Erythroid homoeostasis is primarily controlled by Epo (erythropoietin) receptor signalling; however, the Lyn tyrosine kinase plays an important subsidiary role in regulating the erythroid compartment. Nonetheless, specific erythroid pathways that require Lyn activity and their biological significance remain unclear. To address this, we asked what consequence loss of Lyn had on the ex vivo expansion and maturation of splenic erythroid progenitors and Epo receptor signalling. Pharmacological inhibition of Lyn with PP2 inhibited the survival of terminally differentiated erythroblasts. Less committed erythroid progenitors expanded well, whereas early splenic Lyn−/− erythroblasts had attenuated ex vivo expansion, and late stage Lyn−/− erythroblasts were retarded in completing morphological maturation ex vivo. Furthermore, immortalized Lyn−/− erythroblasts were slower growing, less viable and inhibited in their differentiation. Signalling studies showed that Lyn was required for both positive GAB2/Akt/FoxO3 (forkhead box O3) survival signals as well as negative feedback of JAK2 (Janus kinase 2)/STAT5 (signal transducer and activator of transcription 5) and ERK1/2 (extracellular-signal-regulated kinase 1/2) signals via SHP-1 (Src homology 2 domain-containing protein tyrosine phosphatase 1). During differentiation, Lyn controls survival and cell cycle exit as demonstrated by reduced STAT5 and FoxO3/GSKα/β (glycogen synthase kinase α/β) phosphorylation and diminished p27Kip1 induction in Lyn-deficient erythroblasts. Lyn deficiency alters the balance of pro- and anti-apoptotic molecules (BAD and BclXL), thereby reducing survival and preventing cell cycle exit. Consequently, Lyn facilitates normal erythrocyte production by influencing different stages of erythroid progenitor expansion, and mature cell development and survival signalling.
Epo (erythropoietin) is the primary regulator of committed erythroid progenitors acting through the Epo-R (Epo receptor) to trigger signalling and the activation of JAK (Janus kinase)/STAT (signal transducer and activator of transcription), Ras/Raf/MAPK (mitogen-activated protein kinase) and PI3K (phosphoinositide 3-kinase)/Akt pathways [1–5]. The tyrosine kinase JAK2 is recognised as the primary initiator of Epo-R signalling , whereas the SFK (Src family tyrosine kinase) Lyn has been implicated as a key ancillary kinase [6–11].
Lyn plays an important role in maintaining erythroid homoeostasis as highlighted by the perturbed erythropoiesis of Lyn−/− mice [10,11], namely elevated extramedullary erythropoiesis. Furthermore, Lyn−/− bone marrow erythroblasts are attenuated in their expansion capacity and late stage survival . Previous in vitro studies using the J2E erythroid cell line highlighted a role for Lyn in Epo-R signalling [7,12] which appeared equivalent to that in non-immortalized erythroid cells [7,10,12]. Taken together, these studies revealed a role for Lyn in erythroid differentiation and survival in response to Epo [7,9–12].
Although still not fully elucidated, Lyn appears to be involved in several Epo-R dependent and independent signalling events in erythroid cells, including ERK1/2 (extracellular-signal-regulated kinase 1/2) regulation [12,13] and phosphorylation of STAT5 [7,8]. Both primary and immortalized erythroid cells displayed a requirement for Lyn in the maintenance of GATA-1 and EKLF (erythroid Kruppel-like factor) erythroid transcription factor levels. Loss of Lyn impeded differentiation in response to Epo [10,12] and, conversely, transient overexpression of Lyn enhanced differentiation [7,12]. Lyn is also important for early erythroid progenitor expansion through SCF (stem cell factor) signalling [14,15], as well as late stage development and mature erythrocyte form and function [9,16,17].
Epo, Epo-R and JAK2 are essential for the development of a definitive erythroid compartment [18–20]. Although Lyn−/− mice are capable of definitive erythropoiesis, they do possess an underlying erythroid defect resulting in activation of compensatory stress erythropoiesis [9–11]. Consequently Lyn is clearly required for normal erythropoiesis and influences Epo-R signalling [10,11]. However, the precise pathways requiring Lyn protein and/or its activation and their biological roles remain unclear.
To address these important questions, we analysed the ex vivo expansion/differentiation capabilities of splenic erythroid cells and investigated Epo-R signalling of immortalized fetal liver erythroid progenitors from Lyn−/− mice. Direct chemical inhibition of Lyn activity impeded differentiation/late-stage survival, but not expansion of committed progenitors ex vivo. A null mutation in Lyn reduced early committed erythroid, but not uncommitted progenitor, expansion and also retarded maturation in ex vivo cultures, with reduced BclXL and persistent ERK1/2 activation. Importantly, Lyn deficiency had a major impact on Epo-R signalling in erythroid progenitors, most notably affecting pro-survival GAB2/Akt/FoxO3 (forkhead box O3) signals and negative feedback via SHP-1 (Src homology 2 domain-containing protein tyrosine phosphatase 1) of the JAK2 and ERK1/2 pathways. Taken together, these biological and biochemical experiments suggest that Lyn acts on both early and late stage erythroblasts, and is critical for mediating correct temporal dynamics of Epo-R signalling and maturation, promoting pro-survival signalling as well as negative feedback of JAK2 and restricting ERK1/2 activity.
