Phosphorylation of Serine 248 of C/EBPα Is Dispensable for Myelopoiesis but Its Disruption Leads to a Low Penetrant Myeloid Disorder with Long Latency

Background Transcription factors play a key role in lineage commitment and differentiation of stem cells into distinct mature cells. In hematopoiesis, they regulate lineage-specific gene expression in a stage-specific manner through various physical and functional interactions with regulatory proteins that are simultanously recruited and activated to ensure timely gene expression. The transcription factor CCAAT/enhancer binding protein α (C/EBPα) is such a factor and is essential for the development of granulocytic/monocytic cells. The activity of C/EBPα is regulated on several levels including gene expression, alternative translation, protein interactions and posttranslational modifications, such as phosphorylation. In particular, the phosphorylation of serine 248 of the transactivation domain has been shown to be of crucial importance for granulocytic differentiation of 32Dcl3 cells in vitro. Methodology/Principal Findings Here, we use mouse genetics to investigate the significance of C/EBPα serine 248 in vivo through the construction and analysis of Cebpa S248A/S248A knock-in mice. Surprisingly, 8-week old Cebpa S248A/S248A mice display normal steady-state hematopoiesis including unaltered development of mature myeloid cells. However, over time some of the animals develop a hematopoietic disorder with accumulation of multipotent, megakaryocytic and erythroid progenitor cells and a mild impairment of differentiation along the granulocytic-monocytic lineage. Furthermore, BM cells from Cebpa S248A/S248A animals display a competitive advantage compared to wild type cells in a transplantation assay. Conclusions/Significance Taken together, our data shows that the substitution of C/EBPα serine 248 to alanine favors the selection of the megakaryocytic/erythroid lineage over the monocytic/granulocytic compartment in old mice and suggests that S248 phosphorylation may be required to maintain proper hematopoietic homeostasis in response to changes in the wiring of cellular signalling networks. More broadly, the marked differences between the phenotype of the S248A variant in vivo and in vitro highlight the need to exert caution when extending in vitro phenotypes to the more appropriate in vivo context.


Introduction
The human body contains trillions of blood cells that are continuously replaced through normal cell turnover. Hematopoiesis is the highly orchestrated process responsible for regulating the generation of mature blood cells from a rare population of hematopoietic stem cells (HSC). The HSCs possess the ability to self-renew and differentiate into all blood lineages and are the ultimate reservoir for maintaining the supply of blood cells throughout life. Multiple mechanisms are required in order to meet both the changing demands from the body and to maintain steady-state hematopoiesis [1]. In particular, many transcription factors have been shown to modulate key events in differentiation and proliferation and their function in hematopoiesis has been investigated thoroughly through the examination of knockout mice [2]. One of these transcription factors is CCAAT/enhancer binding protein alpha (C/EBPa), which is not only involved in regulation of hematopoiesis, but also exerts its function in other tissues such as lung, liver and adipose tissue through the induction of lineage-specific gene programs in combination with an ability to promote cell cycle exit [3,4,5]. Within the hematopoietic system, C/EBPa has been shown to be important for the myeloid lineage, since conditional deletion of the Cebpa allele in the hematopoietic compartment of adult mice blocks the transition from common myeloid progenitors (CMPs) to granulocyte-monocyte progenitors (GMPs), thus resulting in complete loss of granulocytes, monocytes and eosinophils [6,7]. Besides this late granulocytic-monocytic differentiation block, fetal livers of newborn Cebpa null mice display increased numbers of progenitors and mature cells of the erythroid lineage, suggesting that C/EBPa might play a role in repressing erythroid differentiation [7]. In line with this, overexpression of C/ EBPa in erythroid progenitor cells, redirects the differentiation potential in a granulocytic direction resulting in an increased level of mature granulocytes and granulocyte-monocyte progenitors with a concomitant decrease of erythroid progenitors [8].
