotinylated PS
Integration of peaks from targeted measurements was performed using Skyline (v126.96.36.199). The top two abundant PS species (PS 18:0_18:1 and PS 18:0_20:4) were monitored. For the identification of biotinylated PS, both FAs and the neutral loss of the biotinylated PS headgroup (m/z, 313) were used. Lipids were quantified on the MS1 level. Lipid species abundance was calculated using peak areas and quantified to the respective internal standard (biotinylated PS to B-PS 14:0_14:0, unlabeled PS to PS 15:0_18:1(d7)). To account for differences in PS total amount throughout the samples, the summed intensity of labeled and unlabeled PS within each sample were used to normalize the amount of labeled PS. A ratio was calculated of externalized:total PS. Day 1 was set to 1 and used as a reference to calculate relative changes during megakaryopoiesis.
Validation of shotgun lipidomics data by targeted LC–MS/MS
Trends of selected lipid species shown in Fig. 4c were validated by targeted LC–MS/MS to increase confidence in the data obtained by shotgun lipidomics. After extraction of lipids by SIMPLEX, the dried lipid phase was resuspended in 50 µl of butanol solvent (1-butanol:IPA:H2O, 8:23:69, v/v/v + 5 mM phosphoric acid) and separated by reversed-phase LC–MS/MS. LC parameters are as described in the section ‘Reversed-phase LC–MS/MS of biotinylated PS.’ The LC was coupled to the Q Exactive HF instrument applying the following ESI source parameters: spray voltage, 4.0 kV and 3.8 kV in positive and negative modes, respectively; capillary temperature, 270 °C; sheath gas flow rate, 50; auxiliary gas flow rate, 15; auxiliary gas heater temperature, 380 °C; S-lens RF level, 60. GLs were analyzed in positive mode, and GPs were analyzed in negative mode. Except for TG and CL, all lipids were quantified on the MS2 level. High-resolution MS full scan and parallel reaction monitoring were performed in one measuring cycle (MS1: 0.0–35.0 min negative mode, resolution 60,000, 350–1500 m/z; 13.0–35.0 min positive mode, resolution 30,000, 400–900 m/z; MS2: 0.0–16.0 min negative mode, resolution 30,000, nCE 24; 13.0–16.0 min positive mode, resolution 30,000, nCE 21). An AGC target of 106 and 105 and a maximum IT of 50 ms and 115 ms were used in full scan and parallel reaction monitoring, respectively. A pooled sample was measured in both polarities separately to verify the identification and acquire MS2 data for TG and CL.
Visualization and network analysis
For further investigation of common patterns in the lipid profiles, we performed both similarity-based clustering and network analysis based on the analysis approach of ref. 12. In the first step, we compared all lipids with each other pairwise. For each lipid, we computed the mean abundances for days 1, 3 and 7, which we denote as a lipid profile. For each lipid pair, we compared their lipid profiles using Pearson correlation. The result is a quadratic similarity matrix m. For the network analysis, we have drawn a graph in which we connected all lipids with each other that had a cosine similarity ≥99% in m. For the cluster analysis, we sorted m row-wise and column-wise equally. To do so, we applied hierarchical clustering on the columns of m with cosine similarity and unweighted average clustering. The result is a sorted matrix m. Since we were interested in lipid profiles even with anticorrelation, we worked only with the absolute values of m. Along the diagonal of the sorted matrix m, we searched for the biggest nonintersecting squared areas, where all values within the squares have a Pearson correlation value of ≥99%. The results are presented in Fig. 3e–h. Networks were generated using Cytoscape (v3.9.1).
Cell vitality assay
Cell vitality of control and inhibitor-treated MKs was determined by a Promega CellTiter-Glo 2.0 Assay, based on the quantification of ATP and indication of metabolically active cells, as done in ref. 78. Experiments were performed according to the manufacturer’s protocol. Per 96 wells, 10,000 MKs were seeded in TPO-supplemented DMEM. Inhibitors of phospholipid synthesis (160 µM FSG67, 5 µM triacsin C), 1 µM ionomycin or DMSO were added to the cells in single wells and cultivated for 0, 3 and 7 days. Ionomycin was added to cells as a positive control for apoptotic cells. Plates were equilibrated to room temperature for approximately 30 min before the addition of CellTiter-Glo 2.0 reagent (equilibrated to 22 °C). The reagent was added to each well in a 1:1 ratio of reagent:cell culture medium, mixed and incubated for 2 min on an orbital shaker to induce cell lysis. After 10 min, luminescence signals were recorded using a GloMax-Multi Detection System (Promega, 9300-002).