MATERIALS AND METHODS
Mice, cell morphology and anaemia induction
C57BL/6 and Lyn−/−(on a C57BL/6 background) mice [21,22] were used at day 12.5–13.5 of embryonic development for the generation of erythroid cell lines, and as 8–15-week-old adults for isolation of splenic erythroid progenitors. All experiments were performed in accordance with National Health and Medical Research Council of Australia guidelines for animal experimentation, with approval from the Animal Ethics Committees of Animal Resource Centre, Murdoch, and Royal Perth Hospital, Perth. Cell morphology was examined microscopically, following cytocentrifugation and Wright’s-Giemsa staining with or without neutral benzidine staining for haemoglobin . Extramedullary erythropoiesis was induced by injection of PHZ (phenylhydrazine) (60 mg/kg of body weight) on two consecutive days, and spleens were collected 2 or 4 days after the last PHZ injection as described previously .
Ex vivo erythroid cell cultures
C57BL/6 and Lyn−/− mice were treated with PHZ to induce emergency erythropoiesis in the spleen, and splenic erythroblastic islands were isolated at 48 h or 96 h post-PHZ induction. Mature erythrocytes and haemoglobin-positive erythroblasts were removed by erythrocyte lysis, which depletes the CD71−/Ter119+ population (Supplementary Figures S1A and S1B at http://www.biochemj.org/bj/459/bj4590455add.htm). Preparations contained approximately 80% erythroid progenitors (erythroblasts) as assayed morphologically (Supplementary Figure S1A), were highly dependent on exogenous Epo for proliferation (Supplementary Figure S1C), constituted a minor proportion (~0.01%) of BFU-E (burst-forming unit-erythroid; Supplementary Figure S1D) and a significant proportion (~10%) of CFU-E (colony-forming unit-erythroid; Supplementary Figure S1E) progenitors. Cells were cultured in either maintenance or maintenance/differentiation medium. Maintenance medium contained complete StemPro-34 SFM (Life Technologies) supplemented with StemPro-34 nutrient supplement, 2 units/ml Epo (Janssen-Cilag) and 10 ng/ml SCF. Maintenance/differentiation medium consisted of maintenance medium supplemented with 50 ng/ml insulin (Novo Nordisk) and 20% FBS. Cells were analysed over 72 h of culture by morphological examination following staining with neutral benzidine and Wright’s-Giemsa, or by cell surface marker expression by flow cytometry. Cultures were also analysed by Western blotting. The tyrosine kinase chemical inhibitors PP2 (Merck Millipore) and JAKI (JAK2 inhibitor 1; Merck Millipore) were used at 2 μM and 1 μM final concentrations respectively. Control cells were treated with the same volume of DMSO vehicle.
Flow cytometry of single cell suspensions of spleen cells and cultured erythroid cell lines was employed to assess cell-surface expression of CD71, Ter119, CD44, CD117, CD4, CD8, ScaI and CD11b as detailed previously  using a BD FACS Aria II flow cytometer (Beckman Coulter) and fluorophore-conjugated antibodies (BD Pharmingen). Phosflow cell signalling analysis was performed essentially as described previously  using Alexa Fluor® 647–STAT5(pY694) (BD Pharmingen), cells were fixed at various time points in 1.5% paraformaldehyde followed by permeabilization in ice-cold methanol before Phosflow staining and analysis. Profiles were analysed using FlowJo (v9.5.2, Tree Star).
Erythroid cell lines
Immortalized erythroid cell lines were generated by exposing fetal liver progenitors (isolated from embryonic day 12.5) to the J2 retrovirus as described previously [27,28]. Individual clones (12 per genotype) were isolated and characterized for viability by eosin dye exclusion and haemoglobin production via benzidine staining in the presence or absence of different doses of Epo, as previously described . Cells were maintained in IMDM (Iscove's modified Dulbecco's medium; Life Technologies) supplemented with 10% FBS. Short-term Epo inductions (5 units/ml) were carried out on cells after 2 h of serum starvation . Differentiation experiments were performed on cells cultured in thyroid hormone-depleted FBS in the presence or absence of Epo (5 units/ml) . Colony formation of erythroid cell lines in the presence or absence of Epo (2 units/ml) was assessed using methylcellulose cultures as previously described .