The activity of C/EBPa is tightly controlled through multiple layers of regulation. First of all, timely expression is required and involves regulation of gene transcription, mRNA translation and protein degradation [9,10]. Secondly, protein interactions have a major impact on the ability of C/EBPa to induce or repress gene transcription [11,12,13]. Thirdly, C/EBPa activity can be altered by posttranslational modifications such as sumyolation and phosphorylation [14,15]. The phosphorylation status of serine 21 (S21) has been shown to have a major impact on the decision to differentiate towards the monocytic or granulocytic lineage in vitro. Upon phosphorylation of S21 or expression of a phosphomimicking mutant in K-562 cells, granulopoiesis is inhibited, thereby favoring monocytic differentiation at the cost of granulocytic differentiation [15,16]. Furthermore, the phosphorylation of serine 248 in the transactivation domain has been suggested to be required for myeloid differentiation since mutating it to an alanine residue abrogates the capacity of C/EBPa to induce granulocytic differentiation of 32Dcl3 cells in vitro [17]. Serine 248 (S248) is phosphorylated by activated Ras signaling and this phosphorylation increases the ability of C/EBPa to promote expression from the G-CSF receptor promoter. Therefore, it appears that phosphorylation and dephosphorylation of C/EBPa at distinct serine residues can directly push the cells towards a specific myeloid branch. However, these phosphorylation sites have only been investigated in an in vitro context and what the functions are in vivo is therefore unknown.
We and others have previously reported on several Cebpa knockin mouse models [18,19,20,21,22,23], which have provided valuable information pertaining the role of C/EBPa in myeloid differentiation and in the development of leukemia. In this study, we use knock-in mutagenesis to elucidate the importance of S248 phosphorylation for myeloid differentiation by introducing an allele of Cebpa with an alanine substituted for serine 248, thereby abrogating phosphorylation of this residue. Surprisingly, we could show that whereas myeloid differentiation of cells expressing C/ EBPa-S248A is blocked in vitro, this is not the case in vivo. Thus, young Cebpa S248A/S248A mice display no phenotypic alterations in the hematopoietic compartment or other tissues. In contrast, aged Cebpa S248A/S248A animals develop a low-penetrant myeloid disorder characterized by a mild impairment of differentiation along the granulocytic-monocytic lineage and by the accumulation of HSCs, multipotent progenitor cells (MPPs), as well as megakaryocytic and early erythroid progenitors.

S248 is required for C/EBPa to promote granulocytic differentiation in vitro
The murine myeloid 32Dcl3 cell line has long been considered a suitable in vitro model system for analyzing myelopoiesis, since it is one of the few cell lines that can terminally differentiate into mature neutrophils. The cell line is diploid and non-leukemic in syngenic murine recipients [24,25]. It proliferates in media containing IL-3 however, upon removal of this cytokine and addition of G-CSF, proliferation ceases and differentiation into neutrophil granulocytes proceeds. It is well documented that ectopic expression of C/EBPa in 32Dcl3 cells is sufficient to induce terminal granulocytic differentiation even in the presence of IL-3, making this a suitable differentiation assay to analyze the effect of C/EBPa mutations on this process [24,25]. In order to investigate if C/EBPa-S248A is defective in granulocytic-monocytic in vitro differentiation as previously reported [17], 32Dcl3 clones expressing either a wild type C/EBPa-estrogen receptor ligand-binding domain fusion protein (C/EBPa-ER) or the C/ EBPa-S248A-ER variant were constructed and clones expressing an equal amount of protein were selected for further analysis ( Figure 1A).
To test whether S248 is required for the ability of C/EBPa to promote granulocytic differentiation C/EBPa-ER was translocated to the nucleus by addition of 4-hydroxytamoxifen (4-OHT). Cells were monitored for three days, and samples were collected each day and analyzed by flow cytometry to assess proliferation ( Figure 1D) and granulocytic differentiation ( Figure 1B, C). As expected 32Dcl3-C/EBPa-ER cells exited cell cycle within a few days, the G1/S ratio increased 15 fold ( Figure 1D) and cells became positive for the granulocytic-monocytic marker Mac-1 and later for Gr-1 ( Figure 1B, C). On the other hand, nuclear translocation of C/EBPa-S248A-ER did not lead to growth arrest and cells failed to express the two differentiation markers.