Subcellular protein fractionation
Subcellular protein fractions were obtained using the Subcellular Protein Fractionation Kit for Cultured Cells (Thermo Scientific, 78840) following the manufacturer’s instructions with some modifications. MKs were isolated after cultivation for 0 days and 7 days in TPO-supplemented DMEM and washed with ice-cold PBS. One hundred thousand cells were pelleted by centrifugation for 2 min at 500 × g. Cell pellets were dried, 100 µl of ice-cold CEB containing protease inhibitors (1:100) were added, and cell pellets were incubated for 30 min at 4 °C while mixing on an end-over-end shaker. After centrifugation for 5 min at 500 × g, the supernatant was collected, and to the remaining cell pellet, 100 µl of ice-cold MEB containing protease inhibitors (1:100) were added, vortexed for 5 s and incubated for 10 min at 4 °C while mixing. The supernatant (membrane fraction) was collected after 5 min of centrifugation at 3,000 × g and frozen at −80 °C until immunoblot analysis.
Immunoblot analysis was performed using the prepared membrane fraction of cultivated MKs (day 0 and day 7) in the absence or presence of the inhibitors. After centrifugation for 15 min at 20,000 × g at 4 °C, the supernatant was collected, and the protein concentration was measured using a Bradford assay from Bio-Rad. For immunoblotting, protein was loaded in 12% gels and electrotransferred onto a nitrocellulose membrane, followed by blocking with 5% nonfat milk or 5% BSA for 1 h at room temperature. Afterward, the membrane was incubated with the primary antibody against CSNK2A1 (1:1,000), PKHO1/CKIP-1 (1:200) or GAPDH (1:1,000) overnight at 4 °C. After washing with TBS-T, the blots were incubated with fluorochrome-conjugated secondary antibodies (1:15,000) for 1 h at room temperature. After washing, antibody binding was detected with an Odyssey infrared imaging system (LI-COR). Bands were quantified with ImageJ (National Institutes of Health)79.
Functional assessment of megakaryopoiesis
To validate the importance of the observed lipidomic changes on megakaryopoiesis, inhibitors of two enzymes involved in phospholipid biosynthesis were used. We added 160 µM FSG67, 5 µM triacsin C or vehicle control (DMSO) to the cell suspensions 24 h after the start of cultivation.
Lipidomic analysis (see the section ‘Lipid analysis’) was performed on MKs isolated via BSA gradient on day 0 (before the addition of inhibitors and control) and after 7 days of cultivation.
Proplatelet formation assay
Proplatelet formation assays were performed in triplicates on 48-well plates. After isolation via BSA gradient after 3 days of cultivation, 15,000 cells per well were seeded. Proplatelet formation was examined every 6 h by microscopy (ECLIPSE Ti2, NIS-Elements imaging software, Nikon). Ratios of proplatelet-forming MKs compared with non-proplatelet-forming MKs were calculated.
Ploidy measurements were performed according to ref. 1. In brief, the bone marrow of femora from B6 mice was flushed and homogenized. Cells were cultivated in 10-cm cell culture dishes containing DMEM (supplemented with 1% penicillin/streptomycin and 10% FBS), and differentiation was initiated by adding (1%) recombinant TPO. After 5 days of cultivation, the cell suspension was carefully pipetted on a two-phase BSA gradient (bottom, 1.5 ml 3% BSA in PBS; top, 1.5 ml 1.5% BSA in PBS) to separate cells by weight. After 40 min, the supernatant was removed, and the cell pellet was washed three times with 500 µl of PBS. Nonspecific bone marrow binding was blocked by incubation with 0.02 mg ml−1 FcR Blocking Reagent. Afterward, MKs were stained using FITC-conjugated anti-CD41 antibody, and the cells were subsequently washed once with 2 mM EDTA in PBS. Then, cells were washed with PBS (5 min at 300 × g) and fixed in PBS containing 1% PFA/0.1% EDTA. Fixed cells were washed with PBS (10 min at 300 × g) and permeabilized in PBS containing 0.1% Triton X-100. Finally, DNA was stained using 50 μg ml−1 propidium iodide staining solution containing 100 μg ml−1 RNase A and 2 mM EDTA in PBS. Analysis was performed by flow cytometry (BD FACSCalibur, BD Biosciences) and FlowJo software (Tree Star, Inc.) (Supplementary Fig. 1).
