Cullin-5 controls the number of megakaryocyte-committed stem cells to prevent thrombocytosis in mice

Cullin proteins coordinate assembly of cullin-RING-E3 ubiquitin (Ub) ligase (CRL) complexes that control posttranslational Ub modification and degradation of cellular proteins. Cullin-5 (Cul5) CRL complexes contain a RING E3 ligase Rbx1/2, the adapter proteins Elongin B and C, and a suppressors of cytokine signaling (SOCS)-box–containing protein (supplemental Figure 1A; available on the Blood website). The SOCS-box motif has been identified in as many as 80 human proteins, most belonging to 1 of 6 families, defined by the additional presence of specific protein interaction domains, including SH2 domains (SOCS family), SPRY domains (SPSB family), ankyrin repeats (ASB family), WD40 motifs (WSB family), and leucine repeats.¹ The SOCS-box is required for incorporation into the CRL complex, while these additional domains are implicated in recruitment of specific proteins for degradation. To date, the roles of CRL-SOCS-box complexes have been investigated via individual analysis of a limited number of candidate SOCS proteins. In contrast, with a focus on hematopoiesis, here we examined the indispensable roles of the CRL-SOCS-box family by generating mice lacking Cul5, the absence of which is predicted to impair the SOCS-box-dependent actions of all proteins that contain this motif, including Ub-mediated degradation of associated substrates.

The SOCS family is best known for regulating the actions of hematopoietic cytokines that utilize the JAK-STAT signaling pathway.² For example, in the absence of SOCS1, uncontrolled interferon-γ (IFN-γ) signaling causes immune cell deregulation and early postnatal lethality.³ SOCS1 is also essential to control signaling by type-I interferons, as well as cytokines that utilize the interleukin-2 (IL-2) family γ-common receptor.^4-7 In contrast, genetic deletion of Socs3 has revealed specific deregulation of signaling via granulocyte-colony stimulating factor (G-CSF), leptin, and cytokines that use the gp130 receptor chain,^8-11 while the cytokine-inducible SH2-containing protein (CIS) has been shown to control IL-15 and granulocyte-macrophage CSF (GM-CSF) signaling in natural killer and myeloid cells.^12,13 However, analyses of knockout mice for each of the 8 SOCS family members have not revealed abnormalities in signaling by several hematopoietic cytokines, including thrombopoietin (TPO) and IL-3, among others, despite the fact that these cytokines activate the JAK-STAT pathway and are known to induce the expression of multiple SOCS proteins.¹⁴ Indeed, the roles of most SOCS-box family proteins in hematopoiesis have not been systematically studied in physiological contexts. While SOCS-box proteins are thought to promote Ub-mediated degradation of substrates via their participation in Cul5-containing CRL complexes, independent direct actions can also occur. For example, SOCS1 and SOCS3 attenuate cytokine signaling by a combination of direct inhibition of the activity of JAK kinases via specific domains independent of the SOCS-box.^15,16 The importance of the SOCS-box/CRL-dependent mechanism is evident in mice specifically lacking the SOCS-box domain of SOCS1 or SOCS3, which display partial loss of cytokine regulation that can result in significant pathology.^17,18 

Recent evidence suggests that multiple pathways operate for production of megakaryocytes, the precursor cells to blood platelets. In addition to stepwise commitment to megakaryocyte production via progressively restricted progenitor cells, a distinct population within the hematopoietic stem cell (HSC) pool exists that can rapidly, perhaps directly, generate megakaryocytes. These megakaryocyte-committed stem cells, which appear to have distinct molecular profiles, including prominent expression of CD41 and the platelet-associated protein von Willebrand factor (VWF), have been associated with responses to inflammation or infection, but may also contribute to steady-state megakaryopoiesis.^19-26 We show here that mice lacking Cul5 develop excess megakaryopoiesis and thrombocytosis. In Cul5-deficient mice, deregulation of HSCs results in more HSCs overall and a biased increase in CD41⁺HSC production, as well as increased megakaryopoietic activity of the CD41⁺HSCs produced. This process is largely independent of TPO and the TPO receptor (myeloproliferative leukemia protein [Mpl]), the major cytokine regulator of megakaryocyte production, and involves signaling via the beta-common (β_c) and/or beta-IL-3 (β_IL3) receptors, with evidence of deregulated responses to IL-3. The phenotype is independent of the interferon alpha/beta receptor, (IFNARI), previously implicated in inflammation-induced activation of stem-like megakaryocyte progenitors.²¹ 