Thus in agreement with previous findings [17], these data show that S248 is necessary for C/EBPa to induce growth arrest and differentiation of neutrophil granulocytes in vitro.

Initial analysis of Cebpa S248A/S248A knock-in mice
In order to investigate the importance of C/EBPa-S248 in vivo, we generated a knock-in mouse line, in which the wild type Cebpa gene was replaced with an allele expressing C/EBPa-S248A. Cebpa S248A/S248A knock-in mice were born and weaned in Mendelian ratios, were physically indistinguishable from both wild type littermates and the more appropriate Cebpa KI/KI controls [20,26], showed no visible signs of illness and were fully fertile (data not shown), thus demonstrating that S248 of C/EBPa is fully dispensable for embryonic survival. Furthermore, inspection of tissues in which C/EBPa function has previously been demonstrated to be important, such as liver, lung, spleen and white adipose tissue revealed no obvious abnormalities neither in terms of morphology nor size in the Cebpa S248A/S248A animals ( Figure S1 and data not shown). These findings suggest that S248 is dispensable for the development and maintenance of these tissues.
Young Cebpa S248A/S248A mice are phenotypically normal in the hematopoietic system C/EBPa is a key regulator of myeloid differentiation and altered C/EBPa activity has been shown to affect hematopoietic development and predispose to myeloid malignancies such as AML [3,21,22]. Moreover, the finding that C/EBPa-S248A was unable to direct granulocytic-monocytic differentiation in vitro prompted us to investigate the hematopoietic system of Ceb-pa S248A/S248A mice. Bone marrows (BM) from 8 week-old Cebpa S248A/S248A and Cebpa KI/KI mice were collected and stained with antibodies for cell-surface markers specific for the different mature lineages and analyzed by flow cytometry. Surprisingly, in contrast to the in vitro data reported above, mutation of S248 did not alter the frequency of mature granulocytes (Mac-1+, Gr-1+) and the prevalence of other BM populations such as B cells (B220+), T cells (CD4+ and/or CD8+), erythroid cells (Ter119+, CD71+/int) was also unaffected (Figure 2A, B). Furthermore, examination of cytospins prepared from BM or spleen from 8 week-old mice showed no aberrations in cellular morphologies or distributions ( Figure 2C, D). These data suggest, that S248 of C/ EBPa is dispensable for the in vivo development of mature hematopoietic lineages.
Since Cebpa null mice have previously been shown to accumulate myeloid progenitors and to harbor a differentiation block upstream of the GMP [6], we analyzed the myeloid progenitor compartment using the antibody panel reported by Pronk et al. [27]. However, we were unable to detect any major alterations in the cellular distributions of myeloid progenitors or HSCs/MPPs ( Figure 3A, B), suggesting that S248 of C/EBPa is not required for steady state lineage commitment or differentiation of myeloid progenitors in young mice.
To evaluate the myeloid progenitors functionally, we next plated BM cells from 8 week-old Cebpa S248A/S248A and Cebpa KI/KI animals in semisolid media to allow for the outgrowth of all myeloid colonies. In contrast to previous studies involving BM progenitors derived from other Cebpa mutant mouse lines [19,21], Cebpa S248A/S248A BMs yielded similar distributions of BFU-E, CFU-GM and CFU-GEMM colonies as the Cebpa KI/KI controls ( Figure 3C). Furthermore, when we analyzed the proliferative consequences of mutating S248 by performing serial replating of colonies derived from the first round of plating, Cebpa S248A/S248A progenitors displayed the same limited proliferative potential as progenitors from the Cebpa KI/KI controls when counted either as colonies ( Figure 3D) or as total cell number (data not shown).
In conclusion, at eight weeks of age Cebpa S248A/S248A mice display no overt phenotype in the hematopoietic compartment.