Mice were anesthetized by intraperitoneal injection of 0.5 mg per g (body weight) medetomidine, 5 mg per g (body weight) midazolam and 0.05 mg per g (body weight) fentanyl. A 1-cm midline incision was made to expose the frontoparietal skull, while carefully avoiding damage to the bone tissue. For immobilization of the head, the mice were placed on a custom-built metal stage equipped with a stereotactic holder. Bone marrow vasculature was visualized by injection of anti-CD105 Alexa Fluor 546 (0.6 µg per g (body weight)), and MKs and platelets were visualized by injection of anti-CD42a (GPIX) Alexa Fluor 488 derivative (0.8 µg per g (body weight)). Images were acquired on an upright two-photon fluorescence microscope (TCS SP8 MP, Leica Microsystems) equipped with a ×25 water objective with a numerical aperture of 1.0. A tunable broadband Ti:sapphire laser (Chameleon, Coherent) was used at 780 nm to capture Alexa Fluor 488 and 546 fluorescence. For each mouse, four to eight z-stacks with a step size of 0.51 µm were recorded from different positions in the bone marrow. Proplatelet-forming MKs were counted, and MK morphology was categorized as normal or fragmented by a blinded experimenter. ImageJ was used to generate movies (Supplementary Videos 1–3).
Immunofluorescence staining on femora cryosections
Femora of inhibitor-treated or csnk2α1lox/lox and csnk2α1Pf4∆/Pf4∆ mice were fixed with 4% PFA in 5 mM sucrose solution (Sigma-Aldrich), transferred into 10% sucrose in PBS and dehydrated using a graded sucrose series (10%–20%–30%). Subsequently, femora were embedded in Cryo-Gel (Leica) and shock-frozen in liquid nitrogen. Cryosections with a thickness of 5 μm were generated using a CryoJane Tape Transfer System (Leica) and probed with a self-conjugated FITC-anti-CD41 antibody (1:100) for specific labeling of MKs and platelets, Alexa Fluor 647-conjugated anti-CD105 antibody (1:300) for endothelium detection, and DRAQ5 (1:1,000) for nuclei staining. Samples were visualized using a Leica Stellaris 5 (LMB R039a) confocal microscope.
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
Gawaz, M., Geisler, T. & Borst, O. Current concepts and novel targets for antiplatelet therapy. Nat. Rev. Cardiol. 20, 583–599 (2023).
Tavassoli, P., Paterson, R. & Grant, J. Actinobaculum schaalii: an emerging uropathogen? Case Rep. Urol. 2012, 468516 (2012).
Wintrobe, M. M. Hematology, the Blossoming of a Science: a Story of Inspiration and Effort (Lea & Febiger, 1985).
Machlus, K. R., Thon, J. N. & Italiano, J. E.Jr. Interpreting the developmental dance of the megakaryocyte: a review of the cellular and molecular processes mediating platelet formation. Br. J. Haematol. 165, 227–236 (2014).
Eckly, A. et al. Biogenesis of the demarcation membrane system (DMS) in megakaryocytes. Blood 123, 921–930 (2014).
Geue, S. et al. Pivotal role of PDK1 in megakaryocyte cytoskeletal dynamics and polarization during platelet biogenesis. Blood 134, 1847–1858 (2019).
Bariana, T. K. et al. Sphingolipid dysregulation due to lack of functional KDSR impairs proplatelet formation causing thrombocytopenia. Haematologica 104, 1036–1045 (2019).
Golfier, S. et al. Shaping of terminal megakaryocyte differentiation and proplatelet development by sphingosine-1-phosphate receptor S1P4. FASEB J. 24, 4701–4710 (2010).
Zhang, L. et al. Sphingosine kinase 2 (Sphk2) regulates platelet biogenesis by providing intracellular sphingosine 1-phosphate (S1P). Blood 122, 791–802 (2013).
Kelly, K. L. et al. De novo lipogenesis is essential for platelet production in humans. Nat. Metab. 2, 1163–1178 (2020).
Valet, C. et al. Adipocyte fatty acid transfer supports megakaryocyte maturation. Cell Rep. 32, 107875 (2020).
Peng, B. et al. Identification of key lipids critical for platelet activation by comprehensive analysis of the platelet lipidome. Blood 132, e1–e12 (2018).