Conditional Cul5 knockout mice were generated using CRISPR/Cas9 technology (supplemental Figure 1B; supplemental Table 1); further details regarding the protocol and specific Cre-recombinase-expressing mice are provided in supplemental Methods. The animal experiments were approved by Walter and Eliza Hall Institute Animal Ethics Committee.

For transplantation, C57BL/6 CD45.1 mice received two 5.5 Gy doses of irradiation given 3 hours apart from a ⁶⁰Co source and then 2.5 × 10⁶ CD45.2 bone marrow (BM) cells from either Cul5^fl/fl or Cul5^fl/flCreERT2 mice. After 5 to 6 weeks, reconstitution of recipient mice (over 75% CD45.2) was confirmed by analysis of peripheral blood, and at 7 weeks after transplantation, cohorts were treated with either tamoxifen (TAM, 4.2 mg in corn oil by oral gavage) or with vehicle (VEH) on 2 consecutive days. Mice were analyzed after 16 weeks.

Cell counts were performed on blood collected into EDTA-coated tubes (Sarstedt, Numbrecht, Germany) using an Advia-2120 hematological analyzer (Siemens, Forchheim, Germany). Sternums were fixed in 10% neutral-buffered formalin, decalcified, and embedded in paraffin for sectioning and staining with haematoxylin and eosin. Megakaryocytes were counted from 10 nonoverlapping fields using a Nikon Eclipse E600 microscope at 600× final magnification. Details of cell cultures are provided in supplemental Methods.

Single-cell suspensions were stained with fluorophore-conjugated antibodies on ice and analyzed on LSRFortessa, LSRII or FACSymphony A3 instruments (Becton Dickinson, Mulgrave, Australia). Cell sorting was performed on a FACSAria III (Becton Dickinson). Dead cells were excluded by FluoroGold (AAT Bioquest, Pleasanton, CA) and data were analyzed using FlowJo (Becton Dickinson). Cell-surface markers used to define hematopoietic cell types are provided in supplemental Table 2, antibodies used in supplemental Table 3 and gating profiles in supplemental Figure 2A-C.

Single-cell BM suspensions were enriched for stem and progenitor cells by negative selection of mature cells using EasySep Mouse Hematopoietic Progenitor Cell Isolation Kits (Stem Cell Technologies, Tullamarine, Australia). Cells were stained with metal-tagged antibodies to cell surface markers (supplemental Table 3) on ice for 1 hour in MaxPar Cell Staining Buffer (Standard Biotools, South San Fransisco, CA). Viability stain (25 μM Cell-ID Cisplatin; Standard Biotools) was added in 37°C serum-free media for 1 minute, cells were fixed in 1.6% paraformaldehyde for 10 minutes and permeabilized in methanol for 15 minutes on ice. After 4 washes, cells were stained with intracellular antibodies (supplemental Table 3) and incubated overnight in 1.6% paraformaldehyde with Cell-IDTM Intercalator-Ir (Standard Biotools). Samples were analyzed on a Helios mass cytometer (Standard Biotools). Gating strategies for Lineage⁻ScaI⁺Kit⁺ (LSK) cells are shown in supplemental Figure 2D. Data were analyzed using FloJo or Cytobank Premium (Beckman Coulter Life Sciences, Lane Cove, Australia).

Cells were lysed in 1% TritonX-100, 150 mM NaCl, 50 mM Tris HCl (pH 7.4), 1 mM EDTA, 1 mM phenylmethylsulphonylfluoride, 2 mM Na₃VO₄, 10 mM NaF and complete protease inhibitors (Merck, Darmstadt, Germany). Proteins were separated in 4% to 12% Bis-Tris NuPAGE protein gels (Thermo Fisher Scientific, Waltham, MA) under reducing conditions, transferred to an Immobilon-P membrane (Merck, Darmstadt, Germany) and immunoblotted with primary antibodies, followed by secondary horseradish peroxidase–conjugated antibody and visualized by enhanced chemiluminescence. Antibodies used are provided in supplemental Table 3.

