Annexin A2 and Ahnak control cortical NuMA-dynein localization and mitotic spindle orientation.

Proper mitotic spindle orientation depends on the correct anchorage of astral microtubules to the cortex. It relies on the remodeling of the cell cortex, a process not fully understood. Annexin A2 (Anx2; also known as ANXA2) is a protein known to be involved in cortical domain remodeling. Here, we report that in HeLa cell early mitosis, Anx2 recruits the scaffold protein Ahnak at the cell cortex facing spindle poles, and the distribution of both proteins is controlled by cell adhesion. Depletion of either protein or impaired cortical Ahnak localization result in delayed anaphase onset and unstable spindle anchoring, which leads to altered spindle orientation. We find that Ahnak is present in a complex with dynein-dynactin. Furthermore, Ahnak and Anx2 are required for correct dynein and NuMA (also known as NUMA1) cortical localization and dynamics. We propose that the Ahnak-Anx2 complex influences the cortical organization of the astral microtubule-anchoring complex, and thereby mitotic spindle positioning in human cells. This article has an associated First Person interview with the first author of the paper.

The spindle assembly checkpoint and the spatial activation of Polo kinase determine the duration of cell division and prevent tumor formation.

The maintenance of a restricted pool of asymmetrically dividing stem cells is essential for tissue homeostasis. This process requires the control of mitotic progression that ensures the accurate chromosome segregation. In addition, this event is coupled to the asymmetric distribution of cell fate determinants in order to prevent stem cell amplification. How this coupling is regulated remains poorly described. Here, using asymmetrically dividing Drosophila neural stem cells (NSCs), we show that Polo kinase activity levels determine timely Cyclin B degradation and mitotic progression independent of the spindle assembly checkpoint (SAC). This event is mediated by the direct phosphorylation of Polo kinase by Aurora A at spindle poles and Aurora B kinases at centromeres. Furthermore, we show that Aurora A-dependent activation of Polo is the major event that promotes NSC polarization and together with the SAC prevents brain tumor growth. Altogether, our results show that an Aurora/Polo kinase module couples NSC mitotic progression and polarization for tissue homeostasis.

Peripheral astral microtubules ensure asymmetric furrow positioning in neural stem cells.

Neuroblast division is characterized by asymmetric positioning of the cleavage furrow, resulting in a large difference in size between the future daughter cells. In animal cells, furrow placement and assembly are governed by centralspindlin that accumulates at the equatorial cell cortex of the future cleavage site and at the spindle midzone. In neuroblasts, these two centralspindlin populations are spatially and temporally separated. A leading pool is located at the basal cleavage site and a second pool accumulates at the midzone before traveling to the cleavage site. The cortical centralspindlin population requires peripheral astral microtubules and the chromosome passenger complex for efficient recruitment. Loss of this pool does not prevent cytokinesis but enhances centralspindlin signaling at the midzone, leading to equatorial furrow repositioning and decreased size asymmetry. These data show that basal furrow positioning in neuroblasts results from a competition between different centralspindlin pools in which the cortical pool is dominant.

Drosophila Tubulin-Specific Chaperone E Recruits Tubulin around Chromatin to Promote Mitotic Spindle Assembly.

Proper assembly of mitotic spindles requires microtubule nucleation not only at the centrosomes but also around chromatin. In this study, we found that the Drosophila tubulin-specific chaperone dTBCE is required for the enrichment of tubulin in the nuclear space after nuclear envelope breakdown and for subsequent promotion of spindle microtubule nucleation. These events depend on the CAP-Gly motif found in dTBCE and are regulated by Ran and lamin proteins. Our data suggest that during early mitosis, dTBCE and nuclear pore proteins become enriched in the nucleus, where they interact with the Ran GTPase to promote dynamic tubulin enrichment. We propose that this novel mechanism enhances microtubule nucleation around chromatin, thereby facilitating mitotic spindle assembly.

Dual control of Kinesin-1 recruitment to microtubules by Ensconsin in Drosophila neuroblasts and oocytes.