Manke, M.-C. et al. ANXA7 regulates platelet lipid metabolism and Ca2+ release in arterial thrombosis. Circ. Res. 129, 494–507 (2021).
Molenaar, M. R. et al. LION/web: a web-based ontology enrichment tool for lipidomic data analysis. GigaScience 8, giz061 (2019).
Peng, B. et al. LipidCreator workbench to probe the lipidomic landscape. Nat. Commun. 11, 2057 (2020).
Han, X. Lipidomics for studying metabolism. Nat. Rev. Endocrinol. 12, 668–679 (2016).
Borgmeyer, M. et al. Multiomics of synaptic junctions reveals altered lipid metabolism and signaling following environmental enrichment. Cell Rep. 37, 109797 (2021).
Coman, C. et al. Simultaneous metabolite, protein, lipid extraction (SIMPLEX): a combinatorial multimolecular omics approach for systems biology. Mol. Cell. Proteomics 15, 1453–1466 (2016).
Michel, L. et al. Targeting early stages of cardiotoxicity from anti-PD1 immune checkpoint inhibitor therapy. Eur. Heart J. 43, 316–329 (2022).
Kuvardina, O. N. et al. RUNX1 represses the erythroid gene expression program during megakaryocytic differentiation. Blood 125, 3570–3579 (2015).
Wang, C. Q. et al. Disruption of Runx1 and Runx3 leads to bone marrow failure and leukemia predisposition due to transcriptional and DNA repair defects. Cell Rep. 8, 767–782 (2014).
Ferreira, R., Ohneda, K., Yamamoto, M. & Philipsen, S. GATA1 function, a paradigm for transcription factors in hematopoiesis. Mol. Cell. Biol. 25, 1215–1227 (2005).
Lepage, A. et al. The αIIbβ3 integrin and GPIb–V–IX complex identify distinct stages in the maturation of CD34+ cord blood cells to megakaryocytes. Blood 96, 4169–4177 (2000).
Li, R. & Emsley, J. The organizing principle of the platelet glycoprotein Ib–IX–V complex. J. Thromb. Haemost. 11, 605–614 (2013).
Nurden, A. T. & Caen, J. P. Specific roles for platelet surface glycoproteins in platelet function. Nature 255, 720–722 (1975).
Clemetson, K. J., Pfueller, S. L., Luscher, E. F. & Jenkins, C. S. P. Isolation of the membrane glycoproteins of human blood platelets by lectin affinity chromatography. Biochim. Biophys. Acta 464, 493–508 (1977).
Sporn, L. A., Chavin, S. I., Marder, V. J. & Wagner, D. D. Biosynthesis of von Willebrand protein by human megakaryocytes. J. Clin. Invest. 76, 1102–1106 (1985).
Peck, B. et al. Inhibition of fatty acid desaturation is detrimental to cancer cell survival in metabolically compromised environments. Cancer Metab. 4, 6 (2016).
Gonzalez-Covarrubias, V. et al. Lipidomics of familial longevity. Aging Cell 12, 426–434 (2013).
Perkins, R. G. & Scott, R. E. Plasma membrane phospholipid, cholesterol and fatty acyl composition of differentiated and undifferentiated L6 myoblasts. Lipids 13, 334–337 (1978).
Heimerl, S. et al. Quantification of bulk lipid species in human platelets and their thrombin-induced release. Sci. Rep. 13, 6154 (2023).
Tani, M., Sano, T., Ito, M. & Igarashi, Y. Mechanisms of sphingosine and sphingosine 1-phosphate generation in human platelets. J. Lipid Res. 46, 2458–2467 (2005).
Niazi, H. et al. Murine platelet production is suppressed by S1P release in the hematopoietic niche, not facilitated by blood S1P sensing. Blood Adv. 3, 1702–1713 (2019).
Haslam, R. J., Koide, H. B. & Hemmings, B. A. Pleckstrin domain homology. Nature 363, 309–310 (1993).
Sutton, R. B., Davletov, B. A., Berghuis, A. M., Südhof, T. C. & Sprang, S. R. Structure of the first C2 domain of synaptotagmin I: a novel Ca2+/phospholipid-binding fold. Cell 80, 929–938 (1995).
Zhang, G., Kazanietz, M. G., Blumberg, P. M. & Hurley, J. H. Crystal structure of the cys2 activator-binding domain of protein kinase Cδ in complex with phorbol ester. Cell 81, 917–924 (1995).