Whole cell lysates were prepared from 6 replicates of LSK cells from each of Cul5^fl/fl and Cul5^fl/flVavCre mice and processed for liquid chromatography tandem mass spectrometry (LC-MS/MS) and bioinformatics analysis; further details are provided in supplemental Methods.

Single-cell RNA sequencing was performed on flow-cytometry sorted LSK cells using the Chromium system (10x Genomics, Pleasanton, CA). Raw sequencing data were processed using the CellRanger pipeline (10x Genomics) and read into Seurat for downstream analysis. Further details are provided in supplemental Methods.

Mice with a conditional knockout allele of Cul5 were generated via CRISPR-mediated recombination (supplemental Figure 1B). Cul5^fl/fl mice were crossed to VavCre mice²⁷ to target Cul5 deletion to the hematopoietic system. Loss of Cul5 was confirmed in western blots of Cul5^fl/flVavCre BM extracts (supplemental Figure 1C). Thrombocytosis was evident in Cul5^fl/flVavCre mice that was not observed in Cul5^+/+, Cul5^fl/fl or Cul5^+/+VavCre control groups (Figure 1A, supplemental Table 4) and was accompanied by increased numbers of megakaryocytes in the BM (Figure 1B, supplemental Figure 3A). Megakaryocyte progenitor (MkP) cells were modestly increased (Figure 1C), and within the LSK stem/progenitor cell fraction, a selective increase in the numbers of phenotypic HSC was evident, while numbers of multipotent progenitor cells (MPP), restricted progenitor cells (HPC1, HPC2, lymphoid-biased HPC1Flt3^hi and myeloid-biased HPCFlt3^lo), common myeloid progenitor (CMP), granulocyte-macrophage progenitor (GMP), megakaryocyte-erythroid progenitor (MEP), and colony-forming erythroid cells (CFU-E)^28,29 were normal (supplemental Table 2; Figure 1D). Numbers of circulating lymphocytes were reduced, (supplemental Table 4) and a selective deficit of B-lymphoid cells in the BM and spleen and elevated myeloid cells in the spleen were observed (supplemental Figure 3B).

Excess megakaryopoiesis in the absence of Cul5 was confirmed in an independent model. Wild-type C57BL/6 mice (CD45.1) were transplanted with BM cells from either Cul5^fl/fl or Cul5^fl/flCreERT2 mice (CD45.2), the latter heterozygous for the CreERT2 allele allowing activation of Cre recombinase following treatment with TAM.³⁰ As observed in Cul5^fl/flVavCre mice, thrombocytosis was evident in TAM-treated Cul5^fl/flCreERT2 mice, that was not observed in the control groups (TAM-treated Cul5^fl/fl, VEH-treated Cul5^fl/flCreERT2 and VEH-treated Cul5^fl/fl, supplemental Figure 4A), accompanied by increased numbers of megakaryocytes, MkP, HSC and HPC2 (supplemental Figure 4B-E). In this model, the number of platelets was higher than in Cul5^fl/flVavCre mice, and circulating neutrophils, monocytes, and eosinophils were also elevated relative to controls (supplemental Table 4). We hypothesized that this may at least in part reflect progression of the phenotype with age: data from Cul5^fl/flVavCre mice were collected at 8 to 11 weeks of age, while mice were irradiated, transplanted, and analyzed 16 weeks after TAM-induced deletion of Cul5. Indeed, in a cohort of aged Cul5^fl/flVavCre mice (22-27 weeks), platelet counts were higher than in the 8- to 11-week cohort and numbers of myeloid blood cells were also significantly elevated relative to aged Cul5^fl/fl controls (supplemental Table 4), suggesting progressive development of a broader myeloproliferative phenotype in the absence of Cul5.