Drosophila Ensconsin (also known as MAP7) controls spindle length, centrosome separation in brain neuroblasts (NBs) and asymmetric transport in oocytes. The control of spindle length by Ensconsin is Kinesin-1 independent but centrosome separation and oocyte transport require targeting of Kinesin-1 to microtubules by Ensconsin. However, the molecular mechanism used for this targeting remains unclear. Ensconsin contains a microtubule (MT)-binding domain (MBD) and a Kinesin-binding domain (KBD). Rescue experiments show that only full-length Ensconsin restores the spindle length phenotype. KBD expression rescues ensc centrosome separation defects in NBs, but not the fast oocyte streaming and the localization of Staufen and Gurken. Interestingly, the KBD can stimulate Kinesin-1 targeting to MTs in vivo and in vitro We propose that a KBD and Kinesin-1 complex is a minimal activation module that increases Kinesin-1 affinity for MTs. Addition of the MBD present in full-length Ensconsin allows this process to occur directly on the MT and triggers higher Kinesin-1 targeting. This dual regulation by Ensconsin is essential for optimal Kinesin-1 targeting to MTs in oocytes, but not in NBs, illustrating the importance of adapting Kinesin-1 recruitment to different biological contexts.

Spindle assembly checkpoint inactivation fails to suppress neuroblast tumour formation in aurA mutant Drosophila.

Tissue homeostasis requires accurate control of cell proliferation, differentiation and chromosome segregation. Drosophila sas-4 and aurA mutants present brain tumours with extra neuroblasts (NBs), defective mitotic spindle assembly and delayed mitosis due to activation of the spindle assembly checkpoint (SAC). Here we inactivate the SAC in aurA and sas-4 mutants to determine whether the generation of aneuploidy compromises NB proliferation. Inactivation of the SAC in the sas-4 mutant impairs NB proliferation and disrupts euploidy. By contrast, disrupting the SAC in the aurA mutant does not prevent NB amplification, tumour formation or chromosome segregation. The monitoring of Mad2 and cyclin B dynamics in live aurA NBs reveals that SAC satisfaction is not coupled to cyclin B degradation. Thus, the NBs of aurA mutants present delayed mitosis, with accurate chromosome segregation occurring in a SAC-independent manner. We report here the existence of an Aurora A-dependent mechanism promoting efficient, timed cyclin B degradation.

Ensconsin/Map7 promotes microtubule growth and centrosome separation in Drosophila neural stem cells.

The mitotic spindle is crucial to achieve segregation of sister chromatids. To identify new mitotic spindle assembly regulators, we isolated 855 microtubule-associated proteins (MAPs) from Drosophila melanogaster mitotic or interphasic embryos. Using RNAi, we screened 96 poorly characterized genes in the Drosophila central nervous system to establish their possible role during spindle assembly. We found that Ensconsin/MAP7 mutant neuroblasts display shorter metaphase spindles, a defect caused by a reduced microtubule polymerization rate and enhanced by centrosome ablation. In agreement with a direct effect in regulating spindle length, Ensconsin overexpression triggered an increase in spindle length in S2 cells, whereas purified Ensconsin stimulated microtubule polymerization in vitro. Interestingly, ensc-null mutant flies also display defective centrosome separation and positioning during interphase, a phenotype also detected in kinesin-1 mutants. Collectively, our results suggest that Ensconsin cooperates with its binding partner Kinesin-1 during interphase to trigger centrosome separation. In addition, Ensconsin promotes microtubule polymerization during mitosis to control spindle length independent of Kinesin-1.

Phosphorylated ERK5/BMK1 transiently accumulates within division spindles in mouse oocytes and preimplantation embryos.

MAP kinases of the ERK family play important roles in oocyte maturation, fertilization, and early embryo development. The role of the signaling pathway involving ERK5 MAP kinase during meiotic and mitotic M-phase of the cell cycle is not well known. Here, we studied the localization of the phosphorylated, and thus potentially activated, form of ERK5 in mouse maturing oocytes and mitotically dividing early embryos. We show that phosphorylation/dephosphorylation, i.e. likely activation/inactivation of ERK5, correlates with M-phase progression. Phosphorylated form of ERK5 accumulates in division spindle of both meiotic and mitotic cells, and precisely co-localizes with spindle microtubules at metaphase. This localization changes drastically in the anaphase, when phospho-ERK5 completely disappears from microtubules and transits to the cytoplasmic granular, vesicle-like structures. In telophase oocytes it becomes incorporated into the midbody. Dynamic changes in the localization of phospho-ERK5 suggests that it may play an important role both in meiotic and mitotic division.