A fraction of Cebpa S248A/S248A mice develop a lowpenetrant hematopoietic disorder with long latency Many hematopoietic diseases are particularly prevalent in the elderly population, but the mechanisms involved are not resolved but may be related to an increase in myeloid-based HSCs upon ageing or accumulation of acquired genetic mutations [28]. We therefore analyzed a cohort of Cebpa S248A/S248A animals at one year of age (12-14 months) to detect whether mutation of S248 led to any phenotypic changes in the hematopoietic compartment in older animals. At this age, several of the Cebpa S248A/S248A mice (4/ 23) had enlarged spleen ( Figure S2D), whereas all of the knock-in wild type controls were phenotypically indistinguishable from the young wild type controls.
Next, we analyzed the distribution of the mature hematopoietic populations within the BM of cohorts of Cebpa KI/KI control and Cebpa S248A/S248A mice. In the BM from Cebpa S248A/S248A mice there was a slight reduction in mature neutrophil granulocytes (Mac-1+, Gr-1+) and a concomitant increase in erythroid progenitors (Ter119+, CD71+), whereas the lymphoid compart-ment was unaffected compared to the Cebpa KI/KI controls ( Figure 4A, B).
To further delineate this apparent skewing towards the erythroid lineage and to test whether this was driven by a few Cebpa S248A/S248A individuals or the entire cohort, we analyzed the myeloid progenitor compartment ( Figure 5A). In contrast to young mice, a subset of the one year-old Cebpa S248A/S248A animals displayed reduced levels of GMPs accompanied by an increase in preMegEs. Whereas 6% (1 out of 17) Cebpa KI/KI mice had a reduced GMP/preMegE ratio, approximately 30% (7 out of 23) of the Cebpa S248A/S248A animals displayed a GMP/preMegE ratio lower than the mean of the Cebpa KI/KI animals minus one standard deviation ( Figure 5B and Figure S2). These phenotypically progressed Cebpa S248A/S248A mice had a significant increase in the preGMs ( Figure 5C), indicating partially impaired differenti-ation towards the granulocyte-monocytic compartment at the preGM to GMP transition, where C/EBPa executes its lineageinstructive function [6,7].
Furthermore, the partly impaired preGM to GMP transition was accompanied by increased levels of megakaryocytic and erythroid progenitors (preMegE, MkP, preCFU-E), suggesting that the partial block in granulocytic-monocytic differentiation directs the cells towards the erythroid lineage ( Figure 5C). This lineageskewing was also detected in colony assays, where Cebpa S248A/S248A BM gave rise to significantly more BFU-Es than the Cebpa KIKI control ( Figure 4C).
In addition to the observed changes in the committed progenitor compartments, a fraction of the one year-old Cebpa S248A/S248A animals with biased lineage choice displayed increased frequencies of the HSC/MPP-enriched Lineage negative, Sca-1+, c-Kit+ (LSK) population compared to Cebpa KI/KI mice. Further resolution of this compartment showed that the progressed Cebpa S248A/S248A BM had a two-fold expansion of MPPs compared to Cebpa KI/KI controls as well as a trend towards increased numbers of HSCs ( Figure 5C). Notably, we observed only a partial overlap between delayed preGM to GMP transition and expansion of the stem and multipotent progenitor compartment ( Figure S2), suggesting that these events occur by different mechanisms.