Thus, loss of Cul5 in hematopoiesis results in enhanced production of HSC and excess megakaryopoiesis and thrombocytosis, effects intrinsic to loss of Cul5 in hematopoietic cells. In contrast to Cul5^fl/flVavCre mice, which lack Cul5 throughout the hematopoietic system, in Cul5^fl/flPF4Cre mice lacking Cul5 specifically in megakaryocytes and platelets, the numbers of platelets, megakaryocytes, and HSC were within normal ranges (Figure 2). This is consistent with a deficiency of Cul5 in the stem cell compartment driving megakaryocytosis and thrombocytosis, rather than intrinsic loss of Cul5 in megakaryocytes and platelets themselves.

Semisolid cultures of BM cells were performed using a broad myeloid stimulus of stem cell factor+IL-3+erythropoietin (EPO). The number of megakaryocyte colony-forming cells (Meg-CFC) was increased in Cul5^fl/flVavCre BM and within the LSK fraction, also accompanied by an increase in total other myeloid colony-forming cells, although no specific other myeloid colony type was significantly increased (Figure 3A-B). Within LSK subpopulations, megakaryocyte colonies developed from cells within the HPC2 subset of control Cul5^fl/fl mice, but not the other subsets (HSCs, multipotent progenitor cells, and HPC1) consistent with previous studies showing enrichment for megakaryocyte potential within this phenotypically-defined population.^24,29 Notably, in Cul5^fl/flVavCre mice, significant numbers of megakaryocyte colonies also developed from cells within the HSC population (Figure 3A). Within the Lin⁻Sca⁻Kit⁺ compartment of control mice, which contains more committed progenitor cells, Meg-CFC occur within the CD150⁺ fraction, which includes the phenotypically defined MkP (CD150⁺CD41⁺). No significant change in the frequency of Meg-CFC in this population was evident in Cul5^fl/flVavCre BM (Figure 3A).

Recent studies have identified cells within the HSC compartment that have the capacity to rapidly generate megakaryocytes/platelets, characterized by cell-surface expression of CD41 and/or intracellular expression of the platelet-associated protein VWF.^21,26 In Cul5^fl/flVavCre BM, the numbers of CD41⁺HSC (Lin⁻ScaI⁺Kit⁺CD150⁺CD48⁻CD41⁺) were significantly increased relative to control, as was the proportion of CD41⁺HSC within the HSC population (Figure 3C). In clonogenic cultures, on average 47% of colonies produced by Cul5^fl/flVavCre CD41⁺HSC were megakaryocytic, compared with 21% from Cul5^fl/fl CD41⁺HSC (Figure 3A). Consistent with the identification of these cells as megakaryocyte-committed stem cells, mass cytometry revealed prominent VWF expression in Cul5^fl/flVavCre HSC compared to barely detected levels in control Cul5^fl/fl cells, and an increased proportion of VWF⁺HSC within the HSC pool (Figure 3D).

Together, these data suggest that the absence of Cul5 results in an increased number of cells overall within the phenotypic HSC pool (6.4-fold) that includes a disproportionate increase in CD41⁺HSC (25-fold) resulting in both higher absolute numbers and increased frequency of CD41⁺HSC. With increased megakaryocytic output from the CD41⁺HSC in clonogenic cultures, this is consistent with deregulation of the HSC pool including a biased increase CD41⁺HSC production, as well as increased megakaryopoietic activity of the CD41⁺HSC produced.

Single cell RNA sequencing was performed on LSK cells purified from Cul5^fl/flVavCre and control Cul5^fl/fl mice. In control Cul5^fl/fl samples, distinct cell clusters with similarity to HSC, MMP2, MMP3 and MMP4 were identified based on comparison of differential gene expression with the published Immgen dataset,³¹ as were megakaryocyte-erythroid progenitor-, granulocyte-macrophage progenitor- and monocytic-like clusters at low abundance (Figure 4A-B). In addition to these clusters, a cluster unique to Cul5^fl/flVavCre samples was evident (Figure 4A-B). Analysis of the top 100 differentially expressed genes in this cluster revealed similarity with a long-term hematopoietic stem cell (LT-HSC) expression profile accompanied by prominent expression of VWF (Figure 4A; supplemental Figure 5; supplemental Table 5), providing further evidence for an expanded megakaryocyte-committed stem cell population in the absence of Cul5.