CDK11(p58) is required for centriole duplication and Plk4 recruitment to mitotic centrosomes.

BACKGROUND: CDK11(p58) is a mitotic protein kinase, which has been shown to be required for different mitotic events such as centrosome maturation, chromatid cohesion and cytokinesis. METHODOLOGY/PRINCIPAL FINDINGS: In addition to these previously described roles, our study shows that CDK11(p58) inhibition induces a failure in the centriole duplication process in different human cell lines. We propose that this effect is mediated by the defective centrosomal recruitment of proteins at the onset of mitosis. Indeed, Plk4 protein kinase and the centrosomal protein Cep192, which are key components of the centriole duplication machinery, showed reduced levels at centrosomes of mitotic CDK11-depleted cells. CDK11(p58), which accumulates only in the vicinity of mitotic centrosomes, directly interacts with the centriole-associated protein kinase Plk4 that regulates centriole number in cells. In addition, we show that centriole from CDK11 defective cells are not able to be over duplicated following Plk4 overexpression. CONCLUSION/SIGNIFICANCE: We thus propose that CDK11 is required for centriole duplication by two non-mutually-exclusive mechanisms. On one hand, the observed duplication defect could be caused indirectly by a failure of the centrosome to fully maturate during mitosis. On the other hand, CDK11(p58) could also directly regulate key centriole components such as Plk4 during mitosis to trigger essential mitotic centriole modifications, required for centriole duplication during subsequent interphase.

Aurora A contributes to p150(glued) phosphorylation and function during mitosis.

Aurora A is a spindle pole-associated protein kinase required for mitotic spindle assembly and chromosome segregation. In this study, we show that Drosophila melanogaster aurora A phosphorylates the dynactin subunit p150(glued) on sites required for its association with the mitotic spindle. Dynactin strongly accumulates on microtubules during prophase but disappears as soon as the nuclear envelope breaks down, suggesting that its spindle localization is tightly regulated. If aurora A's function is compromised, dynactin and dynein become enriched on mitotic spindle microtubules. Phosphorylation sites are localized within the conserved microtubule-binding domain (MBD) of the p150(glued). Although wild-type p150(glued) binds weakly to spindle microtubules, a variant that can no longer be phosphorylated by aurora A remains associated with spindle microtubules and fails to rescue depletion of endogenous p150(glued). Our results suggest that aurora A kinase participates in vivo to the phosphoregulation of the p150(glued) MBD to limit the microtubule binding of the dynein-dynactin complex and thus regulates spindle assembly.

Proteomic screen for potential regulators of M-phase entry and quality of meiotic resumption in Xenopus laevis oocytes.

The quality of oocytes depends largely on the capacity to resume meiotic maturation. In Xenopus laevis, only fully grown oocytes react to progesterone stimulation by resumption of meiotic maturation associated with the entry into the meiotic M-phase. Proteins involved in this process are poorly known. To identify novel proteins regulating M-phase entry, we performed a differential proteomic screen. We compared proteomes of fully grown stage VI oocytes characterized as poorly or highly responsive to progesterone treatment. The comparison of 2-D gels allowed us to identify several spots including two specifically present in highly responsive oocytes and two specifically present in poorly responsive ones. By mass spectrometry we identified the two proteins specifically present in highly responsive oocytes as inosine 5'monophosphate cyclohydrolase and YjgF homologues, and the two specifically present in poorly responsive oocytes as elongation factor 2 (EF2) and S-adenosyl-L-homocysteine hydrolase (SAHH). The proteins specifically expressed in highly responsive oocytes may participate in the stimulation of meiotic maturation and M-phase entry, while the proteins specifically present in poorly maturing oocytes may participate in the inhibition of meiotic resumption.

EP45 accumulates in growing Xenopus laevis oocytes and has oocyte-maturation-enhancing activity involved in oocyte quality.

