Krüppel-Like Factor 1 (KLF1), KLF2, and Myc Control a Regulatory Network Essential for Embryonic Erythropoiesis

Abstract

The Krüppel-like factor 1 (KLF1) and KLF2 positively regulate embryonic β-globin expression and have additional overlapping roles in embryonic (primitive) erythropoiesis. KLF1^−/− KLF2^−/− double knockout mice are anemic at embryonic day 10.5 (E10.5) and die by E11.5, in contrast to single knockouts. To investigate the combined roles of KLF1 and KLF2 in primitive erythropoiesis, expression profiling of E9.5 erythroid cells was performed. A limited number of genes had a significantly decreasing trend of expression in wild-type, KLF1^−/−, and KLF1^−/− KLF2^−/− mice. Among these, the gene for Myc (c-Myc) emerged as a central node in the most significant gene network. The expression of the Myc gene is synergistically regulated by KLF1 and KLF2, and both factors bind the Myc promoters. To characterize the role of Myc in primitive erythropoiesis, ablation was performed specifically in mouse embryonic proerythroblast cells. After E9.5, these embryos exhibit an arrest in the normal expansion of circulating red cells and develop anemia, analogous to KLF1^−/− KLF2^−/− embryos. In the absence of Myc, circulating erythroid cells do not show the normal increase in α- and β-like globin gene expression but, interestingly, have accelerated erythroid cell maturation between E9.5 and E11.5. This study reveals a novel regulatory network by which KLF1 and KLF2 regulate Myc to control the primitive erythropoietic program.

INTRODUCTION

There are two developmental phases of erythropoiesis: primitive (embryonic) and definitive (adult). In mice, primitive erythropoiesis initiates in the yolk sac as early as embryonic day 7.5 (E7.5) and definitive erythropoiesis begins in the fetal liver by E11.5. The molecular mechanisms that control primitive erythropoiesis are generally less well understood than those that control definitive erythropoiesis. At E9.5, primitive proerythroblast cells synchronously enter the bloodstream from the yolk sac and follow a stepwise developmental program (48). In the circulation, these nucleated proerythroblasts proliferate until approximately E11.5 (18). In parallel, these cells undergo significant upregulation in the expression of embryonic globin, as well as other red cell genes, including surface protein glycophorin A (GPA), heme biosynthesis, and iron transport genes (18, 37, 40). Additionally, these primitive erythroid cells go through cellular and morphological changes in the maturation process from proerythroblasts to orthochromatophilic erythroblasts involving cytoskeletal reorganization, nuclear condensation, and enucleation between E12.5 and E16.5 (27, 41).

Understanding the roles of key transcription factors in primitive erythropoiesis will define molecular mechanisms controlling this process. Krüppel-like factors (KLFs) are a family of transcription factors that bind GC-rich sequences such as CACCC elements via their three carboxy-terminal Cys2/His2 zinc fingers (9, 59). The 17 members in the KLF gene family have important functions in cell differentiation, proliferation, and tissue development (47). Erythroid KLF (EKLF or KLF1) was the first family member to be identified in mice and humans (49) and is a master regulator of definitive erythropoiesis. It is expressed specifically in erythroid cells and positively regulates the adult β-globin gene (20, 49). KLF1^−/− mice develop fatal anemia during fetal liver erythropoiesis, due to a defect in the maturation of red blood cells, and die by E16 (11, 52, 58). In KLF1^−/− mice, definitive erythroid cells have reduced amounts of transcripts encoding major red cell membrane proteins and heme synthesis enzymes (22, 34, 51). KLF1 is necessary for normal chromatin looping between the β-globin locus control region enhancer and the adult β-globin promoter (21). Furthermore, KLF1 is required for the coassociation of KLF1-regulated genes in nuclear transcription factories (70).

KLF1 is also important in embryonic erythropoiesis. KLF1 directly binds to and positively regulates the human ε- and γ-globin promoters and the mouse Ey- and βh1-globin gene promoters in embryonic erythroid cells (2). KLF1^−/− primitive red blood cells have an accumulation of Heinz bodies reminiscent of α-globin chain aggregation and defective membrane morphology (22, 34). Recent evidence indicates that certain mutations in the human KLF1 gene correlate with hereditary persistence of fetal hemoglobin (3, 10). In contrast to its direct positive effect in embryonic cells, KLF1 has an indirect repressive effect on the γ-globin gene in adult cells, most likely via upregulation of BCL11A (10, 81).