The proteome of Cul5^fl/flVavCre LSK cells was also compared with that of control Cul5^fl/fl cells. Among the 1066 proteins found to be differentially expressed, based on an adjusted P value ≤.05, 648 exhibited significant downregulation in protein expression within Cul5^fl/flVavCre LSK cells relative to the Cul5^fl/fl control and 418 exhibited significant upregulation (Figure 5B; supplemental Table 6). As expected Cul5 was significantly down-regulated and notably, VWF and CD41 (Itga2b) were among the upregulated proteins. Consistent with an expanded megakaryocyte-committed population, the proteome of Cul5^fl/flVavCre LSK cells showed significant enrichment for the set of proteins previously identified as characteristic of stem-like MkP²¹ (Figure 5A-B). Proteins present in both datasets included megakaryocyte-associated proteins VWF and CD41 (Itga2b) as well as several interferon-response proteins (Figure 5B). Indeed, this previous study suggested that the number of stem-like MkP increases under conditions of inflammation.²¹ Accordingly, molecular signatures of inflammation and interferon responses were more closely examined, as were other pathways known to regulate HSC behavior. Proteins associated with inflammatory response, IFN-α response, IFN-γ response and JAK/STAT signaling were upregulated in the proteome of Cul5^fl/flVavCre LSK cells compared to Cul5^fl/fl controls (Figure 5C; supplemental Figure 6; supplemental Table 7). Upregulation of cholesterol homeostasis and downregulation of oxidative phosphorylation and reactive oxygen species pathways were also noted.

Type-I interferon signaling through the IFNARI receptor is required for emergence of stem-like megakaryocyte progenitors in response to inflammatory stimuli.²¹ However, the absence of IFNARI did not ameliorate the effects of Cul5 deficiency: numbers of platelets, megakaryocytes, and HSCs in IFNARI^−/−Cul5^fl/flVavCre mice were no different to those observed in Cul5^fl/flVavCre mice (supplemental Figure 7A-C). Likewise, disabling SOCS1, a major physiological regulator of interferon signaling,⁶ from interaction with the Cul5-containing CRL by deletion of the SOCS-box (Socs1^SB/SB18) did not increase platelet counts (supplemental Figure 7D).

Upregulation of the JAK/STAT signaling pathway in the proteome of Cul5^fl/flVavCre LSK cells suggests involvement of cytokines in Cul5-deficient megakaryocytosis. TPO, acting through its specific cell surface receptor, the myeloproliferative leukemia protein (Mpl), is the major cytokine regulator of megakaryocyte and platelet production.³² To determine the role of TPO signaling in enhanced megakaryopoiesis in Cul5^fl/flVavCre mice, we generated these mice on Mpl– and TPO-deficient backgrounds (Mpl^−/−Cul5^fl/flVavCre and Thpo^−/−Cul5^fl/flVavCre). The absence of Cul5 resulted in increased platelet numbers on both c-Mpl– and TPO-null backgrounds relative to Mpl^−/− and Thpo^−/− mice respectively (Figure 6A), as well as increased numbers of megakaryocytes, HSC and CD41⁺HSC (Figure 6B-C; supplemental Figure 8A). As observed on an otherwise wild-type background, Cul5^fl/flVavCre mice lacking Mpl or TPO showed few consistent changes in numbers of hematopoietic progenitor cells relative to Cul5^fl/flMpl^−/− or Cul5^fl/flThpo^−/− mice themselves (Figure 6D). The reduction in circulating lymphocytes was evident and the increase in eosinophils was more pronounced (supplemental Table 4). Although the absolute numbers of platelets, megakaryocytes and HSC did not reach the values observed in Cul5^fl/flVavCre mice, the fold-increase in these cells in Cul5-deficient Mpl^−/− or Thpo^−/− mice, relative to Cul5 wild-type Mpl^−/− or Thpo^−/− mice, were at least the same magnitude, if not greater, than that observed in Cul5-deficient mice on a wild-type background (Figure 6A-C). This suggests that the effects of Cul5 deletion may be largely if not completely independent of the Mpl/TPO pathway.