Pearson, M. A., Reczek, D., Bretscher, A. & Karplus, P. A. Structure of the ERM protein moesin reveals the FERM domain fold masked by an extended actin binding tail domain. Cell 101, 259–270 (2000).
Park, M.-J. et al. SH2 domains serve as lipid-binding modules for pTyr-signaling proteins. Mol. Cell 62, 7–20 (2016).
Huang, K.-P. The mechanism of protein kinase C activation. Trends Neurosci. 12, 425–432 (1989).
Murray, N. R. & Fields, A. P. Phosphatidylglycerol is a physiologic activator of nuclear protein kinase C. J. Biol. Chem. 273, 11514–11520 (1998).
Saito, K., Scharenberg, A. M. & Kinet, J.-P. Interaction between the Btk PH domain and phosphatidylinositol-3,4,5-trisphosphate directly regulates Btk. J. Biol. Chem. 276, 16201–16206 (2001).
Niggli, V., Andréoli, C., Roy, C. & Mangeat, P. Identification of a phosphatidylinositol‐4, 5‐bisphosphate‐binding domain in the N‐terminal region of ezrin. FEBS Lett. 376, 172–176 (1995).
Turro, E. et al. A dominant gain-of-function mutation in universal tyrosine kinase SRC causes thrombocytopenia, myelofibrosis, bleeding, and bone pathologies. Sci. Transl. Med. 8, 328ra30 (2016).
Lannutti, B. J., Shim, M.-H., Blake, N., Reems, J. A. & Drachman, J. G. Identification and activation of Src family kinases in primary megakaryocytes. Exp. Hematol. 31, 1268–1274 (2003).
Huang, J. et al. Abivertinib inhibits megakaryocyte differentiation and platelet biogenesis. Front. Med. 16, 416–428 (2022).
Boni, L. T. & Rando, R. R. The nature of protein kinase C activation by physically defined phospholipid vesicles and diacylglycerols. J. Biol. Chem. 260, 10819–10825 (1985).
Newton, A. C. Lipid activation of protein kinases. J. Lipid Res. 50, S266–S271 (2009).
Garcia, P. et al. The pleckstrin homology domain of phospholipase C-δ1 binds with high affinity to phosphatidylinositol 4,5-bisphosphate in bilayer membranes. Biochemistry 34, 16228–16234 (1995).
Nishizuka, Y. Intracellular signaling by hydrolysis of phospholipids and activation of protein kinase C. Science 258, 607–614 (1992).
Bader, G. D. & Hogue, C. W. V. An automated method for finding molecular complexes in large protein interaction networks. BMC Bioinformatics 4, 2 (2003).
Shannon, P. et al. Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res. 13, 2498–2504 (2003).
Bosc, D. G. et al. Identification and characterization of CKIP-1, a novel pleckstrin homology domain-containing protein that interacts with protein kinase CK2. J. Biol. Chem. 275, 14295–14306 (2000).
Olsten, M. E. K., Canton, D. A., Zhang, C., Walton, P. A. & Litchfield, D. W. The pleckstrin homology domain of CK2 interacting protein-1 is required for interactions and recruitment of protein kinase CK2 to the plasma membrane. J. Biol. Chem. 279, 42114–42127 (2004).
Canton, D. A., Olsten, M. E. K., Niederstrasser, H., Cooper, J. A. & Litchfield, D. W. The role of CKIP-1 in cell morphology depends on its interaction with actin-capping protein. J. Biol. Chem. 281, 36347–36359 (2006).
Nie, J. et al. CKIP-1: a scaffold protein and potential therapeutic target integrating multiple signaling pathways and physiological functions. Ageing Res. Rev. 12, 276–281 (2013).
Manke, M.-C., Ahrends, R. & Borst, O. Platelet lipid metabolism in vascular thrombo-inflammation. Pharmacol. Ther. 237, 108258 (2022).
Machlus, K. R. et al. Synthesis and dephosphorylation of MARCKS in the late stages of megakaryocyte maturation drive proplatelet formation. Blood 127, 1468–1480 (2016).
Münzer, P. et al. CK2β regulates thrombopoiesis and Ca2+-triggered platelet activation in arterial thrombosis. Blood 130, 2774–2785 (2017).
Fan, J. et al. A novel role of CKIP-1 in promoting megakaryocytic differentiation. Oncotarget 8, 30138–30150 (2017).
Zehentmeier, S. et al. Static and dynamic components synergize to form a stable survival niche for bone marrow plasma cells. Eur. J. Immunol. 44, 2306–2317 (2014).