In addition to KLF1, its close relative KLF2 (formerly lung KLF or LKLF) is expressed in erythroid cells (6, 7, 14, 46). KLF2 is important for normal primitive erythropoiesis in the yolk sac and, like KLF1, positively regulates mouse and human embryonic β-like globin gene expression in vivo (2). KLF2^−/− erythroid cells have abnormal morphology by E10.5 (6); subsequently, embryos develop vascular endothelial cell and heart defects and die by E14.5 (42, 76). Microarray gene expression assays showed that in KLF2^−/− mouse E9.5 erythroid precursor cells, 89 genes are significantly downregulated, including genes enriched in erythroid cells (63), such as those that encode the cell signaling factors CD24a antigen, cytotoxic T-lymphocyte-associated protein 2 alpha (Ctla2α), adenylate cyclase 7, and reelin (60). KLF2 can partially compensate for KLF1 function in regulating the mouse embryonic β-globin genes, because KLF1 and KLF2 double knockout (KO) mice die earlier in development at a critical phase of primitive erythropoiesis (5).

KLF1^−/− KLF2^−/− (double KO) embryos have features that are unique compared to those of single KO mice (5). The embryos are more anemic at E10.5, and Ey- and βh1-globin mRNA levels are considerably lower in double KO embryonic yolk sacs than in either KLF1^−/− or KLF2^−/− mice. The erythroid cells in E9.5 KLF1^−/− KLF2^−/− embryos have severely aberrant shapes due to cytoplasmic blebbing and also have altered cell maturation. The more severe blood cell abnormalities in double than in single KOs predict that red cell genes other than that for globin are also coordinately regulated by KLF1 and KLF2 during primitive erythropoiesis.

Only a few genes, other than those for KLF1 and KLF2, are known to specifically control primitive erythropoiesis in mammals. Flk-1 (4), vascular endothelial growth factor (VEGF/VEGF-A) (45), SCL/Tal1 (45, 69), LMO2/Rbtn2 (77), and Gfi-1b (68) are involved in embryonic erythropoiesis, but these factors have more global roles in the development of other hematopoietic lineages and in endothelial cells. GATA1 and GATA2 are specifically involved in embryonic erythropoiesis. Mice lacking GATA1 or GATA2 die between E10.5 and E11.5, exhibit pallor, and have embryonic erythroid cells arrested at an early proerythroblast-like stage or in reduced numbers (28, 75). However, little is known about the gene interaction networks controlled by these factors in primitive erythroid cells.

To identify regulatory pathways and interactions of major significance in primitive erythropoiesis, we investigated its mechanistic control by KLF1 and KLF2. Expression profiling of wild-type (WT), KLF1^−/−, and KLF1^−/− KLF2^−/− E9.5 erythroid cells revealed a limited number of genes expressed with a decreasing trend in these three genotypes. Among these genes, that for Myc was of particular interest, because it is synergistically regulated by KLF1 and KLF2 and mapped at a central node of the most significant regulatory network identified. Importantly, KLF1 and KLF2 bind to and transactivate the promoters of the Myc gene. The role of Myc in embryonic erythropoiesis was then functionally characterized by specific ablation in primitive proerythroblast cells. The absence of Myc induced a block in the normal expansion of the erythroid cell population and in the increase of globin gene expression after E9.5. Interestingly, circulating Myc null erythroid cells displayed early-onset maturation that did not preclude severe anemia in the embryos. Taken together, this study reveals a regulatory network involving KLF1, KLF2, and Myc that is essential for the normal developmental program of primitive erythropoiesis.

MATERIALS AND METHODS

Animals.

The KLF1 KO mouse model was generated by Perkins et al. (58), and mice were obtained from Jackson Laboratory. KLF2 KO mice were generated by Wani et al. (76). KLF1 and KLF2 double KO mice were generated by Basu et al. (5). These mouse models were bred to an FVB/N genetic background to allow valid comparisons of the microarray data.