Immunoblotting and immunoprecipitation
Cells were lysed in raft buffer [150 mM NaCl, 1% Igepal CA-630 (Sigma–Aldrich), 0.5% n-dodecyl-β-D-maltoside (Sigma–Aldrich), 0.2% octyl-D-glucoside (Sigma–Aldrich), 0.25% CHAPS (Signal–Aldrich), 20 mM Tris/HCl, pH 8.0, 1× Complete™ protease inhibitor cocktail (Roche), 2 mM benzamidine, 1 mM sodium orthovanadate, 1 mM EDTA, 1 mM EGTA, 10 mM NaF, 2.5 mM sodium pyrophosphate and 10 mM 2-glycerophosphate). Signalling studies were undertaken by immunoprecipitation and immunoblotting [30,31]. For immunoprecipitation, clarified cell lysates were incubated with antibodies against Epo-R (sc-697, Santa Cruz Biotechnology) for 2 h at 4°C, collected with Protein G–Sepharose beads for 16 h before washing, SDS/PAGE and analysis by immunoblotting. Anti-EKLF antibodies were a gift from Professor J.J. Bieker (Mount Sinai School of Medicine, New York, NY, U.S.A.) . Additional antibodies used for immunoblotting were anti-pY (PY100; 9411), anti-pERK1/2 (Thr202/Tyr204; 4376), anti-ERK1/2 (9102), anti-pSFK (Tyr416; 2101), anti-pLyn (Tyr507; 2731), anti-pSTAT5 (Tyr694; 9351), anti-pSTAT3 (Tyr705; 9145), anti-STAT3 (4904), anti-pp38MAPK (Thr180/Tyr182; 4631), anti-pAkt (Ser473; 4060), anti-Akt (9272), anti-pJAK2 (Tyr1007/Tyr1008; 3776), anti-JAK2 (3230), anti-pBAD (Ser136; 4366), anti-pBAD (Ser112; 9291), anti-BAD (9292), anti-pFoxO1/3a/4 (Thr24/Thr32/Thr28; 2599), anti-FoxO3 (2497), anti-pGSK (phospho-glycogen synthase kinase)-3α/β (Ser21/Ser9; 8566), anti-p27Kip1 (3688), anti-pSHP-1 (Tyr464; 8849), anti-SHP-1 (3759), anti-pSHP-2 (Tyr542; 3751), anti-pSHP-2 (Tyr580; 5431), anti-SHP-2 (3397), anti-pPLCγ1 (phospho-phospholipase Cγ1; Tyr783; 2821), anti-histone H3 (9715) (Cell Signaling Technology), anti-β-actin (AC-15), anti-14-3-3ζ (ab9063) (Abcam), anti-STAT5 (sc-1081), anti-Lyn (sc-15), anti-BclXL (sc-634) (Santa Cruz Biotechnology) and anti-phosphotyrosine (4G10) (Merck Millipore). Proteins were revealed using secondary antibodies coupled to horseradish peroxidase (GE Healthcare or Cell Signaling Technology) and detection by ECL (GE Healthcare) or with fluorescently-labelled secondary antibodies and an Odyssey scanner (LI-COR Biosciences).
PP2 inhibits ex vivo erythroid cell late stage survival
Since we have previously shown that Lyn deficiency leads to erythroid compartment defects [10,11] and Karur et al.  have shown expansion/development alterations to bone marrow erythroblast cultures, we address in the present study whether direct chemical inactivation of Lyn alters ex vivo erythropoiesis of mixed cell cultures, using erythroblastic islands isolated from spleens of C57BL/6 mice treated with PHZ. When cultured in medium proficient for erythroid expansion and differentiation (maintenance/differentiation medium), these cells proliferate as erythroblastic islands (erythroblasts surrounding a central macrophage, Supplementary Figures S1 and S2 at http://www.biochemj.org/bj/459/bj4590455add.htm) and mature into haemoglobin-containing enucleated reticulocytes (Figure 1A). Addition of JAKI to these cultures profoundly inhibited their Epo-responsive expansion capacity, whereas the SFK inhibitor PP2 showed no significant effect upon their capacity to expand (Figure 1B). Total erythroid cells expanded 3–4-fold in both control and PP2 treated cultures, whereas no significant expansion occurred in the presence of JAKI (Figure 1B). Transient expansion of highly proliferative erythroid progenitors (CD71high/Ter119low) was also unaffected by PP2, but absent in JAKI cultures (Figure 1C). However, PP2 did show a significant effect upon the integrity of reticulocytes produced in these ex vivo cultures (Figure 1A). Although haemoglobin-positive enucleated reticulocytes started to appear in PP2 cultures at 48 h as was seen in control cultures, they failed to be maintained by 72 h, with only ghost-like haemoglobin-negative cells present at this latter time. In JAKI-treated cultures, although cells did not expand, a subpopulation did undergo morphological maturation including condensation, haemoglobin production and enucleation, but the reticulocytes that formed by 48 h did not remain viable at 72 h (Figure 1A). Epo-induced STAT5 phosphorylation dynamics assayed by Phosflow staining were not significantly altered in PP2 cultures assayed over 120 min, whereas JAK2 inhibition significantly inhibited the activation of STAT5 (Figure 1D).