The capacity of oocytes to fully support meiotic maturation develops gradually during oocyte growth. Growing oocytes accumulate proteins and mRNAs required for this process. However, little is known about the identity of these factors. We performed a differential proteomic screen comparing the proteomes of growing stage-IV oocytes, which do not undergo meiotic maturation in response to progesterone, with fully grown stage-VI ones, which do. In 2D gels of stage-VI oocytes, we identified a group of four protein spots as EP45 (estrogen-regulated protein 45 kDa), which belongs to the family of serine protease inhibitors and is also known as Seryp or pNiXa. Western blot analysis after mono- and bi-dimensional electrophoreses confirmed the accumulation of certain forms of this protein in oocytes between stages IV and VI. EP45 mRNA was not detectable in oocytes or ovaries, but was expressed in the liver. A low-mobility isoform of EP45 was detected in liver and blood, whereas two (occasionally three or four) higher-mobility isoforms were found exclusively in oocytes, suggesting that liver-synthesized protein is taken up by oocytes from the blood and rapidly modified. Alone, overexpression of RNA encoding either full-length or N-terminally truncated protein had no effect on meiotic resumption in stage-IV or -VI oocytes. However, in oocytes moderately reacting to low doses of progesterone, it significantly enhanced germinal-vesicle breakdown, showing a novel and unsuspected activity of this protein. Thus, EP45 accumulates in growing oocytes through uptake from the blood and has the capacity to act as an 'oocyte-maturation enhancer' ('Omen').

Complex relationship between TCTP, microtubules and actin microfilaments regulates cell shape in normal and cancer cells.

Translationally controlled tumor-associated protein (TCTP) is a ubiquitous and highly conserved protein implicated in cancers. Here, we demonstrate that interactions of TCTP with microtubules (MTs) are functionally important but indirect, and we reveal novel interaction of TCTP with the actin cytoskeleton. Firstly, immunofluorescence in Xenopus XL2 cells revealed cytoplasmic fibers stained with TCTP but not with tubulin antibodies, as well as MTs free of TCTP. Furthermore, TCTP localized to a subset of actin-rich fibers in migrating cells. Secondly, Xenopus laevis TCTP did not affect in vitro assembly/disassembly of MTs and lacked MT-binding affinity both in pull-down assays and in cell-free extracts. Although TCTP also failed to bind to purified filamentous actin (F-actin), it was associated with microfilaments in cell-free extracts. Thirdly, TCTP concentrated in mitotic spindle did not colocalize with MTs and was easily dissociated from these structures except at the poles. Finally, RNA interference knockdown of TCTP in XL2 and HeLa cells provoked drastic, MT-dependent shape change. These data show that although TCTP interacts with MTs, it does not behave as classic MT-associated protein. Our evidence for an association of TCTP with F-actin structures, and for an involvement in cell shape regulation, implicates this protein in integrating cytoskeletal interactions both in interphase and mitosis providing a new avenue to fully understand the role of TCTP.

Differential proteomic screen to evidence proteins ubiquitinated upon mitotic exit in cell-free extract of Xenopus laevis embryos.

Post-translational modification of proteins via ubiquitination plays a crucial role in numerous vital functions of the cell. Polyubiquitination is one of the key regulatory processes involved in regulation of mitotic progression. Here we describe a differential proteomic screen dedicated to identification of novel proteins ubiquitinated upon mitotic exit in cell-free extract of Xenopus laevis embryo. Mutated recombinant His6-tagged ubiquitin (Ubi (K48R)) was added to mitotic extract from which we purified conjugated proteins, as well as associated proteins in nondenaturing conditions by cobalt affinity chromatography. Proteins eluted from Ubi (K48R) supplemented and control extracts were compared by LC-MS/MS analysis after monodimensional SDS-PAGE. A total of 144 proteins potentially ubiquitinated or associated with them were identified. Forty-one percent of these proteins were shown to be involved in ubiquitination and/or proteasomal degradation pathway confirming the specificity of the screen. Twelve proteins, among them ubiquitin itself, were shown to carry a "GG" or "LRGG" remnant tag indicating their direct ubiquitination. Interestingly, sequence analysis of ubiquitinated substrates carrying these tags indicated that in Xenopus cell-free embryo extract supplemented with Ubi (K48R) the majority of polyubiquitination occurred through lysine-11 specific ubiquitin chain polymerization. The potential interest in this atypical form of ubiquitination as well as usefulness of our method in analyzing atypical polyubiquitin species is discussed.

Temporal regulation of embryonic M-phases.