Next we wanted to assess in more detail whether there existed any correlations between the different aberrant phenotypes (enlarged spleen, expanded LSK compartment, erythroid-biased differentiation, etc) that we observed in old Cebpa S248A/S248A animals. We therefore subjected a second cohort of old animals (18-24 months) to the full experimental regime and in support of our initial findings, this group of Cebpa S248A/S248A animals displayed essentially the same phenotypic characteristics as the previously analyzed 1-year old cohort: Thus, whereas all (8 out of 8) Cebpa KI/KI mice had a phenotypical normal spleen, 29% (6 out of 21) of the Cebpa S248A/S248A animals displayed splenomegaly at various degrees ( Figure S3C). Twenty-nine % (6 out of 21) of the Cebpa S248A/S248A animals had an expansion of the LSK compartment compared to 13% (1 out of 8) of Cebpa KI/KI ( Figure S3B). Moreover, 29% (6 out of 21) of Cebpa S248A/S248A mice had a reduced GMP/preMegE ratio in the BM compared to 0% (0 out of 8) of the Cebpa KI/KI mice ( Figure S3A). The phenotypes of the older Cebpa S248A/S248A mice are summarized in Table 1 and Figure 6. Specifically when mice are binned according to GMP/ preMegE ratios, the fraction of Cebpa S248A/S248A animals associated with decreased GMP/preMegE ratios are significantly elevated compared to Cebpa KI/KI controls.
Importantly, the full complement of analyses performed on the 18-24 month cohort allowed us to draw a number of correlations between the different aberrant phenotypes in the old Cebpa S248A/ S248A animals. Thus, whereas we detected a low partial overlap of mice with decreased GMP/preMegE ratio and increased LSK compartment, there was no detected correlation between mice with enlarged spleen and reduced GMP/preMegE ratio. In contrast, there was a considerable overlap of mice with expanded LSK compartment and enlarged spleen ( Figure S3D-G).
At this age the phenotypically progressed Cebpa S248A/S248A mice had slightly lower levels of mature monocytes/granulocytes in the BM and an increased amount of mature erythroid cells in the BM compared to Cebpa KI/KI mice ( Figure S4A). In line with this, BM cells from the progressed Cebpa S248A/S248A mice displayed significantly more BFU-E colonies in CFU assays, whereas the CFU-GM and CFU-GEMM were unaltered compared to BM from Cebpa KI/KI mice ( Figure S4B).
Taken together, a fraction of old mice defective in the phosphorylation of S248 of C/EBPa exhibit a slowly developing low penetrant hematopoietic disorder(s) associated with distur-  (Figure 7 and Figure  S5). Consistent with the phenotype in the Cebpa S248A/S248A mice, there was only a partial overlap between the disturbance in the myeloid compartment and increased numbers of LSK cells ( Figure  S5), which supports the notion that the partly halted granulocyticmonocytic differentiation and the accumulation of HSCs and MPPs occur by two distinct mechanisms.
To test whether the S248A allele results in a stem cell repopulation advantage or disadvantage compared to the wild type knock-in allele, we performed serial competitive BM transplantations, in which whole BM (CD45.2) from eight weekold Cebpa S248A/S248A and Cebpa KI/KI mice was mixed in a 1:1 ratio with competitor BM (CD45.1) and transplanted into lethally irradiated CD45.1 recipients. Recipient BM was harvested 18 weeks post-transplantation and donor (CD45.2) versus competitor (CD45.1) contribution was analyzed. As shown in figure 8A, there were no significant differences in these ratios between recipients transplanted with Cebpa S248A/S248A and Cebpa KI/KI donor cells, respectively. To test whether a repopulation phenotype could be uncovered when further proliferative stress was applied to the hematopoietic system, we transplanted pooled whole BM from 6-7 primary recipients ( Figure 8B) into secondary recipients, which were analyzed 21 and 34 weeks post-transplantation. At 21 weeks after secondary transplantation, the LSK compartment of mice transplanted with Cebpa S248A/S248A was significantly increased compared to mice transplanted with Cebpa KI/KI BM and this increased contribution in the stem cell compartment of Cebpa S248A/ S248A mice resulted in a significant increase of donor BM cells at 34 weeks post-transplantation ( Figure 8C, D). It should be noted that a selective advantage of donor-versus competitor cells was also observed in mice transplanted with Cebpa KI/KI secondary donors, since the donor/competitor ratio had also increased relative to input cells in these animals. Therefore, some selective advantage must be assigned to the genetic background of the cells used. However, the effect was significantly higher in the Cebpa S248A/ S248A secondary recipients suggesting that the S248A mutation confer a mild selective advantage to the HSC/MPP compartment.