Although TPO is the major cytokine regulator of megakaryopoiesis, cytokines such as IL-6 and IL-11, that act through the gp130 receptor, or IL-3 and GM-CSF acting via the β_c receptor (and for IL-3 in mouse also via β_IL3) can stimulate megakaryopoiesis, at least in vitro (reviewed in Behrens and Alexander³²). Because mice lacking the SOCS-box from SOCS3 (Socs3^SB/SB),¹⁷ in which gp130-family cytokine signaling is deregulated, did not exhibit increased platelet numbers (supplemental Figure 7E), we focused on the β_c/β_IL3 receptor systems.

Cul5^fl/flVavCre mice were crossed with mice lacking both the β_c and β_IL3 receptor chains (β_c^−/−β_IL3^−/−)³³ to abolish IL-3 and GM-CSF (and IL-5) signaling. Numbers of platelets and megakaryocytes were increased in β_c^−/−β_IL3^−/−Cul5^fl/flVavCre mice relative to β_c^−/−β_IL3^−/−Cul5^fl/fl controls, although absolute numbers and fold increases were less than that of Cul5^fl/flVavCre mice (Figure 7A-B; supplemental Figure 8B). In contrast, the numbers of HSC and CD41⁺HSC, were not significantly different in β_c^−/−β_IL3^−/−Cul5^fl/flVavCre mice relative to Cul5^fl/flVavCre (Figure 7C) and no changes were evident in hematopoietic progenitor or circulating blood cell populations (Figure 7D; supplemental Table 4).

Megakaryocyte colony formation from Cul5^fl/flVavCre BM revealed higher numbers with IL-3 stimulation compared to Cul5^fl/fl controls (supplemental Figure 9), while numbers of GM-CSF-stimulated megakaryocyte colonies were low in BM of both genotypes. Similarly, IL-3 caused pronounced proliferation of Cul5^fl/flVavCre LSK cells in liquid cultures, significantly in excess of that observed from Cul5^fl/fl cells, which was not consistently observed with GM-CSF (Figure 7E). Mass cytometry was used to measure levels of phospho(p)STAT5 in Cul5^fl/flVavCre and control LSK cells stimulated with IL-3 or GM-CSF. IL-3 was the more potent stimulus, inducing increased pSTAT5 in control cells within 60 minutes, while little, if any, increase was observed with GM-CSF. In Cul5^fl/flVavCre cells, the peak of IL-3-stimulated pSTAT5 was higher, and unlike in control cells, remained elevated for up to 4 hours. The induction of pSTAT5 by GM-CSF in Cul5^fl/flVavCre cells was markedly reduced relative to IL-3 (Figure 7F). Thus, the increased numbers of megakaryocytes and platelets resulting from Cul5-deficiency is dependent on β_c/β_IL3 family cytokine(s) with evidence of a significant contribution from IL-3. The lack of a reduction in HSCs in β_c^−/−β_IL3^−/−Cul5^fl/flVavCre mice implies that this system may be more involved in megakaryocyte development from increased numbers of megakaryocyte-committed stem-like cells than the generation itself of excess numbers of these cells.

Cul5 coordinates assembly of dozens of CRL complexes for posttranslational ubiquitination and degradation of myriad cellular proteins. These complexes are distinguished by the presence of different SOCS-box proteins that recruit different specific substrates. Here we generated mice lacking Cul5 within hematopoiesis, a modification that should impair the CRL-mediated function of all SOCS-box proteins and define roles of the broader SOCS family in the blood-forming system. A key discovery was the major role that Cul5 plays in regulation of megakaryopoiesis. In the absence of Cul5 throughout hematopoiesis, but not when specifically deleted in megakaryocytes and platelets, mice developed thrombocytosis accompanied by excess numbers of megakaryocytes and changes within the HSC compartment consistent with a surfeit of stem cells with megakaryocytic potential. Previous studies have identified cells within the HSC pool that have the capacity to rapidly or directly generate megakaryocytes.^21,26 Several observations suggest that the excess megakaryopoiesis in Cul5-deficient mice results from increased production of megakaryocyte-committed HSC. Cells with a surface marker profile typical of HSC and with high CD41 expression were in excess in Cul5^fl/flVavCre mice, and mass cytometry revealed prominent VWF expression in Cul5-deficient HSC. Single-cell analysis identified a distinct cluster of cells in Cul5^fl/flVavCre mice that combined a typical HSC expression profile with high VWF expression. Finally, unlike control HSC, the Cul5-deficient HSC population generated high numbers of megakaryocytes in clonogenic cultures, with increased megakaryocytic output from the CD41⁺HSC population. The data are consistent with deregulation of the HSC pool including a biased increase in CD41⁺HSC production, as well as increased megakaryopoietic activity of the CD41⁺HSC produced.