Temporal regulation of M-phases of the cell cycle requires precise molecular mechanisms that differ among different cells. This variable regulation is particularly clear during embryonic divisions. The first embryonic mitosis in the mouse lasts twice as long as the second one. In other species studied so far (C. elegans, Sphaerechinus granularis, Xenopus laevis), the first mitosis is also longer than the second, yet the prolongation is less pronounced than in the mouse. We have found recently that the mechanisms prolonging the first embryonic M-phase differ in the mouse and in Xenopus embryos. In the mouse, the metaphase of the first mitosis is specifically prolonged by the unknown mechanism acting similarly to the CSF present in oocytes arrested in the second meiotic division. In Xenopus, higher levels of cyclins B participate in the M-phase prolongation, however, without any cell cycle arrest. In Xenopus embryo cell-free extracts, the inactivation of the major M-phase factor, MPF, depends directly on dissociation of cyclin B from CDK1 subunit and not on cyclin B degradation as was thought before. In search for other mitotic proteins behaving in a similar way as cyclins B we made two complementary proteomic screens dedicated to identifying proteins ubiquitinated and degraded by the proteasome upon the first embryonic mitosis in Xenopus laevis. The first screen yielded 175 proteins. To validate our strategy we are verifying now which of them are really ubiquitinated. In the second one, we identified 9 novel proteins potentially degraded via the proteasome. Among them, TCTP (Translationally Controlled Tumor Protein), a 23-kDa protein, was shown to be partially degraded during mitosis (as well as during meiotic exit). We characterized the expression and the role of this protein in Xenopus, mouse and human somatic cells, Xenopus and mouse oocytes and embryos. TCTP is a mitotic spindle protein positively regulating cellular proliferation. Analysis of other candidates is in progress.

Temporal regulation of the first mitosis in Xenopus and mouse embryos.

Cell cycle regulation in Eukaryotes is based on common molecular actors and mechanisms. However, the canonical cell cycle is modified in certain cells. Such modifications play a key role in oocyte maturation and embryonic development. They can be achieved either by introduction of new components, pathways, substrates, changed interactions between them, or by elimination of some factors inherited by the cells from previous developmental stages. Here we discuss a particular temporal regulation of the first embryonic M-phase of Xenopus and mouse embryos. These two examples help to understand better the general regulation of M-phase of the cell cycle.

Cyclin B2/cyclin-dependent kinase1 dissociation precedes CDK1 Thr-161 dephosphorylation upon M-phase promoting factor inactivation in Xenopus laevis cell-free extract.

Cyclin-dependent kinase 1 (CDK1) is the enzymatic subunit of M-phase Promoting Factor (MPF). It is positively regulated by phosphorylation on Thr-161 and association with a cyclin B molecule. The role of Thr-161 dephosphorylation upon MPF inactivation remains unclear; nevertheless, degradation of cyclin B is thought to be a direct cause of MPF inactivation. However, MPF inactivation actually precedes cyclin B degradation in Xenopus cell-free extracts. Here we study in details the temporal relationship between histone H1 kinase (reflecting MPF activity) inactivation, Thr-161 dephosphorylation, CDK1-cyclin B2 dissociation and cyclin B2 proteolysis in such extracts. We show an asynchrony between inactivation of histone H1 kinase and degradation of cyclin B2. CDK1 dephosphorylation on Thr 161 is an even later event than cyclin B2 degradation, reinforcing the hypothesis that cyclin B dissociation from CDK1 is the key event inactivating MPF. Cyclins synthesized along with MPF inactivation could deliver shortly living active MPF molecules, potentially increasing the asynchrony between histone H1 kinase inactivation and cyclin B2 degradation. We confirm this by showing that in the absence of protein synthesis, such a tendency is lower, but nevertheless, still detectable. Finally, to characterise better CDK1/cyclin B dissociation, we show that CDK1 begins to dissociate from cyclin B2 before the very beginning of cyclin B2 degradation and that the diminution in CDK1-associated cyclin B2 is faster than the decline of its total pool. Thus, neither cyclin B2 degradation nor Thr-161 dephosphorylation participates directly in CDK1 inactivation as measured by histone H1 kinase decline upon the exit from mitotic M-phase in Xenopus embryo extract.

Absence of reciprocal feedback between MPF and ERK2 MAP kinase in mitotic Xenopus laevis embryo cell-free extract.