Stegner, D. et al. Thrombopoiesis is spatially regulated by the bone marrow vasculature. Nat. Commun. 8, 127 (2017).
Rebholz, H., Zhou, M., Nairn, A. C., Greengard, P. & Flajolet, M. Selective knockout of the casein kinase 2 in D1 medium spiny neurons controls dopaminergic function. Biol. Psychiatry 74, 113–121 (2013).
Cotte, A. K. et al. Lysophosphatidylcholine acyltransferase 2-mediated lipid droplet production supports colorectal cancer chemoresistance. Nat. Commun. 9, 322 (2018).
Kuhajda, F. P. et al. Pharmacological glycerol-3-phosphate acyltransferase inhibition decreases food intake and adiposity and increases insulin sensitivity in diet-induced obesity. Am. J. Physiol. Regul. Integr. Comp. Physiol. 301, R116–R130 (2011).
Heib, T., Gross, C., Müller, M.-L., Stegner, D. & Pleines, I. Isolation of murine bone marrow by centrifugation or flushing for the analysis of hematopoietic cells — a comparative study. Platelets 32, 601–607 (2021).
Zougman, A., Selby, P. J. & Banks, R. E. Suspension trapping (STrap) sample preparation method for bottom-up proteomics analysis. Proteomics 14, 1006–1000 (2014).
Seiser, S. et al. Octenidine-based hydrogel shows anti-inflammatory and protease-inhibitory capacities in wounded human skin. Sci. Rep. 11, 32 (2021).
Weiss, T. et al. Schwann cell plasticity regulates neuroblastic tumor cell differentiation via epidermal growth factor-like protein 8. Nat. Commun. 12, 1624 (2021).
Kovarik, J. J. et al. A multi-omics based anti-inflammatory immune signature characterizes long COVID-19 syndrome. iScience 26, 105717 (2023).
Cox, J. & Mann, M. MaxQuant enables high peptide identification rates, individualized p.p.b.-range mass accuracies and proteome-wide protein quantification. Nat. Biotechnol. 26, 1367–1372 (2008).
Cox, J. & Mann, M. 1D and 2D annotation enrichment: a statistical method integrating quantitative proteomics with complementary high-throughput data. BMC Bioinformatics 13, S12 (2012).
Nolte, H., MacVicar, T. D., Tellkamp, F. & Krüger, M. Instant Clue: a software suite for interactive data visualization and analysis. Sci. Rep. 8, 12648 (2018).
Peng, B. et al. A comprehensive high-resolution targeted workflow for the deep profiling of sphingolipids. Anal. Chem. 89, 12480–12487 (2017).
Herzog, R. et al. LipidXplorer: a software for consensual cross-platform lipidomics. PLoS ONE 7, e29851 (2012).
Herzog, R. et al. A novel informatics concept for high-throughput shotgun lipidomics based on the molecular fragmentation query language. Genome Biol. 12, R8 (2011).
Thomas, C. P. et al. Identification and quantification of aminophospholipid molecular species on the surface of apoptotic and activated cells. Nat. Protoc. 9, 51–63 (2014).
Rampler, E. et al. LILY-lipidome isotope labeling of yeast: in vivo synthesis of 13C labeled reference lipids for quantification by mass spectrometry. Analyst 142, 1891–1899 (2017).
Josefsson, E. C. et al. Platelet production proceeds independently of the intrinsic and extrinsic apoptosis pathways. Nat. Commun. 5, 3455 (2014).
Schneider, C. A., Rasband, W. S. & Eliceiri, K. W. NIH Image to ImageJ: 25 years of image analysis. Nat. Methods 9, 671–675 (2012).
This study was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation), project numbers 374031971 – CRC240, BO3786/3-1 and BO3786/7-1, 453989101 – CRC1525 (O.B.), and the Austrian Science Fund, FWF DMP I6303. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. The authors received support from the University of Vienna through seed funding and funding derived from the Vienna Doctoral School in Chemistry (DoSChem) at the Faculty of Chemistry, University of Vienna (R.A.), as well as from the DigiOmics4Austria: BMBWF Ausschreibung “(Digitale) Forschungsinfrastrukturen” (R.A.). We further thank V. O’Donnell (Systems Immunity Research Institute, School of Medicine, Cardiff University) for providing PS surface labeling protocols and D. Eißler (DFG Heisenberg Group Cardiovascular Thromboinflammation and Translational Thrombocardiology, University of Tübingen, and Department of Cardiology and Angiology, University of Tübingen) for excellent technical assistance.