Consistent with this model, differential proteomics analysis also revealed upregulated VWF and CD41 expression in Cul5^fl/flVavCre LSK cells, as well as enrichment of proteins previously associated with stem-like megakaryocyte progenitors.²¹ Inflammatory and interferon response proteins were found to be enriched in Cul5-deficient LSK cells, as were proteins linked to JAK/STAT signaling. The IFNRI receptor is required for sustaining baseline levels of CD41⁺HSC in a model of myeloproliferative disease³⁴ and for the upregulation of megakaryocyte proteins following inflammatory stimuli.²¹ However, numbers of CD41⁺HSC, as well as megakaryocytosis and thrombocytosis in IfnarI^−/−Cul5^fl/flVavCre mice were no different to Cul5^fl/flVavCre mice. Thus, increased numbers of CD41⁺HSC in the absence of Cul5 is driven by a mechanism distinct from that previously associated with inflammation. Stimuli other than type I interferons may contribute. For example, previous studies have also implicated IL1β in the platelet bias of aged stem cells in mice³⁵ and in the activation of CD41⁺HSC following viral infection.²⁵ Other studies have shown that VWF⁺LT-HSCs display a similar frequency of platelet/myeloid- and platelet-biased output in young mice that becomes dominated by the latter in older animals.²⁰ The megakaryocyte-committed stem cells in Cul5^fl/flVavCre mice may more resemble these cells, although in the absence of Cul5, the megakaryocyte/platelet bias over other myeloid cells is more pronounced in younger mice with excessive numbers of the latter progressing with age. The data is consistent with Cul5-deficiency particularly favoring production of platelet-biased stem cells but also increasing the numbers of more balanced HSCs.

Deregulation of cytokine responses was anticipated in the absence of Cul5 given the known actions of several SOCS proteins in attenuating cytokine signaling² and is consistent with the upregulation of proteins linked to the JAK/STAT pathway in Cul5^fl/flVavCre LSK cells. The excess megakaryopoiesis in Cul5^fl/flVavCre mice was attenuated in the absence of TPO, the major cytokine regulator of megakaryopoiesis, or its receptor Mpl, although the fold-increase in HSC, CD41⁺HSC, megakaryocyte and platelet populations in Mpl^−/−Cul5^fl/fl VavCre and Thpo^−/−Cul5^fl/fl vavCre mice over their respective Cul5^fl/fl controls, were at least as great as those in Cul5^fl/fl VavCre mice themselves. In contrast, in the absence of both the β_c and β_IL3 receptor chains, in which IL-3, GM-CSF and IL-5 action is ablated, the numbers of megakaryocytes and platelets were reduced in Cul5^fl/flVavCre mice, but the excess numbers of HSC and CD41⁺HSC were not altered. These observations are consistent with a model in which Cul5 deficiency manifests at multiple stages of megakaryopoiesis. In this model, the absence of Cul5 releases mechanisms that act on the available stem cell pool, and the presence or absence of TPO/Mpl dictates the size of the pool that develops, consistent with the well-established role in normal HSC production.^36,37 In the absence of Cul5, there is a substantial, if not complete, TPO/Mpl-independent contribution. β_c and/or β_IL3 signaling appears to have little or no role in regulation at the HSC level, but in the absence of Cul5, primarily contributes to enhanced production of megakaryocytes and platelets from the excess megakaryocyte-committed HSCs.