Inhibition of SFKs via PP2 inhibits ex vivo erythroid maturation/survival, but not expansion
Dynamics of erythroblast cell expansion in ex vivo cultures is altered by genetic ablation of Lyn
To determine whether Lyn was required for ex vivo erythroid progenitor expansion, we analysed splenic erythroblastic island cultures from Lyn-deficient and control mice. Mice were treated with PHZ to induce extramedullary erythropoiesis and spleen cells were isolated 4 days later. Previously we have shown that Lyn−/− mice display significantly elevated numbers of erythroid progenitors in the spleens of PHZ-treated mice, particularly at 1 and 2 days post-treatment . However, by day 4 after PHZ treatment, splenic erythroid progenitor parameters are equivalent in both Lyn−/− and Lyn+/+ mice, including CD71/Ter119/CD117 profiles (Supplementary Figure S2) and CFU-E/BFU-E numbers . Consequently, we utilized cells from day 4 post-PHZ treated spleens to compare Lyn+/+ and Lyn−/− erythroblasts cultured in erythroid maintenance medium. Although both Lyn−/− and Lyn+/+ cultures initially (24 h) expanded their CD71high/Ter119+ late-erythroblasts to similar degrees, by 48 h and 72 h, Lyn−/− cultures showed significantly elevated levels of CD71high/Ter119+ cells with a concomitant proportional decrease in more mature CD71low/Ter119+ cells (Figure 2A). Interestingly, we also observed significant differences in the expansion of earlier c-Kit (CD117)-positive erythroblasts (Figure 2B). The absence of Lyn diminished the expansion capacity of c-Kit-positive progenitors expressing high levels of the erythroid-specific marker Ter119 at 24, 48 and 72 h of culture, whereas c-Kit-positive cells co-expressing low levels of Ter119 were elevated by 72 h of culture (Figure 2B). Taken together, these results suggest that ablation of Lyn restricts the normal maturation capacity of erythroblasts, causing accumulation of erythroblasts with less mature cell surface phenotypes in ex vivo cultures.
Lyn deficiency alters erythroid progenitor expansion in ex vivo cultures
Lyn−/− erythroblasts have delayed morphological maturation ex vivo with reduced BclXL induction and persistent ERK1/2 activation
Splenic erythroblastic islands were isolated from PHZ-treated Lyn+/+ and Lyn−/− mice at day 4 post-PHZ treatment as detailed above. These cells were then cultured in maintenance/differentiation medium to examine their capacity to morphologically mature ex vivo. Significantly, Lyn−/− erythroblast cultures had delayed morphological maturation; with a significant percentage of erythroblasts still present at 72 h, whereas in control cultures most cells had matured into enucleated reticulocytes (Figures 3A and 3B). In control cultures, high levels of BclXL protein appear at 72 h as the erythroid cells enucleate into reticulocytes (Figure 3C). However, in Lyn−/− cultures, the level of BclXL is reduced compared with controls (Figure 3C). Furthermore, activated ERK1/2 is present in Lyn−/− cultures, which is not seen in those of Lyn+/+ cells (Figure 3C).
Erythroid progenitors from Lyn−/− mice have impeded differentiation ex vivo
Erythroid cells immortalized from Lyn−/− embryos have reduced viability and clonogenicity, extended doubling times and impeded capacity to differentiate in response to Epo
To determine the biochemical and molecular alterations caused by Lyn deficiency, cell lines (12 lines for each genotype) were generated from control and Lyn−/− mice by immortalization of erythroid cells from fetal liver with the J2 retrovirus . The morphology of the J2-Lyn−/− cell lines was typical of cells immortalized at the pro-erythroblast stage and similar to the lines generated from control mice (J2-WT) (Figure 4A). In addition, flow cytometric analyses revealed that these cell lines had similar cell surface marker profiles, expressing both CD71 (transferrin receptor) and CD44, and lacking CD117 (c-Kit), CD11b, CD4, CD8 and Ter119 (Figure 4B, and results not shown). A subpopulation of ~20% of J2-Lyn−/− cells was noted to have low CD44 expression. Multiple clones displayed very similar profiles and expressed equivalent levels of the erythroid transcription factors GATA-1 and EKLF (Figure 4C). Previously we found that adult Lyn−/− erythroblasts from PHZ-treated mice had low EKLF levels , we speculate that the fetal liver cells we have immortalized in the present study are at an early stage of erythroid development before a lack of Lyn has influenced the expression of EKLF. Taken together, these results suggest that the cell lines from control and Lyn−/− mice were immortalized at similar stages of maturation.
Immortalized erythroblasts from Lyn−/− mice have reduced viability and differentiation
Significantly, the J2-Lyn−/− lines displayed reduced clonogenicity, in both the number and size of colonies produced in methylcellulose cultures supplemented with Epo (2 units/ml), compared with control J2-WT lines (Figures 4D–4F). Furthermore, lines lacking Lyn also had slower rates of proliferation (doubling times were twice as long) and reduced viability in complete culture medium in the absence of Epo (Figures 4G and 4H). J2-Lyn−/− lines were also significantly impaired in their ability to differentiate (produce haemoglobin) in response to Epo (Figure 4I).
J2-Lyn−/− cells were then exposed to different concentrations of Epo in medium with enhanced differentiation competency (IMDM supplemented with 10% thyroid hormone-depleted FBS)  and the biological responses compared with J2-WT cells. Control cells displayed a typical Epo-dependent viability signal in differentiation medium [T3 (3,3′, 5-tri-iodothyronine)-depleted FBS] at 48 h, with the highest concentrations of Epo producing viability equivalent to that in maintenance medium (non-T3-depleted FBS) (Figure 4J). Significantly, J2-Lyn−/− cells exhibited very low viability, compared with J2-WT cells when cultured in differentiation medium for 48 h, and although there was an Epo-dependent viability signal, unlike control cells this failed to be sufficient to generate viability levels equivalent to that seen in maintenance medium (Figure 4J). Taken together these results demonstrate that Lyn is required for viability, proliferation and differentiation of immortalized erythroid progenitors.