Characterization of ErGFPcre (EpoRCreⁱⁿ) knock-in and Myc^fl/fl mutant mice (provided by U. Klingmüller, S. Philipsen, and F. Alt) was previously described (17, 33). To generate double transgenic mice (Myc^fl/fl; EpoR-Cre), EpoRCreⁱⁿ and Myc^fl/fl mice were crossed and the progeny were bred with Myc^fl/fl or Myc^fl/+ mice. Genomic analysis for the KLF1 KO, KLF2 KO, Myc^flox, and EpoRCre transgene was performed by PCR (33). The primers for the Myc^flox PCR were as follows: forward, 5′ GCCCCTGAATTGCTAGGAAGACTG 3′; reverse, 5′ CCGACCGGGTCCGAGTCCCTATT 3′. The conditions were 94^°C for 5 min; 30 cycles of 94°C for 30 s, 55^°C for 40 s, and 72°C for 50 s; and 72°C for 7 min. Transgenic mice were selected as homozygous for the murine β-globin haplotype “diffuse.” All experiments conformed to the standards of the Canadian Council on Animal Care and the Virginia Commonwealth University Institutional Animal Care and Use Committee.

Cellular and histological analyses.

For red cell counts, blood from the embryo and yolk sac (WT, KLF1^−/−, KLF2^−/−, and KLF1^−/− KLF2^−/−) or from the embryo only (control and Myc^fl/fl; EpoR-Cre) was allowed to drain into 4°C 1× phosphate-buffered saline for 15 min. Cells were counted with a hemocytometer. Cytospins were done for 7 min using a Shandon apparatus, and cells were fixed and stained with Giemsa. The control mice for all Myc^fl/fl; EpoR-Cre experiments were of the genotypes Myc^fl/+ and Myc^fl/fl. Heterozygous KO mice were used in timed matings to obtain E9.5 WT, KLF1^−/−, KLF2^−/−, and KLF1^−/− KLF2^−/− embryonic yolk sacs, which were embedded in freezing medium and cryosectioned for laser capture microdissection (LCM) (63).

LCM.

The Arcturus^XT LCM instrument (Applied Biosystems, Inc.) was used to isolate erythroid cells from yolk sac cryosections obtained from WT, KLF1^−/−, and KLF1^−/− KLF2^−/− embryos. The manufacturer's protocol was followed in preparing the instrument, isolating the cells, and visually inspecting the isolated erythroid cells on the LCM cap. RNA extractions from the LCM-purified samples yielded 10 to 20 ng of RNA. RNA quality was assessed by capillary electrophoresis, and intact samples were processed for microarray hybridization as previously described (64).

Microarrays and data analysis.

Fifteen micrograms of labeled erythroid cell cRNA was fragmented, and 10 μg was hybridized to Affymetrix GeneChip Mouse Genome 430A 2.0 arrays for 18 h (Affymetrix Inc.). The robust multiarray average (RMA) algorithm was used to obtain probe set expression summaries (36). Thereafter, control probe sets were removed, leaving 22,626 probe sets for statistical analysis. A moderated t test (71) was used to compare the 8 WT and 3 KLF1^−/− samples using the limma Bioconductor package (29, 71) in the R programming environment (http://www.R-project.org) and to identify the probe sets that were significantly differentially expressed. The Jonckheere-Terpstra trend test (35), a nonparametric test for ordered differences among classes, was performed to identify probe sets having significantly decreasing expression across the 8 WT, 3 KLF1^−/−, and 3 KLF1^−/− KLF2^−/− samples.

qRT-PCR.