While we cannot exclude contributions from IL-5 or GM-CSF, the former has not previously been linked to megakaryopoiesis and while GM-CSF has modest reported actions in vitro, little evidence exists for an in vivo role.³² Moreover, GM-CSF did not significantly stimulate megakaryocyte colony formation from Cul5^fl/flVavCre BM, nor excess proliferation of Cul5-deficient LSK cells in culture. In contrast, IL-3 showed markedly enhanced activity on Cul5^fl/flVavCre cells in these assays and STAT5 phosphorylation was prolonged in IL-3-stimulated cells in the absence of Cul5. Together these observations support a significant contribution from IL-3 in the enhanced megakaryopoiesis of Cul5 deficiency. Moreover, the increased numbers of neutrophils, monocytes and eosinophils in older Cul5^fl/flVavCre mice may also result from deregulated β_c and/or β_IL3 signaling, reflecting enhanced effects of the known actions of IL-3, GM-CSF and IL-5 on production of these cell lineages.³⁸ 

The contribution of multiple mechanisms to the phenotypes observed in Cul5-deficient mice is perhaps to be expected considering that the loss of Cul5 should disable dozens of CRL complexes with different individual SOCS-box proteins and different biological actions. To date, no SOCS-box protein and thus specific CRL, has been associated with regulation of megakaryopoiesis in vivo: numerous models of targeted disruption have been generated and no such phenotype has been reported. Our observations now clearly establish an indispensable negative regulatory role in megakaryopoiesis: identification of the specific SOCS-box protein(s) and substrates, including potential roles in TPO and IL-3 signaling, await further studies.

In summary, our data reveal a previously undescribed mechanism of negative regulation of megakaryocyte-committed stem cells, disruption of which results in unrestrained megakaryopoiesis and thrombocytosis and ultimately a broader myeloproliferation. Cul5-deficient megakaryopoiesis is distinctive in that it is largely independent of TPO/Mpl and involves signaling via the β_c and/or β_IL3 receptors, with evidence of deregulated responses to IL-3. This process is independent of the IFNARI interferon receptor, previously implicated in inflammation-induced activation of megakaryocyte-committed stem cells. Potential contributions of megakaryocyte-committed stem cells in pathologies such as the myeloproliferative diseases are emerging.³⁴ Any disruption of Cul5-coordinated regulation of hematopoiesis in disease awaits further investigation, as does any potential strategy for targeting this process in therapy.

The authors thank Janelle Lochland, Sandra Mifsud, and Jason Corbin for skilled technical assistance and the Walter and Eliza Hall Institute of Medical Research (WEHI) Bioservices, Histology and Flow Cytometry Laboratories for expert support. This work was performed in part at the Materials Characterisation and Fabrication Platform at the University of Melbourne and the Victorian Node of the Australian National Fabrication Facility.

This work was supported by the Australian National Health and Medical Research Council (project grant 1062840, program grant 1113577, investigator grant 1173342 to W.S.A.; Early Career Fellowship [grant 461284] to C.A.W.; and Independent Research Institutes Infrastructure Support Scheme grant), a Medical Research Future Fund grant (MRF2008972, to J.J.B.), the Australian Cancer Research Foundation, and the Victorian State Government Operational Infrastructure Support scheme. K.B. received a fellowship (grant 341020624) from the Deutsche Forschungsgemeinschaft. A.E.W. received a Research Training Program Scholarship from the University of Melbourne and the Wendy Dowsett Scholarship from WEHI. D.J.H. is supported by The Lorenzo and Pamela Galli Chair in Medical Biology. C.A.d.G. is supported by the DHB Foundation. The generation of CRISPR/Cas9 modified mice used in this study was supported by Phenomics Australia and the Australian Government through the National Collaborative Research Infrastructure Strategy program.

Contribution: M.K., K.B., C.D.H., E.M.V., C.A.W., and L.D.R., designed and performed experiments and analyzed data; A.J.K. and M.J.H. designed and generated CRISPR/Cas9 modified mice; C.A.d.G. and A.E.W. designed, performed and analyzed single-cell RNA-sequence data; J.Y., L.F.D., and S.J.E.-C. designed, performed, and analyzed proteomics data; W.S.A. designed experiments, analyzed data, and led preparation of the manuscript; D.J.H., J.J.B., and N.A.N. contributed to experimental design and data interpretation; and all authors contributed to preparation of the manuscript.

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

Correspondence: Warren S. Alexander, Blood Cells and Blood Cancer Division, The Walter and Eliza Hall Institute of Medical Research, 1G Royal Parade, Parkville, VIC, 3052, Australia; email: alexandw@wehi.edu.au.