The authors declare no competing financial interests.
Peer review information
Nature Cardiovascular Research thanks John Hwa, Matthew A Mitsche and Angel Garcia Alonso for their contribution to the peer review of this work.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Extended Data Fig. 1 Protein regulation highlighting key proteins involved in lipid metabolism. Related to Fig. 1.
a, Volcano plots of -log10 p-values over log2 fold changes of all identified proteins on day 0, 1, 3 and 7 of MK maturation. P-values were corrected using the Benjamini-Hochberg correction with an FDR cut-off of 0.05. b,c, Proteins of fatty acid transport pathways (b) as well as identified phospholipases (c) are depicted. A two-sided t-test was used for statistical analysis. Benjamini-Hochberg correction was applied to p-values using an FDR cut-off < 0.05 (*P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001). d, Network highlighting proteins involved in lipid metabolism. Edges are correlations of r ≥ 0.85. Nodes represent proteins and the node color the associated lipid category (left). The network on the right is color-coded based on the log2 fold change of proteins regulated from day 0 to day 7. Pure red indicates an FC ≥ 2 and pure blue an FC ≤ -2. Data are combined from 3 independent biological experiments and mean values are shown. Error bars represent standard deviations.
Extended Data Fig. 2 Fuzzy c-means clustering of down-regulated proteins during MK maturation. Related to Fig. 1.
Protein cluster (C1-C18) showing overall downregulation. Number of proteins and their median are denoted in individual plots as well as the associated cluster (C1-C18). The assignment of proteins to clusters can be found in the Source Data. Threshold = 85.
Extended Data Fig. 3 Fuzzy c-means clustering of up-regulated proteins during MK maturation. Related to Fig. 1.
Protein cluster (C19-C39) showing overall upregulation. Number of proteins and their median are denoted in individual plots as well as the associated cluster (C19-C39). The assignment of proteins to clusters can be found in the Source Data. Threshold = 85.
Extended Data Fig. 4 Structural elucidation at the FA level of lipids in the main lipid categories. Related to Fig. 2.
a-d, Molecular structural analysis of lipids identified with shotgun lipidomics at the MS2 level. The MS2 spectra shown in each panel are used as structural assessment by customized MFQL search files using the LipidXplorer search engine. Potential fragmentation patterns are shown above each panel. Annotated fragments are denoted in the chemical structure of the individual species. For lipid identification, FAs and loss of FAs as well as class-specific headgroups or neutral losses were used. Quantification of lipid species on the molecular species level is based on FA intensities.
Extended Data Fig. 5 Quantitative elucidation at the FA level of lipids in the main lipid categories. Related to Fig. 2.
a-d, Radar charts displaying the relative intensity of lipid species of mature MKs organized according to their lipid category, including GP, SP, GL, and STs. GPs and GLs were analyzed with shotgun lipidomics while SPs and STs were analyzed with targeted lipidomics.
Extended Data Fig. 6 Validation of shotgun lipidomics data by targeted RPLC-MS/MS. Related to Fig. 4.
Trends of selected class representative lipid species were confirmed by targeted reverse-phase LC-MS/MS. Shown in blue are data obtained from shotgun measurements, shown in grey are results from LC experiments. For better comparison of observed lipid trends, lipid quantities are shown as relative amount compared to day 1. All data show the mean of 3 biological replicates. One biological replicate was comprised of 5 individual animals. Error bars represent standard deviations.
Extended Data Fig. 7 MS2 spectra for the structural elucidation of representative lipid species. Related to Fig. 4.
All MS2 spectra were acquired using a collision energy of 21 % in positive ion mode and 24 % in negative ion mode at R m/z200 of 30000. GP (glycerol phosphate backbone), FA (fatty acyl), HG (class specific head group fragment), IP (inositol phosphate), I (inositol) and NL (neutral loss of class specific headgroup), indicate structure specific fragments. Scan polarity is indicated on the upper left side of each panel. The fragments annotated were used for lipid identification. Quantification of lipid species on the molecular species level is based on FA intensities. CL and TG were quantified on MS1 level based on precursor intensities.