Lyn is required for both pro-survival signalling through GAB2/Akt/FoxO3 and feedback inhibition of JAK2/STAT5 and ERK1/2 via SHP-1
To examine the effects of loss of Lyn on Epo signalling, the immortalized fetal liver erythroid cell lines were analysed biochemically. Anti-phosphotyrosine immunoblots showed a substantive lack of Epo-induced tyrosine phosphorylation events (clear banding differences are indicated, Figure 5A) and, as expected, a lack of prominent bands corresponding to Lyn (p53/56 kDa) (Figure 5A), as well as changes prior to Epo addition. These data indicate that Lyn is responsible for many of the phosphotyrosine intracellular signalling events in erythroid cells before and following exposure to Epo. To explore these changes in greater detail, a series of immunoblots were conducted with antibodies directed against specific Epo-R signalling molecules (Figures 5B–5D). As anticipated, Lyn (and pLyn Tyr508) was not detectable in J2-Lyn−/− cells, and active SFKs were virtually absent, supporting our previous observation that Lyn is the major SFK in erythroid cells . Significantly, a lack of Lyn did not impede the initiation of Epo-R signalling in terms of tyrosine phosphorylation of the receptor, JAK2 or STAT3/STAT5 (Figure 5B). Indeed, J2-Lyn−/− cells displayed elevated levels of pEpo-R, pJAK2 and pSTAT3/pSTAT5. Interestingly, the kinetics of JAK2 and STAT phosphorylation were altered in J2-Lyn−/− cells. At 30 min post-Epo stimulation, JAK2 and STAT3/STAT5 were maximally phosphorylated, whereas in control cells this occurred at 10 min with decreases in phosphorylation at 30 min. This could potentially be explained by the lack of SHP-1 phosphorylation and its activation in J2-Lyn−/− cells (Figure 5C), as SHP-1 is primarily involved in dephosphorylating and thus inactivating JAK2 . This could also explain the elevated Epo-R and STAT3/STAT5 phosphorylation at 30 min as a result of elevated JAK2 activation. The Epo-stimulated activation of SHP-2 was also significantly reduced in cells lacking Lyn. GAB2 is a major substrate of Lyn in haemopoietic cells, mediating PI3K/Akt activation and SHP-2 recruitment, and corroborating this established link; in J2-Lyn−/− cells essentially no tyrosine phosphorylation of GAB2 could be observed. Indeed, little Epo-induced activation of Akt could be observed in J2-Lyn−/− cells, compared with control cells; however, there was an elevated basal level of pAkt. Significant basal ERK1/2 activation was observed in cells lacking Lyn that was further increased upon Epo induction (Figure 5C), reflecting that which we have previously found for primary Lyn−/− erythroid cells isolated from mice [10,11]. The lack of SHP-1 phosphorylation in J2-Lyn−/− cells could explain the elevated ERK1/2 activation, as SHP-1 is known to dephosphorylate ERK1/2 and mediate its inactivation . The lack of GAB2 tyrosine phosphorylation could also explain the substantive decrease in SHP-2 phosphorylation in J2-Lyn−/− cells, as there would be reduced recruitment of SHP-2 to the GAB2 complex and thus reduced recruitment to the Epo-R signalling complex, due to a lack of Lyn-mediated GAB2 tyrosine phosphorylation. Epo-induced phosphorylation of PLCγ1 appeared unaffected by a lack of Lyn.
Western blot analysis of Epo-R signalling in erythroid Lyn−/− cells compared with Lyn+/+ cells
Considering the alterations to GAB2/Akt, we then looked at the phosphorylation status of several Akt substrates (Figure 5D). For this FoxO3 phosphorylation appeared to reflect pAkt dynamics in J2-WT cells (maximal at 10 min), whereas pGSK3α/β and pBAD(S112) appeared to correlate with the basal pAkt levels in J2-Lyn−/− cells. Interestingly, levels of BclXL were lower in J2-Lyn−/− cells compared with control cells as was CIS (cytokine-inducible Src homology 2-containing protein), whereas levels of SOCS1 (suppressor of cytokine signalling 1), SOCS3 and p27Kip1 did not appear significantly different in cells lacking Lyn.
Collectively, these data demonstrate that a lack of Lyn has a significant impact on signal transduction. Since these cells are immortalized at the proerythroblast stage , these data suggest that Lyn plays a major role in transmitting Epo-dependent inhibitory signals via SHP-1 in addition to inducing a GAB2/Akt/FoxO3 survival pathway in immature erythroid cells.
We then looked at direct chemical inhibition of Lyn with PP2, and JAK2 with JAKI in J2-WT cells to assess whether the alterations to signalling observed in J2-Lyn−/− cells could be via a lack of Lyn kinase activity and/or Lyn adaptor activity (Figure 6). JAKI almost fully repressed the majority of Epo-induced signalling, including Epo-R, JAK2, STAT5 (Figure 6A), GAB2, Akt, FoxO3, SHP-2 and ERK1/2 phosphorylation (Figure 6B), illustrating the central importance of JAK2 for initiating receptor signalling. Furthermore, Lyn activation was also reduced by JAK inhibition, supporting Lyn activation being downstream of JAK2. Interestingly, SHP-1 phosphorylation was significantly elevated by JAK2 inhibition, suggesting that Epo-induced receptor engagement, in the absence of significant JAK2 kinase activity, can still lead to SHP-1 activation. Potentially this could be independent of Lyn due to the reduced Lyn activity apparent with JAKI. Alternatively, it could be mediated by residual Lyn activity being directed more to SHP-1 in the absence of modulating Epo-R signals that require JAK2 activity.