Quantitative reverse transcription-PCR (qRT-PCR) was done with RNA extracted from mouse peripheral blood cells (E9.5, E10.5, or E11.5) or K562 cells with TRIzol reagent. cDNA was prepared from total RNA using the iScript cDNA synthesis kit (Bio-Rad, Inc.) or RT with Moloney murine leukemia virus reverse transcriptase (8). The NCBI database (http://www.ncbi.nih.gov) was used to establish that the primers are gene specific. qRT-PCR experiments were performed using the ABI Prism 7300 system (Applied Biosystems, Inc.) or Mx3005P quantitative PCR (qPCR) using SYBR green chemistry. A dissociation curve was used to verify that only a single product was amplified. Mouse cyclophilin A, ribosomal protein S16, or human cyclophilin A mRNA was used as an internal standard for normalization of input cDNA, as appropriate (8). qRT-PCR was performed in duplicate or triplicate for each biological replicate. Relative values were calculated using MxPro v4.01 software. Statistical analyses of qRT-PCR data were performed with a two-sample Student t test, and all findings were judged to be significant using an α ≤ 0.05 level of significance. The forward and reverse primers used in the qRT-PCR assays for the mouse genes were as follows: Gmpr, GGCAGAAGCTGAAACTCT and TCCACGTCCCCTTTGTAA; Pklr, ACGACTCAACTTCTCCCA and GCAAAACTTTCAGCCGC; Tfrc, AAAATGTGAAGCTCATTGTGAAA and CAACACCAGCACCCAAA; Ddx3x, GACCTGCCTAGTGATATCGA and AAGATCCAGTAAATCCTTTGTGA; Myc, TGCTGCATGAGGAGACA and TCGGGATGGAGATGAGC; Ctse, CGAGTGTCAATGAACCCCT and TGCAGTACACAGAAGGGA; Csda, CGAGGACGCGGAGAA and TCTTTGGTGTCATTTCGGTT; cyclophilin A, CACAAACGGTTCCCAGTT and ACCTTCCCAAAGACCACA; Gapdh, GACAACTTTGGCATTGTGG and AGTGGATGCAGGGATGA; εy, CAAGCTACATGTGGATCCTGAGAA and GCCGAAGTGACTAGCCAAAAC; βh1, AGGCAGCTATCACAAGCATCTG and AACTTGTCAAAGAATCTCTGAGTCCAT; ζ, CGAGCTGCATGCCTACAT and GCCATTGTGACCAGCAGACA. The qRT-PCR primers used for GPA were described previously (8). The human gene forward and reverse primers for qRT-PCR were as follows: KLF2, TGCCATCTGTGCGATCGT and GGCTACATGTGCCGTTTCATG; Myc, GACTCCAGCGCCTTCTC and CTTCCTCATCTTCTTGTTCCTCC; cyclophilin A, CCGAGGAAAACCGTGTACTATTAG and TGCTGTCTTTGGGACCTTG.

Chromatin immunoprecipitation (ChIP) assay.

The evolutionarily conserved region (ECR) browser interface was used to navigate through alignments of the human and mouse genes for Myc (56). An ECR is a 100-bp region of at least 70% similarity between mouse and human gene sequences. Mulan was used to align the mouse and human Myc promoters (55). The user-defined motif search tool of rVISTA within the Mulan interface was used to find KLF1 (CCNCNCCC) (25) and KLF2 binding sites (CCACCC or CCGCCC) (39) within ECRs. The website http://www.dcode.org was used to generate the schematic representation in Fig. 4.

Fig 4.

ChIP assays at evolutionarily conserved KLF1 and KLF2 binding sites in the promoters of the mouse gene for Myc. The consensus site for KLF1 binding is CCNCNCCC, and that for KLF2 is CCRCCC (CCACCC or CCGCCC). Each tick mark at the bottom of the diagram represents an ECR between the mouse and human genes for Myc. The ECR sequences are designated intergenic, intron, or coding. P1 (promoter 1) of the gene for Myc is at the left and includes the −126 KLF1/KLF2 consensus site, and P2 is to the right and includes the +2074 consensus binding site. These −126 and +2074 binding sites are within ECRs. ChIP assay values obtained with antibodies specific for KLF1 and KLF2 were compared to that obtained with IgG, which was set to 1. **, P = 0.0158; *, P = 0.0405 (unpaired t tests). The arrows indicate transcription start sites.

ChIP assays were performed with E11.5 erythroid cells as previously described (2). Chromatin was precipitated with anti-KLF1 (Abcam catalog no. AB-2483), anti-KLF2 (KLF2_Ng) (39), and nonspecific rabbit or goat IgG antibodies. Specific primers surrounding the putative KLF binding sites were used to determine KLF1 and KLF2 relative enrichment, which was normalized to IgG. Specific primers for the promoter of mouse β-actin were used as a negative control. The sequences of the primers used for qPCR are as follows: Myc P1 (5′ promoter), TCCTCTTTCCCCGGCTC and TCCTCCTCTCGCTTCCC; Myc P2 (3′ promoter), AGTAAAAGAGTGCATGCCTCC and GTACCCCAATCCTGAACCAC; ACTB (β-actin), ACCCCATTGAACATGGCATT and TGTAGAAGGTGTGGTGCCAGAT.

Transfection assays.