MPF and MAP kinase ERK2 are two major M-phase kinases. They interact with each other in a complex way during meiotic maturation of Xenopus laevis oocytes. Here we study their interrelationship during first mitosis in X. laevis embryo cell-free extract perturbing the polyubiquitination pathway as a tool. Recombinant ubiquitin K48R (Ub-K48R) mutant protein arrests mitotic cyclin B degradation in the extract. This results in both increased accumulation of phosphorylated form of cyclin B2 and MPF activity as well as mitotic phosphorylation of its substrates. Ub-K48R also increased the mitotic phosphorylation of ERK2. Simultaneous addition of Ub-K48R and the proteasome inhibitor MG 132 strengthened and further prolonged MPF activity, MCM4 phosphorylation and accumulation of phosphorylated forms of cyclin B2. ERK2 phosphorylation levels increased and persisted longer than upon action of Ub-K48R alone. This shows a synergistic effect of inhibition of two different steps of ubiquitin-proteasome pathway on MPF activity and mitotic phosphorylation and ubiquitination of specific M-phase proteins. On the other hand, complete inhibition of ERK2 activation using U0126 had no effect either on MPF activity or on MCM4 phosphorylation either in control or in Ub-K48R-supplemented extracts. Experimental reduction of MPF activity by addition of recombinant p21(Cip) protein resulted in significant reduction of ERK2 phosphorylation. Thus, the reciprocal feedback observed between MPF and ERK2 in meiosis is not observed during mitotic M-phase in cell-free Xenopus embryo extracts. ERK2 phosphorylation is regulated by the levels of MPF activity, however no influence of ERK2 on MPF activity could be detected. These results show a fundamental difference in the relationship between the two major M-phase kinases in meiotic and mitotic cell cycle.

Cyclin B dissociation from CDK1 precedes its degradation upon MPF inactivation in mitotic extracts of Xenopus laevis embryos.

Cyclin B is a regulatory subunit of CDK1 within MPF complex. Degradation of cyclin B via ubiquitin-proteasome pathway seemed to be absolutely required for the M-phase exit. However, inhibition of the proteasome proteolytic activity upon the exit from the meiotic metaphase II-arrest in Xenopus cell-free extract revealed that the proteasome-dependent dissociation of cyclin B from CDK1 is sufficient to inactivate MPF without cyclin B degradation. In this study we analyze whether the same mechanism operates during the exit from mitotic M-phase. We show in Xenopus cell-free extract undergoing the first or the second embryonic mitosis that CDK1 oscillations are not affected by proteasome inhibition with MG132 or ALLN despite effective inhibition of cyclins B degradation. The majority of cyclins B1 and B2 surviving CDK1 inactivation is CDK-free and cyclin B2 becomes resistant to phosphatase lambda dephosphorylation. The pool of cyclins B remaining after CDK1 inactivation in the presence of MG132 is mitotically inert, while exogenous or newly synthesized cyclin B activates CDK1. This suggests that cyclins B remain sequestered within the proteasome upon MPF inactivation in the presence of MG132. Comparison of the dynamics of the decline of total and CDK-bound pools of cyclins B1, B2 and B4 upon mitotic exit in absence of protein synthesis reveals that CDK-bound cyclins B diminish clearly faster. Our results thus show that cyclin B dissociation from CDK1 precedes cyclins B degradation upon CDK1 inactivation in mitotic embryo extracts and that proteasome proteolytic activity is dispensable for both activation and inactivation of CDK1 in such extracts.

STAT5 acts as a repressor to regulate early embryonic erythropoiesis.

STAT5 regulates definitive (adult stage) erythropoiesis through its ability to transduce signals from the erythropoietin receptor. A function for STAT-dependent signaling during primitive (embryonic) erythropoiesis has not been analyzed. We tested this in the Xenopus system, because STAT5 is expressed at the right time and place to regulate development of the embryonic primitive ventral blood island. Depletion of STAT5 activity results in delayed accumulation of the first globin-expressing cells, indicating that the gene does regulate primitive erythropoiesis. Our results suggest that in this context STAT5 functions as a repressor, since forced expression of an activator isoform blocks erythropoiesis, while embryos expressing a repressor isoform develop normally. The erythroid phenotype caused by the activator isoform of STAT5 resembles that caused by overexpression of fibroblast growth factor (FGF). We show that STAT5 isoforms can function epistatic to FGF and can be phosphorylated in response to hyperactivated FGF signaling in Xenopus embryos. Therefore, our data indicate that STAT5 functions in both primitive and definitive erythropoiesis, but by different mechanisms.