Megakaryocyte-erythroid lineage promiscuity in EKLF null mouse blood

Commitment towards megakaryocyte versus erythroid blood celllineages occurs in the megakaryocyte-erythroid progenitor, wheremutually exclusive expression of either EKLF (Klf1) or Fli1defines alternative outcomes. Here we show there is a markedincrease in the number of circulating platelets in mice lackingthe erythroid transcription factor EKLF. In addition, committederythroid cells retain key signatures of megakaryocytes bothon the cell surface and at the mRNA level. We also show thatthe effect of EKLF on megakaryocyte-erythroid progenitor lineagedecision and commitment is cell autonomous in bone marrow reconstitutionassays where stem cells lacking EKLF favor the megakaryocytedifferentiation pathway. We conclude the megakaryocyte programis aberrantly activated in EKLF null erythroid cells.

Key words: megakaryocyte, EKLF, megakaryocyte differentiation.

Introduction

TOP

Introduction

Erythroid Krüppel-like factor (EKLF) is the founding memberof the Krüppel-like factor family of transcription factorsand is critical for many aspects of erythropoiesis.^1–3Previous cell line and biochemical studies have shown an importantrole for EKLF in determination of erythroid versus megakaryocytelineages.^4–7 For example, studies performed in mouse EScells have demonstrated that the forced expression of EKLF reducesmegakaryocytic, but increases erythroid differentiation potential.⁵ Endogenous EKLF is expressed prior to erythroid commitmentto push megakaryocyte-erythroid progenitor (MEP) cells towardsthis pathway,⁷ a conclusion supported by an observed increasein megakaryocyte numbers during thrombopoietin (TPO) induceddifferentiation of EKLF^–/– primary fetal liver cultureswhen compared to WT cultures.⁵ Additional studies using specificsiRNAs targeting EKLF in murine erythroleukemia (MEL) cellsdemonstrated that the loss of EKLF enhances the megakaryocyteprogram.⁴

The biological mechanism responsible for lineage determinationin the MEP may involve direct functional interactions betweenFli1 and EKLF at megakaryocytic and erythroid gene promotersin an elaborate cross-antagonsim.⁸ It appears that the antagonismof megakaryocytic gene promoters is dependent on a specificEKLF sumoylation event that is critical for its function asa transcriptional repressor.⁶ To date no studies have examinedthe consequences of loss of EKLF in vivo with respect to megakaryopoiesispartly because until recently it was not appreciated that plateletsare produced very early during development.⁹

Design and Methods

TOP

Design and Methods

Mouse lines and procedures
EKLF^–/– erythroid tissues were collected from EKLF^+/–timed matings and genotypes confirmed as previously described.³ EKLF^+/– EGFP-actin transgenic mice were generated bycrossing EKLF^+/– mice¹⁰ with EGFP-actin transgenic mice¹¹ on a congenic Balb/c background. Balb/c bone marrow transplantrecipient mice were lethally irradiated (850Gy) and transplantedat eight weeks of age by injection of a 50:50 mix of WT fetalliver cells with EKLF^–/– fetal liver cells markedwith EGFP (EKLF^–/– GFP), or a 50:50 mix of WT andWT-EGFP livers (CON GFP) that served as a control.

Flow cytometry
Analysis of peripheral blood samples was performed using anLSRII flow cytometer (BD Biosciences, NJ, USA) with antibodiesfor CD71 (CD71-PE, 553267) (BD Pharmingen, CA, USA) and CD41(CD41-FITC, 553848)(BD Pharmingen). CD71-PE positive cells andEGFP positive cells were sorted from fetal liver and bone marrowpreparations, respectively, using an Influx cell sorter (Cytopeia,WA, USA).

Gene expression profiling
cDNA was prepared from CD71 positive sorted cells as previouslydescribed.³ Primers for real-time RT-PCR were designed usingPrimer Express software (Applied Biosystems, CA, USA). Quantitativereal time RT-PCR was performed using SYBR Green chemistry onan ABI-Prism 7500 sequence detection system (Applied Biosystems)(Primer sequences are shown in Online Supplementary Table S1).

Histology
Splenic sections were prepared from bone marrow transplantedmice, stained with hematoxylin and eosin and imaged using aBX51 microscope fitted with a DP70 digital camera imaging system(Olympus, Tokyo, Japan).