In summary, the Cebpa S248A allele results in a low-penetrant cell-intrinsic partly impairment of granulocytic-monocytic differentiation accompanied by skewing towards the megakaryocytic and erythroid compartment in a fraction of aged animals. Furthermore, these data show that Cebpa S248A/S248A HSCs have a competitive advantage compared to their wild type counterparts resulting in the accumulation of HSCs and MPPs in some of the Cebpa S248A/S248A mice.

Discussion
Protein phosphorylation represents an important layer of cellular regulation, which may affect the stability, activity and functional property of the modified protein. Whether phosphorylation of key regulators plays a role in normal hematopoiesis has only been addressed to a limited extent and mainly in a cell culture context. The key hematopoietic regulator C/EBPa contains a number of phosphorylation sites that confer regulation under a variety of conditions [15,29,30]. Previously, Ras-dependent phosphorylation of C/EBPa at S248 was reported to increase its transactivational activity and promote increased granulocyticmonocytic differentiation in vitro, whereas Erk-dependent phosphorylation of S21 inhibited the same process [15,17]. Furthermore, it has been shown that S21 must be dephosphorylated in order to induce not only expression of granulocytic-monocytic markers but also the erythroid-specific CD71 [16]. These findings suggest that the phosphorylation status of C/EBPa is potentially involved in regulating both granulopoiesis and erythropoiesis.
In order to extend this analysis to the proper in vivo context, we decided to generate mice, in which the wild type Cebpa gene was replaced by an allele containing a S248A substitution. Before initiating the phenotypic analysis of these animals we verified the previous reports on the requirement of S248 for in vitro differentiation of granulocytic-monocytic cell lines [17,31]. Indeed, by introducing C/EBPa-ER and C/EBPa-S248A-ER into 32Dcl3 cells, we could show that the S248A variant is unable to promote myeloid differentiation. Whereas C/EBPa-ER induces the expression of myeloid markers such as Mac-1 and Gr-1 accompanied by cell cycle exit, 32Dcl3 cells expressing C/ EBPa-S248A-ER fail to induce expression of these markers and  Table 1  continue to proliferate. This suggests that phosphorylation of S248 is required in order for the cells to differentiate into mature granulocytic-monocytic cells in vitro.
Given the strong requirement for S248 for in vitro granulocyticmonocytic differentiation, we were surprised to find that young Cebpa S248A/S248A mice and their relevant controls contain a similar number of mature and immature myeloid cells. In line with this, BM cells plated in semi-solid media result in outgrowth of an equal amount of colonies with similar distributions of myeloid progenitors from Cebpa S248A/S248A and Cebpa KI/KI mice. This suggests that in young mice, S248 is dispensable for the induction of granulocytic-monocytic differentiation, and consequently, that phosphorylation of S248 is neither required for the preGM to GMP transition nor for the differentiation into mature neutrophil granulocytes in mice at 8 weeks of age.
Interestingly, a fraction of older Cebpa S248A/S248A mice develop a myeloid disorder in which differentiation towards the megakaryocytic and erythroid lineage is promoted at the expense of granulocytic-monocytic differentiation as evident by an increase in early megakaryocytic and erythroid progenitors and a corresponding decrease in their granulocytic-monocytic equivalents. Another, partly overlapping, fraction of older Cebpa S248A/ S248A mice presents with a mild expansion of the HSC-containing LSK compartment.