Effects of chemical inhibitors PP2 and JAKI on Epo-R signalling in erythroid J2-WT cells
Inhibition of Lyn kinase activity via PP2 both mitigated and enhanced different downstream Epo-R pathways as we hypothesized from our J2-Lyn−/− results. Epo-induced phosphorylation of Epo-R and JAK2 were not inhibited by PP2 addition (Figure 6A). However, STAT5 phosphorylation was elevated in the presence of PP2, potentially due to reduced SHP-1 activation (Figures 6A and 6B). The activation of Akt was repressed by PP2 to a similar extent as JAK2 inhibition (Figure 6B), illustrating the importance of Lyn for Epo-R-dependent induction of the Akt pathway, potentially via a failure to induce tyrosine phosphorylation of the adaptor GAB2, which feeds into the Akt pathway. Consequently tyrosine phosphorylation of the Akt substrate FoxO3 was reduced. Interestingly, the serine phosphorylation (Ser159) of GAB2 was increased with PP2, even though there was reduced Akt activity (the kinase shown to phosphorylate this site ), but correlates with the elevated ERK1/2 activity (Figure 6B), suggesting that ERK1/2 is responsible for phosphorylating this site in GAB2. Furthermore, Epo-induced phosphorylation of GSK3α/β was elevated by Lyn inhibition (Figure 6B), similar to that seen in Lyn−/− cells (Figure 5D), suggesting that Akt may not be the primary kinase that phosphorylates GSK3α/β in these cells. The activation of ERK1/2 was significantly elevated by Lyn inhibition (Figure 6B), reflecting that seen in Lyn−/− cells (Figure 5C) and correlates with GSK3α/β phosphorylation, suggesting that ERK1/2 may in part be responsible for GSK3α/β phosphorylation/inactivation following Epo stimulation. Elevation of ERK1/2 levels with the addition of PP2 was also observed prior to Epo addition (Figure 6C), and may also feed into the inhibitory serine phosphorylation of GAB2. Interestingly, both total and phosphorylated FoxO3 appears to be predominantly cytoplasmic in both J2-WT and J2-Lyn−/− cells before and after Epo stimulation (Figure 6D).
To relate these short-term signalling alterations to the differentiation/survival alterations in extended Epo inductions (Figures 4G and 4H), signalling/viability/cell cycle molecules were examined over an extended timeframe in medium that is highly conducive for erythroid differentiation (T3-depleted FBS) [30,36]. For J2-WT cells at 48 h in the presence of Epo, when differentiation reaches a maximum [30,36], high levels of pSTAT5 are present, and pAkt is detectable, pFoxO3 is increased and pGSK3α/β levels are maintained (Figure 7A). Furthermore, in J2-WT cells the level of the pro-apoptotic molecule BAD increases in differentiation medium even in the presence of Epo; however, BclXL levels remain high in response to Epo, presumably mediating their continued overall survival signal by tipping the balance of pro- and anti-apoptotic molecules towards a pro-survival state. In J2-WT cells a significant increase in the cell cycle inhibitor p27Kip1 is observed in the presence of Epo, indicative of differentiating cells that have exited the cell cycle . In stark contrast, J2-Lyn−/− cells failed to maintain high pSTAT5 levels. Furthermore, they failed to mount any significant inactivation of FoxO3 or maintain pGSK3α/β levels. Commensurate with this was a significant increase in total BAD levels and a failure to maintain BclXL protein (Figure 7A). Taken together, these data could explain the very low survival of J2-Lyn−/− cells in differentiation medium even in the presence of Epo (Figures 4G and 4H). There was also a failure to significantly induce p27Kip1, indicating the cells had not exited the cell cycle, reflecting their lack of differentiation induction.
Next, we sought to confirm the Epo-receptor signalling alterations we saw in the immortalized erythroid cells (Figure 5) using primary erythroid cells isolated from the spleen of PHZ-treated Lyn+/+ and Lyn−/− mice. Encouragingly, Epo-induced pSTAT5 levels were elevated, whereas pFoxO3, pSHP-1 and pSHP-2, as well as tyrosine phosphorylation of GAB2, were reduced in Lyn−/− cells (Figure 7B), consistent with our findings from the immortalized J2-Lyn−/− cells. The levels of pAkt and pERK1/2 were high in the Lyn−/− cells before addition of Epo, which correlated with the immortalized cell data. Furthermore, although the Lyn−/− cells had elevated pAkt before Epo addition, this did not translate into elevated pFoxO3 levels. This contrasted with the results from Lyn+/+ cells, where the Epo-induced Akt activity correlated with pFoxO3 levels.