Methylcellulose colony assays
EGFP positive cells sorted from bone marrow preparations wereplated at 3x10⁴ cells/mL in MethoCult GFM3434 methylcellulose-basedmedium (Stem Cell Technologies, BC, Canada). CFUe were scoredafter three days. BFUe, BFUe/mk, and CFUmk were scored after12 days. Colonies were scored based on morphology previouslydescribed.¹²

Results and Discussion

TOP

Results and Discussion

The objective of this study was to characterize the fate ofMEPs in EKLF^–/– mice which die by E15.5 from anemia.^10,13 We initially observed a dramatic increase in the numberof circulating CD41⁺ platelets at E12.5 and E14.5 in EKLF^–/–embryos when compared to WT embryos (Figure 1A and C). Surprisingly,further analysis of CD71⁺ erythroid cells at E12.5 and E14.5revealed that EKLF^–/– cells also expressed CD41at significant levels on the cell surface (Figure 1A). We founda marked increase in the mean level of CD41 at the cell surfacefor primitive (E12.5) and definitive (E14.5) EKLF^–/–erythroid cells compared to WT cells (Figure 1B). This observationsuggested that the loss of EKLF was leading to failure of appropriateerythroid versus megakaryocyte lineage choices in the progenyof the MEP.

View larger version (32K):
[in this window]
[in a new window]
[Download PPT slide]
Figure 1. Increased platelets and lineage infidelity in EKLF^–/– mice. (A) Representative contour plots generated from E12.5 and E14.5 EKLF^–/– or wild-type (WT) peripheral blood by combined staining for CD71 (erythroid) and CD41 (megakaryocyte). The WT level of CD41 staining is indicated by a gray area for comparison (e: erythroid cells; p: platelets). (B) Quantification of the mean level of CD41 staining observed in the peripheral blood of E12.5 and E14.5 embryos for each genotype (Mean±SEM, $≥$ 3 for each genotype, *P<0.05 by Student’s t test). (C) Quantification of the percentage of platelets contained in the peripheral blood of E12.5 and E14.5 embryos for each genotype (Mean±SEM, n $≥$ 3 for each genotype, *P<0.05 by Student’s t test). (D) Gene expression determined by RT-PCR for CD71 sorted E14.5 fetal liver cells (Mean±SEM normalized to the housekeeping gene HPRT and presented on a log scale, n=9 for each, *P<0.05 by Student’s t test).

We hypothesized that these biphenotypic cells (CD71⁺, CD41⁺)had arisen due to either an expansion of a normal but rare celltype, or an aberrant gene expression program caused by the lossof EKLF. To confirm that the latter was the most likely explanationwe sorted CD71 positive erythroid cells from both EKLF^–/–and WT fetal livers at E14.5 and analyzed the expression ofmegakaryocyte genes^14,15 at the mRNA level by real-time RT-PCR.We found a significant increase in the expression of GpIIb inthe EKLF^–/– sorted cells compared to WT sorted cellsas expected since this gene encodes the CD41 antigen (Figure 1D). Levels of Gata1 and Gata2 were also significantly increased,although the change was less dramatic (Figure 1D). The genesfor platelet factor 4 (Pf4) and the TPO receptor (Mpl) werealso mildly increased in EKLF^–/– erythroid cells;however, this increase was not found to be significant. Interestingly,we found only a minor change (not significant) in the expressionlevel of the Fli1 gene, a critical determinant of the megakaryocytelineage. Expression of β_Maj-Globin (Hbb-b1) was almostabsent in the EKLF^–/– sorted cells, and served asa useful control (Figure 1D). We conclude that the loss of EKLFleads to failure to silence megakaryocyte specific genes inerythroid progeny of MEPs; that is a lineage infidelity thatresults in the majority of cells becoming abnormally biphenotypic.It is possible that the molecular mechanism responsible forthe lineage promiscuity found in EKLF^–/– erythroidcells involves both the loss of sumoylation dependent EKLF repressionand a loss of BKLF repression of megakaryocyte genes, and emergesindependently of a change in Fli1 gene expression.^6,16