K562 cells (human myeloid leukemia; ATCC CCL-243) were grown and maintained at 37°C and 5% CO₂ in RPMI 1640 supplemented with 10% fetal bovine serum. K562 cells were cotransfected with a Myc promoter/luciferase reporter construct, a Renilla luciferase control reporter (pRL-TK; Promega, Inc.), and a KLF1 cDNA expression construct (49). The human Myc promoter/luciferase fusion gene constructs were obtained from Addgene (31, 32). Transfection assays were performed using Amaxa Nucleofector Solution V (Lonza Bio, Inc.) according to the company's optimized protocol for K562 (ATCC) cells. The Dual-Luciferase reporter assay system (Promega, Inc.) was used to measure firefly and Renilla luciferase activities with a TD-20120 luminometer (Turner Designs, Inc.) at 48 h after transfection. The promoterless pBv-luc construct was transfected as a negative control. Three separate transfections were performed, and the mean of the trials was determined. For KLF2 knockdown transient-transfection assays, 5 × 10⁵ K562 cells were transfected with Scramble small interfering RNA (siRNA) or siRNA Hs-KLF2-7 (Qiagen) at 10 nm and 20 nm using Lipofectamine RNAi-Max (Invitrogen) according to the manufacturer's recommendations. Cells were harvested at 36 to 48 h posttransfection. The methods for subsequent qRT-PCR are described above. The statistical analyses of transfection assay results were performed using a two-sample Student t test, and all findings were judged to be significant when α was ≤0.05.

Microarray data accession number.

The raw data from the microarrays (.CEL files) have been deposited in GEO under accession number GSE36427.

RESULTS

Expression profiling of KLF1^−/− and KLF1^−/− KLF2^−/− embryonic erythroid cells.

To explore the complex network of genes controlled by KLF1 and KLF2 in the primitive erythropoietic program, gene expression profiling was undertaken. We reasoned that genes important for KLF1^−/− KLF2^−/−-specific phenotypes are synergistically regulated by KLF1 and KLF2. E9.5 embryos were selected for this study because KLF1^−/− KLF2^−/− embryos are normal in gross appearance at that time point, unlike at E10.5 (5). E9.5 erythroid precursor cells were isolated from embryonic yolk sacs by LCM because it facilitated the enrichment of erythroid precursor cells. Total RNA from 8 WT, 3 KLF1^−/−, and 3 KLF1^−/− KLF2^−/− samples were used for Affymetrix 430A 2.0 microarray assays (63–65).

WT and KLF1^−/− samples were compared using a moderated t test, and 146 genes were identified as being significantly differentially expressed. The genes that are downregulated in KLF1^−/− compared to WT samples are indicated by a negative n-fold change in Table 1, and those upregulated are indicated by a positive n-fold change. Of the 47 genes that are significantly downregulated in embryonic erythropoiesis, many are also downregulated in adult KLF1^−/− erythroid cells, such as those for the transferrin receptor (Tfrc), erythrocyte protein band 4.9 (Epb4.9), and α-hemoglobin-stabilizing protein (Ahsp) (34). However, some genes, such as those for E2F2, p18 (Cdkn2c), and p21 (Cdkn1a) (61, 72), that are dysregulated in KLF1^−/− definitive erythroid cells are unaltered in primitive erythroid cells. On the other hand, a number of genes are significantly downregulated in KLF1^−/− compared to WT primitive erythroid cells, including those for pyruvate kinase liver and red blood cell (Pklr), adenylate cyclase 7 (Adcy7), and Myc.

Table 1.

Genes that are up- and downregulated in E9.5 KLF1^−/− versus WT yolk sac erythroid cells^a