Lack of pro-survival pathway activation and cell cycle exit in J2-Lyn−/− erythroid cells grown in differentiation medium, Epo-R signalling changes in primary Lyn−/− erythroblasts and model of Lyn signalling
We have previously shown that the erythroid compartment is perturbed in Lyn-deficient mice [10,11]. This is caused by stem/progenitor intrinsic defects reducing their expansion capacity and late-stage development [9,38], which results in extramedullary emergency erythropoiesis in the spleen in an attempt to compensate for a developing anaemia. We demonstrate in the present study that Lyn is required for proficient expansion and maturation of splenic/emergency erythroid progenitors in ex vivo cultures. Furthermore, in immortalized fetal liver erythroid lines and primary erythroid cells from PHZ-treated Lyn−/− mice, both pro-survival GAB2/Akt/FoxO3 signalling and feedback inhibition of JAK2/STAT5 and ERK1/2 via SHP-1 are significantly altered.
We established an ex vivo erythroblastic island culture system to study the effect of signalling pathway inhibition on the expansion and differentiation of erythroid cells, to more closely reflect their in vivo milieu, i.e. the presence of stromal macrophages that nurse the developing erythroblasts [39–43]. Direct pharmacological inhibition of SFKs (including Lyn) within this system significantly inhibited the terminal maturation of these committed erythroid cells, but not their expansion. This supports our previous in vitro and in vivo delineation of an important function of Lyn in erythropoiesis [7,10,11], and suggests a specific requirement for differentiation/late-stage survival. We then looked at the effect of Lyn deficiency on ex vivo erythroblastic island cultures, via comparison of cells isolated from Lyn+/+ and Lyn−/− mice. Utilizing medium optimized for expansion/maintenance to focus on these aspects of erythropoiesis, we found that Lyn−/− cultures had significantly altered erythroid progenitor expansion profiles. Specifically, maturation associated down-regulation of surface CD71 (transferrin receptor) was retarded in Lyn−/− cultures. Although Lyn−/− c-Kit+ cells co-expressing low levels of Ter119 expanded to a greater degree, c-Kit+/Ter119high cells expanded less, compared with control cultures. These results suggest that not only is Lyn important for late-stage maturation, but also for traversing through the different stages of erythroid progenitor commitment. Indeed, Lyn appears to be required in the transition from early to late erythroid progenitors particularly in the conversion of SCF (via c-Kit)-responsive into Epo-responsive progenitors . This is consistent with our findings with ex vivo emergency/stress-erythroid Lyn−/− cultures. Thus primary bone marrow and emergency/stress erythroid progenitors from Lyn−/− mice both have expansion/stage transition defects, potentially via reduced capacity to correctly interchange c-Kit/Epo-R responsiveness [9,38,44].
Turning to cultures utilizing medium proficient for expansion and differentiation, we found a significant alteration in the maturation dynamics in cells with a null mutation in Lyn. Cultures of Lyn−/− cells displayed a lag in their progression through morphological maturation, with a significant percentage of erythroblasts still present after 72 h of culture, whereas control cultures had almost completely differentiated into enucleated reticulocytes. This retardation of morphological maturation caused by a lack of Lyn was also reflected in reduced BclXL levels. These data lend further support to Lyn being required for correct transitioning through erythroid progenitor stages and morphological maturation [9,38,44].
To delineate the changes to Epo-R-dependent and -independent signalling dynamics in Lyn−/− erythroid cells, we immortalized fetal liver erythroblasts. We used day 12.5 embryos to obviate establishment of compensatory pathways that could occur in adult mice. Significantly, Lyn−/− lines exhibited reduced viability, clonogenicity and differentiation, as well as lengthened doubling times. Not only did J2-Lyn−/− cells have lower viability in maintenance cultures, but also Epo was unable to provide a strong viability signal in differentiation medium (thyroid hormone-depleted FBS) [30,36], unlike that seen for control J2-WT cells. This showed significant Epo-R-independent signalling alterations and proliferation/viability alterations in Lyn−/− immortalized erythroid cells. Although there was evidence of a viability signal propagated via Epo-R in J2-Lyn−/− cells, this was insufficient, even at the highest Epo doses tested (10 units/ml), to compensate for that produced by thyroid hormone [30,36]. In addition, biochemical data demonstrates that an absence of Lyn significantly alters signalling downstream of Epo-R, including the GAB2/Akt/FoxO3 cascade, and feedback inhibition of JAK2/STAT and ERK1/2 via SHP-1, as well as has significant impact on Epo-R-independent signalling, viability and differentiation. We and others have shown an important role for Lyn in phosphorylating SHP-1 in myeloid, mast and B-cells to mediate the subsequent SHP-1 regulation of the ERK1/2 pathway [11,22,34,45]. Consequently, it appears that Lyn is very important in diverse haemopoietic cells/haemopoietic receptor systems in mediating activation of SHP-1 to provide negative control of ERK1/2 signalling. The elevated ERK1/2 activity in Lyn−/− cells and in cells incubated with PP2 correlates (before and after Epo stimulation) with increased phosphorylation of several Akt substrate sites on GAB2 and GSK3α/β, suggested that ERK1/2 could potentially phosphorylate these sites. Additionally, Lyn is also important for regulation of JAK/STAT signal strength [8,10,46]. Lyn has been found to phosphorylate Epo-R and STAT5 , as well as STAT3 , but through its activation of SHP-1 can also mediate feedback inhibition of JAK/STAT signalling [33,47], which is corroborated by our data on Lyn−/− erythroblasts.