In order to verify these observations, and to demonstrate thatthe effect of loss of EKLF on lineage choice was cell autonomous,we performed bone marrow reconstitution assays. The other aimof this study was to test whether EKLF acts as a tumor suppressorgene, in a similar manner to that which has been proposed forother Kruppel-like factors.¹⁷ EKLF^–/– erythroidprogenitors are more readily transformed by co-operating oncogenes,¹⁸ and we hypothesized EKLF^–/– MEPs might similarlyobtain additional mutations and generate leukemia. We generateda 50:50 mix of WT fetal liver cells with EKLF^–/–fetal liver cells marked with EGFP to reconstitute the bonemarrow compartment of lethally irradiated recipient mice (Figure 2A). A mixture of EKLF^–/– and WT fetal livers wasnecessary as reconstitution with EKLF^–/– bone marrowalone would lead to rapid death from anemia. Reconstitutionof lethally irradiated mice with a 50:50 mix of WT and WT-EGFPlivers served as a control for this experiment. Successful reconstitutionof the bone marrow was confirmed at eight weeks post transplantby FACS using markers for non-erythroid lineages (B220, Gr1,and Mac1) and fluorescence microscopy of peripheral blood smearsto look for EGFP positive cells which were derived from theEKLF^–/– compartment. As expected, we found equalreconstitution (~50:50 GFP+ to GFP-) of EKLF^–/–cells to each of the non-erythroid lineages investigated (datanot shown).

View larger version (44K):
[in this window]
[in a new window]
[Download PPT slide]
Figure 2. Lineage promiscuity and increased platelet production is transplantable. (A) A schematic presentation of the experimental design for bone marrow reconstitution assays. Mice receive a mixture of fetal liver cells containing either EKLF^–/– cells (EKLF^–/– GFP) or WT cells (CON GFP). (B) Spleen weights as a percentage of total body weight for EKLF^–/– transplants (EKLF^–/– GFP) or control transplants (CON GFP). Mean + SEM, n $≥$ 4 for each group, *P<0.05 by Student’s t test. (C) Histological examination of transplant mouse spleens by hematoxylin and eosin staining. Images acquired at 40x magnification with scale bar indicated. (D) Methylcellulose colony assays using GFP positive sorted bone marrow from EKLF^–/– GFP or CON GFP mice. CFUe were scored at day 3, other colonies were scored at day 12. Mean + SEM, n=4, *P<0.05 by Student’s t test.

At the conclusion of the experiment (12 months post transplant)mice were sacrificed and the erythroid compartment examinedby FACS, histologically, and by methylcellulose colony assays.We did not find a single leukemia suggesting either EKLF doesnot function as a tumor suppressor or additional mutations arerequired (n. more than 20 individual mice for experimental andcontrol groups). However, we found a significant increase inspleen weights in the EKLF^–/– GFP cohort comparedto the control cohort (CON GFP, Figure 2B). We attributed thisto an increase in red pulp caused by an expansion of the erythroidcompartment (Figure 2C). We believe this was the result of anincrease in red cell production by WT HSCs to compensate forthe EKLF^–/– HSCs which are unable to produce healthyred cells. We also observed an increase in megakaryocyte numbersin the spleens of these mice by hematoxylin and eosin staining(yellow arrows, Figure 2C). In colony assays using FACS sortedGFP positive bone marrow, the EKLF^–/– GFP mice producedsignificantly more megakaryocyte colonies (CFUmk) and mixederythroid/megakaryocyte colonies (BFUe/mk) than the controlmice (Figure 2D), demonstrating an increased megakaryocyte potentialfor EKLF^–/– HSCs and/or MEPs. We found no significantchange in the number of BFUe and CFUe as we expected from previouswork (Figure 2D).¹⁰ We conclude from these observations thatthe loss of EKLF in vivo leads to megakaryocyte-erythroid lineagepromiscuity. Although this may partly result from failure tosilence Fli1 in erythroid progenitor cells as suggested previously,^4,8 this work suggests other molecular mechanisms may be atplay.

Footnotes

Funding: this work was supported by a grant from the CancerCouncil Queensland (519718/ACP), and an Australian ResearchCouncil Discovery Grant (DP0770471/ACP). MRT is the recipientof an Australian Postgraduate Award.

The online version of this article has a supplementary appendix.

Authorship and Disclosures

MRT designed and performed research, and wrote the paper. ACPadvised on research and contributed to manuscript preparation.

The authors reported no potential conflicts of interest.

Received for publication April 13, 2009. Revision received June 18, 2009. Accepted for publication July 8, 2009.

References

TOP