Extended Data Fig. 8 Exclusion of media lipid composition and apoptosis as confounding factors for observed lipidomic changes in MKs. Related to Fig. 4.
a, Distribution of main lipid classes from mature MKs (day 3, n = 3) as well as FBS (n = 11) used for cell culture supplementation. Lipid quantities were normalized to volume. b, Proteomic analysis of pro- and anti-apoptotic markers in MKs. A two-sided t-test was used for statistical analysis. Benjamini-Hochberg correction was applied to p-values using an FDR cut-off < 0.05 (*P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001). c, ATP cell vitality assay showing MKs treated with inhibitors. Ionomycin was used as positive control for apoptotic cells. The measured luminescence was normalized to the control for each day to display changes relative to the control baseline (dark grey). d, PS externalization during megakaryopoiesis. The two most abundant PS species were monitored. Externalization was calculated by dividing the amount of biotinylated PS on the cell surface by the total biotinylated PS, as described in the methods. Lipid quantities of day 1 were used as reference and set to 1. All other days were calculated as ratios relative to day 1 (n = 5). All data show the mean of at least 3 biological replicates. One biological replicate was comprised of 5 individual animals. In a-b, error bars represent standard deviations. In d, boxplot whiskers represent the minimum and maximum. The boundaries of the box represent the 25th and 75th percentile. Middle black lines represent the mean and black squares indicate the median.
Extended Data Fig. 9 Analysis of the megakaryocytes upon inhibition of phospholipid biosynthesis. Related to Fig. 5.
a, Lipidome analysis showing the relative quantities (mean ± SD, n = 4) per lipid class of control and inhibitor treated MKs on day 7. Non-treated day 0 MKs were used as baseline control and lipid quantities were set to 1 and the standard deviation is depicted as dotted lines. Changes for all lipid classes were calculated as ratios relative to control day 0. A two-sided t-test was used for statistical analysis. All days were tested against day 7 control (*P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001). b, Arithmetic means ± SD (n = 5-7) of MKs of Triacsin C, FSG67 or vehicle treated mice per visual field of in vivo imaging in the BM vasculature of the frontoparietal skull of mice. c, Arithmetic mean ± SD (n = 5-7) of MKs and MK area of Triacsin C, FSG67 or vehicle treated mice per visual field in immunostained BM sections.
Extended Data Fig. 10 Lipid-dependent kinases in MKs involved in megakaryopoiesis and proplatelet formation. Related to Fig. 6.
a, Network showing significantly regulated kinases (blue) and lipid-binding proteins (yellow). Edges are correlations with a minimum interaction confidence score 0.5. Nodes are proteins. b, Domain composition of depicted kinases. Membrane-targeting modules are the C1 (orange), C2 (yellow), pleckstrin homology (PH) or PH-like (purple), and FERM (light blue) domains. Additional interaction domains like Src-homology (SH) domains are shown in dark blue. Kinase domains are shown in gray. Note, the schematic representation is not reproduced in the correct domain proportions and does not reflect the actual protein size. c, Regulation of kinases with a lipid binding domain (mean ± SD, n = 3). A two-sided t-test was used for statistical analysis. Benjamini Hochberg correction was applied to p-values using an FDR cut-off of 0.05 (*P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001). d, Arithmetic means of MKs and MK area of csnk2α1lox/lox and csnk2α1Pf4∆/Pf4∆ mice per visual field (n = 30) in immunostained BM sections.
Supplementary Fig. 1 and Tables 1 and 2.
2P-IVM of DMSO-treated mice in vivo. MKs and platelets were stained with anti-GPIX antibody derivatives (green), and vessels were labeled with anti-CD105 antibodies (magenta).
2P-IVM revealing reduced proplatelet formation in triacsin C-treated mice in vivo. MKs and platelets were stained with anti-GPIX antibody derivatives (green), and vessels were labeled with anti-CD105 antibodies (magenta).
2P-IVM revealing reduced proplatelet formation in FSG67-treated mice in vivo. MKs and platelets were stained with anti-GPIX antibody derivatives (green), and vessels were labeled with anti-CD105 antibodies (magenta).
Statistical source data.
Statistical source data.
Statistical source data.
Statistical source data.
Statistical source data.
Statistical source data.
Unprocessed western blots.
Statistical source data.
Statistical source data.
Statistical source data.
Statistical source data.
Statistical source data.
Statistical source data.
About this article
Cite this article
de Jonckheere, B., Kollotzek, F., Münzer, P. et al. Critical shifts in lipid metabolism promote megakaryocyte differentiation and proplatelet formation. Nat Cardiovasc Res 2, 835–852 (2023). https://doi.org/10.1038/s44161-023-00325-8