AffyID	Entrez ID	Gene symbol	Expression summary for:		Fold change	P value
			WT	KLF1−/−
1419175_a_at	12231	Btn1a1	6.94	9.00	4.19	0.0000205
1449360_at	12984	Csf2rb2	6.49	8.22	3.33	0.0065116
1421811_at	21825	Thbs1	8.86	10.58	3.30	0.0041092
1434853_x_at	54484	Mkrn1	9.75	11.41	3.16	0.0046076
1449280_at	71690	Esm1	6.07	7.66	3.02	0.0000491
1460241_a_at	20454	St3gal5	6.41	8.00	3.00	0.0073785
1418480_at	57349	Ppbp	6.55	8.08	2.89	0.0000965
1456014_s_at	108101	Fermt3	8.98	10.48	2.84	0.0007647
1436905_x_at	16792	Laptm5	8.05	9.47	2.68	0.0076590
1455504_a_at	NA	NA	9.41	10.82	2.67	0.0023565
1460232_s_at	NA	NA	7.31	8.69	2.59	0.0008638
1451425_a_at	54484	Mkrn1	8.51	9.86	2.54	0.0026376
1418835_at	21664	Phlda1	8.86	10.19	2.51	0.0014728
1432004_a_at	13430	Dnm2	5.19	6.50	2.48	0.0041875
1424966_at	94346	Tmem40	6.31	7.56	2.36	0.0055598
1435386_at	22371	Vwf	5.52	6.76	2.35	0.0018190
1450852_s_at	14062	F2r	9.92	11.15	2.35	0.0043781
1416811_s_at	NA	NA	6.85	8.07	2.33	0.0018710
1418412_at	21987	Tpd52l1	6.73	7.93	2.30	0.0002513
1451263_a_at	11770	Fabp4	7.76	8.94	2.25	0.0006915
1423062_at	16009	Igfbp3	6.55	7.71	2.23	0.0004194
1448471_a_at	13024	Ctla2a	9.62	10.75	2.19	0.0097492
1437726_x_at	12260	C1qb	6.06	7.15	2.12	0.0018801
1448954_at	78593	Nrip3	6.17	7.22	2.07	0.0007463
1416488_at	12452	Ccng2	8.24	9.29	2.07	0.0004951
1448736_a_at	15452	Hprt	10.18	11.20	2.02	0.0023947
1450344_a_at	19218	Ptger3	6.15	7.15	1.99	0.0036792
1439259_x_at	105501	Abhd4	9.43	10.42	1.98	0.0006879
1422798_at	66797	Cntnap2	7.90	8.88	1.97	0.0058018
1426970_a_at	NA	NA	7.15	8.10	1.93	0.0043149
1422474_at	18578	Pde4b	6.97	7.91	1.92	0.0042260
1422977_at	14724	Gp1bb	6.23	7.15	1.90	0.0000247
1416882_at	67865	Rgs10	9.10	10.02	1.89	0.0006363
1452352_at	13025	Ctla2b	8.62	9.54	1.89	0.0079715
1451196_at	383295	Ypel5	7.97	8.89	1.89	0.0008458
1427168_a_at	12818	Col14a1	7.32	8.23	1.89	0.0012275
1434059_at	230088	B230312A22Rik	7.36	8.24	1.84	0.0034893
1451453_at	13143	Dapk2	6.85	7.72	1.83	0.0037859
1433593_at	383295	Ypel5	5.58	6.44	1.81	0.0011630
1420664_s_at	19124	Procr	9.01	9.86	1.81	0.0080605
1418435_at	54484	Mkrn1	7.99	8.80	1.76	0.0076558
1435275_at	333182	Cox6b2	6.90	7.71	1.76	0.0058952
1424923_at	20715	Serpina3g	6.99	7.79	1.74	0.0036619
1426454_at	11857	Arhgdib	8.86	9.65	1.73	0.0075277
1423319_at	15242	Hhex	7.72	8.51	1.73	0.0023488
1425158_at	57246	Tbx20	5.45	6.24	1.72	0.0095330
1448995_at	56744	Pf4	11.21	11.96	1.68	0.0064027
1422444_at	16403	Itga6	7.25	7.98	1.66	0.0019243
1438261_at	56222	Cited4	6.32	7.05	1.65	0.0002694
1415874_at	24063	Spry1	8.32	9.04	1.65	0.0005373
1415949_at	12876	Cpe	6.99	7.71	1.64	0.0009311
1423306_at	106878	2010002N04Rik	7.48	8.19	1.64	0.0016944
1417214_at	80718	Rab27b	5.42	6.11	1.61	0.0016003
1433964_s_at	108101	Fermt3	6.55	7.23	1.61	0.0065248
1419259_at	20163	Rsu1	10.12	10.79	1.60	0.0031843
1417023_a_at	11770	Fabp4	6.47	7.14	1.59	0.0037652
1426450_at	224860	Plcl2	7.69	8.34	1.57	0.0005064
1450016_at	12450	Ccng1	8.91	9.55	1.56	0.0059483