Collectively, these data contribute to the increasing body of evidence pointing towards C/EBPa as an important regulator of cell fate decisions in progenitors more primitive than GMPs. Conditional Cebpa knockout in the hematopoietic system in adult mice blocks the transition from CMP to GMP resulting in loss of granulocytes and monocytes but increases the numbers of myeloid  Figure S5) and was termed progressed. Numbers of mice in the other groups of transplanted mice were as follows: Cebpa KI/KI (n = 7) and Cebpa S248A/S248A (n = 3) mice. P values designate significance between progressed Cebpa S248A/S248A and Cebpa KI/KI , ns = not significant (mean +/2 standard deviation). doi:10.1371/journal.pone.0038841.g007 blasts and MEPs [6]. Consistently, Cebpa 2/2 fetal liver contains increased numbers of erythroid progenitors and erythroid cells [32]. Furthermore, ectopic expression of C/EBPa in HSCs induces granulocyte-monocytic differentiation and inhibits erythroid development resulting in an increase in mature granulocytes and loss of megakaryocytic and erythroid progenitors [33]. In line with this, mice with mutations in the Cebpa allele that either reduce the transcriptional activity of the protein or destroy its growth inhibitory function, all lack mature neutrophil granulocytes and have increased numbers of erythroid cells [19,20,21,23]. However, as these mice age, they progress from the neutropenic phenotype to a myeloid proliferative condition and later to an AML-like disease, showing first of all, that mutations in Cebpa result in a predisposition to myeloid diseases. Similarly, expression of the C/EBPa-S248A variant might act as an initiating factor in a multi-hit myeloid disorder associated with a biased lineage choice. As the development of this disorder might require additional genetic lesions this model also explains why only a fraction of the aged Cebpa S248A/S248A mice develop disease whereas others appear normal.
Growing evidence suggests that tight regulation of lineagespecific transcription factors plays a major role in the HSC compartment. In accordance, it has been shown that mice with various mutations in Cebpa have deregulated stem cell pools [19,20,21,23]. The data presented in this work also support a functional role for C/EBPa in HSCs since a fraction of the 1-2 years old Cebpa S248A/S248A mice present with an expanded LSK compartment. Furthermore, upon serial whole BM transplantation we observed a competitive advantage of Cebpa S248A/S248A BM cells compared to Cebpa KI/KI controls and after the second round of transplantation the Cebpa S248A/S248A BM cells have overtaken the recipient BM. Formally, we cannot exclude the possibility that the donor-derived cells in one or more of the primary recipients had an expansion of the LSK compartment. Therefore it is unclear whether the observed increased selective advantage of Cebpa S248A/S248A BM cells is due to an acquired event leading to an increase of functional stem cells or an increase of self-renewal per se in older Cebpa S248A/S248A HSCs. In either case, our data demonstrates that the S248A mutation confers a competitive advantage to HSCs, when these are subjected to proliferative stress.
The development of hematopoietic disorders in older Cebpa S248A/S248A animals may in principle be explained by two nonexclusive models: One formal possibility is that C/EBPa-S248 is receiving input from a signalling transduction pathway that changes its activity in an age-dependent manner. Precedence for such a model comes from studies of rodent livers, where C/EBPa display different phosphorylation patterns in young and old animals. Specifically, the proportion of C/EBPa that is phosphorylated on S193 by cyclin D3-cdk4/6 increase with age, which correlates with the expression pattern of cyclin D3, and results in a reduced ability to eliminate the growth repressive potential of C/ EBPa through dephosphorylation of S193 after partial hepatectomy [34,35,36].
Alternatively, the hematopoietic disorders in the old Cebpa S248A/ S248A animals arise as a result of different additional genetic/ epigenetic events in the Cebpa S248A/S248A mice, which facilitate the progression into an erythroid-biased condition, the expansion of the LSK compartment or both. We favour this second model as the Cebpa S248A/S248A mice with an expanded LSK compartment and an erythroid-biased lineage choice only partially overlap, which suggests that these phenotypes are driven by distinct molecular mechanisms. However, this does not exclude the possibility that mutation of S248 makes hematopoietic cells more susceptible to age-dependent changes in signal transduction pathways operating through C/EBPa.
In conclusion, our data show that S248 is dispensable for normal steady-state hematopoiesis, and that Cebpa S248A/S248A mice develop a low-penetrant myeloid disorder with age associated with a mild skewing towards the erythroid lineage and a partial differentiation block at the preGM to GMP transition. Additionally, Cebpa S248A/S248A BM display a competitive advantage during serial transplantation suggesting that phosporylation of S248 may normally serve to restrict HSC self-renewal in ageing mice.