The alterations to GAB2/Akt/FoxO3 in Lyn−/− cells could be the cause of a failure to suppress anti-survival signals and promote cell cycle exit upon Epo-induced differentiation. Alternatively, a lack of Lyn could reduce the differentiation/morphological maturation capacity of J2-Lyn−/− erythroblasts that results in reduced survival signals being present as Lyn is also an important factor in mature erythrocytes post-Epo-R signalling influences [16,17,48]. Significantly, we have recently found that erythroid cells from mice with a gain-of-function Lynup allele, which generates a constitutively active Lyn kinase, display elevated Akt/FoxO3 pathway activation and enhanced survival, as well as having a major effect on mature erythrocyte morphology . This correlates well with the reduced Akt/FoxO3 and viability we see here in Lyn−/− erythroid cells, and also further shows the importance of Lyn for mature erythrocyte form/function. Interestingly, although short-term Epo-R signalling through JAK2/STAT was elevated in J2-Lyn−/− cells, this was not the case in long-term Epo-stimulated differentiation cultures. Potentially, a failure to provide feedback inhibition of Epo-R/JAK2 phosphorylation through SHP-1 activation in J2-Lyn−/− cells may promote Epo-R complex internalization/degradation  and mitigate subsequent long-term sustained viability activation through STAT5 and its promotion of BclXL levels. It is conceivable that the decreased viability initiated by a lack of FoxO3 phosphorylation (inactivation) occurs by up-regulated expression of pro-apoptotic genes, e.g. BAD (as we show here), Bim and PUMA (p53 up-regulated modulator of apoptosis) [50–52]. Akt-mediated phosphorylation of FoxO3  is implicated in Epo-induced viability and transient inhibition of cell cycle exit [54,55]. During late-stage differentiation FoxO3 appears to be required for promotion of the differentiation of erythroblasts through mediating the cell cycle exit as well as reducing the oxidative stress caused by high levels of haemoglobin in differentiating erythroblasts . We also observed that total as well as phosphorylated FoxO3 were predominantly cytoplasmic in both wild-type and Lyn−/− cells before and during the 30 min time course of Epo stimulation. This is consistent with previous findings showing that FoxO3 nuclear localization in erythroid cells occurs late in their maturation . Taken together, these data suggest a viability pathway involving Akt/FoxO3 in suppressing BAD levels, or other pro-apoptotic molecules (i.e. PUMA, Bim) [50–52], is partially dependent upon Lyn, as well as maintenance of sustained pSTAT5/BclXL during differentiation. Without Lyn, these immortalized erythroid progenitors have a reduced ability to correctly respond to Epo in the short-term, as well as a reduced ability under differentiation conditions (which is initiated by but does not require Epo during morphological maturation) to exit the cell cycle and maintain viability while differentiating.
We propose that altered intracellular signalling, that is both Epo-dependent and -independent caused by a lack of Lyn (Figure 7C), significantly perturbs the viability of erythroid cells, particularly during their late-stage development. A potential consequence of this in mice with a null mutation in Lyn is that erythroid progenitor and precursor levels are elevated in an attempt to overcome a defect in their late-stage maturation/viability, but that this expansion is insufficient to fully overcome this, resulting in the development of age-related anaemia [10,11]. Additionally, compensatory signalling mechanisms are almost certainly activated in the adult animal that reduces the severity of the viability signalling/maturation phenotype we observe in the immortalized fetal liver erythroblasts. This has been observed in mice lacking STAT5  and indeed in adult bone marrow cultures from Lyn−/− mice that display stage-specific compensatory activation of Akt .
Previously we showed that a loss of Lyn had a significant impact upon erythropoiesis [10,11]. In the present study we have delineated that both positive viability signals (GAB2/Akt/FoxO3) and feedback inhibitory signalling (SHP-1/2) are significantly altered in Lyn-deficient in erythroid cells (Figure 7C), influencing cell survival and their ability to mature/differentiate. Together, these results illustrate that Lyn is very important for normal erythropoiesis, erythroid cell development during both Epo-dependent and -independent signalling.
erythroid Kruppel-like factor
extracellular-signal-regulated kinase 1/2
forkhead box O
glycogen synthase kinase
Iscove’s modified Dulbecco’s medium
JAK2 inhibitor 1
mitogen-activated protein kinase
p53 up-regulated modulator of apoptosis
stem cell factor
Src family kinase
Src homology 2 domain-containing protein tyrosine phosphatase
suppressor of cytokine signalling
signal transducer and activator of transcription
Neli Slavova-Azmanova and Nicole Kucera designed, supported and performed experiments, and analysed data. Alison Louw, Jiulia Satiaputra, Peter Singer, Adley Handoko, Leah Stone and David McCarthy performed experiments. S. Peter Klinken supported the research and contributed to writing the paper. Margaret Hibbs supported the research, designed experiments, analysed results and contributed to writing the paper. Evan Ingley designed and supported the research, designed and undertook experiments, analysed data and contributed to writing the paper.
We thank Irma Larma for technical assistance.
This work was supported by the National Health and Medical Research Council [grant numbers 513714 and 634352], the Medical Research Foundation of Royal Perth Hospital, and the Cancer Council of Western Australia. E.I. is supported by a Senior Research Fellowship from the Cancer Council of Western Australia. M.L.H. is supported by a Senior Research Fellowship from the National Health and Medical Research Council, Australia.