In more general terms, the stark difference between the in vitro and in vivo phenotypes of the S248A mutant highlights the need to exert caution when extrapolating in vitro data to a more appropriate in vivo setting. Moreover, the phenotypic progression in old Cebpa S248A/S248A mice to a condition with a partial resemblance to the in vitro phenotype of mutating S248 may suggest that granulocytic-monocytic cell lines could be wired in a manner-either genetically or in terms of active signal transduction networks-that render them more relevant as models for an aged hematopoietic system.

Ethic statement
All mouse work was performed according to national and international guidelines and approved by the Danish Animal Ethical Committee. This study was approved by the review board at the Faculty of Health Sciences, University of Copenhagen (P10-014).

Generation of C/EBPa-expressing cells
A plasmid expressing rat C/EBPa was kindly provided by G. Behre. pBabePuro-C/EBPa-S248A-ER TM was constructed by ligating a Sfi I/EcoRI fragment containing the S248A substitution in place of the identical wild type fragment in pbabePuro-C/ EBPa-ER TM .
Phoenix-E cells were transiently transfected with pbabePuro-C/ EBPa-ER TM or pbabePuro-C/EBPa-S248A-ER TM and viruscontaining supernatant was collected after two and three days. 32Dcl3 cells were infected and selected by addition of puromycin as described in [18]. Cells were limited-diluted in 96-well culture dishes to obtain clones, which were expanded and tested for expression of C/EBPa or C/EBPa-S248A by western blotting.

Cell cycle analysis
The cell cycle distribution was analyzed at day 0-3 after addition of 4-OHT as previously described [18].

Mouse work and procedures
The Cebpa-S248A mutant was generated by using the Quickchange mutagenesis kit (Stratagene) and confirmed by sequencing. Cloning of the targeting construct, electroporation, selection of E14.1 ES cells, blastocyst injection and breeding of chimera was performed as described previously [20]. The Cebpa S248A and Cebpa KI alleles were backcrossed to C57BL/6 for at least 6 generations [20].
BM transplantations were carried out by tail vein injection of whole CD45.2 BM cells into CD45.1 recipients, which had been subjected to a lethal dose (900 rad) of gamma-irradiation 16 hours prior to transplantation. In the non-competitive setup, 1 to 2 million donor cells were used. In the competitive setup, primary transplantations were carried out by co-injecting 500.000 donor cells and 500.000 competitor cells (CD45.1). For secondary transplantations, BM cells from 6-7 primary recipients from the competitive transplantations were pooled, and 5 million of these cells were used for as secondary donor cells.

Flow cytometry
C/EBPa-ER and C/EBPa-S248A-ER expressing cell lines were stained with Mac-1 and Gr-1 and analyzed on a FACSCalibur. Briefly, 500.000 cells were incubated with 1 ml Fc receptor block (anti-CD16/32, BD) in 20 ml PBS3%FCS on ice for 15 min. The cells were washed with cold PBS3%FCS and stained in the dark on ice with antibodies against Mac-1 and Gr-1 or corresponding isotype controls (eBioscience) for 20 min. Cells were resuspended in PBS3%FCS and run on a FACSCalibur (BD).

Cytospins, colony assays and serial replating experiments
For the analysis of colony-forming potential, BM cells (5,000-20,000 cells/35-mm dish) were seeded in methylcellulose-based medium (M3434, StemCell Technologies Inc.) supplied with erythropoietin, IL-3, IL-6, and stem cell factor. After 10-12 d in culture, the colonies were scored as CFU-GM, BFU-E, or CFU-GEMM. In the serial replating experiments, a similar number of BM cells were seeded in M3434 medium, cultured for 7 days, and the number of colonies was counted. The cells were harvested, washed with PBS, diluted and replated in fresh M3434 medium and cultured for an additional 7 d. This procedure was repeated for 5 weeks.
Preparation of cytospins was performed as described in [21].