1448657_a_at	56812	Dnajb2	8.41	9.03	1.54	0.0031640
1452358_at	24004	Rai2	3.93	4.55	1.53	0.0000153
1427040_at	16543	Mdfic	8.38	8.98	1.52	0.0062468
1452366_at	234356	Csgalnact1	6.14	6.74	1.52	0.0073674
1448694_at	16476	Jun	6.84	7.43	1.50	0.0008474
1421268_at	22234	Ugcg	6.49	7.07	1.50	0.0062467
1434087_at	17769	Mthfr	6.96	7.54	1.50	0.0082196
1416315_at	105501	Abhd4	7.33	7.91	1.50	0.0065987
1417978_at	66892	Eif4e3	7.72	8.30	1.49	0.0018258
1423212_at	13619	Phc1	8.22	8.79	1.49	0.0074183
1455220_at	212398	Frat2	7.35	7.92	1.48	0.0019347
1421633_a_at	12950	Hapln1	5.95	6.51	1.47	0.0086383
1426734_at	224093	Fam43a	6.99	7.53	1.46	0.0063603
1417426_at	19073	Srgn	7.89	8.43	1.46	0.0098012
1419298_at	269823	Pon3	5.49	6.02	1.44	0.0032081
1418104_at	78593	Nrip3	5.32	5.83	1.43	0.0048456
1428794_at	432572	Specc1	7.03	7.54	1.42	0.0032746
1420124_s_at	102791	Tcta	7.39	7.90	1.42	0.0025608
1425570_at	27218	Slamf1	6.83	7.33	1.41	0.0089085
1415850_at	19414	Rasa3	7.01	7.50	1.40	0.0049996
1416714_at	15900	Irf8	5.98	6.46	1.40	0.0004215
1416843_at	18582	Pde6d	6.72	7.20	1.39	0.0057671
1431597_a_at	78593	Nrip3	4.44	4.90	1.38	0.0007931
1423286_at	12404	Cbln1	7.91	8.36	1.37	0.0053469
1425506_at	107589	Mylk	8.11	8.56	1.36	0.0066188
1448134_at	27355	X99384	7.79	8.24	1.36	0.0040177
1426624_a_at	66090	Ypel3	8.03	8.48	1.36	0.0098438
1418144_a_at	18720	Pip5k1a	7.39	7.82	1.36	0.0053209
1425125_at	18302	Oit3	6.72	7.16	1.35	0.0034544
1435383_x_at	17984	Ndn	11.13	11.57	1.35	0.0051937
1422699_at	11684	Alox12	5.37	5.80	1.35	0.0096007
1437401_at	16000	Igf1	6.44	6.86	1.34	0.0042858
1434555_at	11737	Anp32a	9.53	9.95	1.33	0.0032632
1435382_at	17984	Ndn	11.08	11.49	1.33	0.0087450
1449402_at	60322	Chst7	6.27	6.67	1.32	0.0037640
1422817_at	14729	Gp5	6.54	6.94	1.32	0.0072048
1436448_a_at	19224	Ptgs1	5.83	6.23	1.32	0.0032874
1437849_x_at	67416	Armcx2	9.95	10.35	1.32	0.0051815
1417363_at	22719	Zfp61	5.67	6.06	1.31	0.0090568
1455180_at	102371	Gcom1	5.55	5.93	1.31	0.0066221
1416935_at	22368	Trpv2	6.15	6.53	1.30	0.0057705
1427007_at	74131	Sash3	7.24	7.62	1.30	0.0070703
1417755_at	106021	Topors	5.54	5.88	1.27	0.0086579
1448024_at	18162	Npr3	6.12	6.46	1.27	0.0059614
1417523_at	56193	Plek	6.52	6.86	1.26	0.0014985
1448443_at	20713	Serpini1	5.44	5.75	1.24	0.0041357
1455367_at	213236	Dnd1	5.46	5.75	1.22	0.0050743
1428816_a_at	14461	Gata2	6.86	7.15	1.22	0.0071113
1427287_s_at	16439	Itpr2	5.18	5.44	1.20	0.0060864
1426604_at	24014	Rnasel	4.38	4.63	1.18	0.0096496
1450442_at	11519	Add2	7.44	7.11	−1.26	0.0046469
1422866_at	12817	Col13a1	6.77	6.43	−1.26	0.0015366
1455035_s_at	67134	Nop56	9.69	9.29	−1.32	0.0029171
1428390_at	72515	Wdr43	9.75	9.33	−1.33	0.0069107
1429080_at	67973	Mphosph10	9.02	8.60	−1.34	0.0017344
1419660_at	67008	1600012F09Rik	8.85	8.40	−1.36	0.0027637
1424021_at	65103	Arl6ip6	8.09	7.63	−1.37	0.0087204
1427954_at	270802	BC048403	4.82	4.36	−1.38	0.0095131
1420918_at	170755	Sgk3	6.97	6.51	−1.38	0.0061668
1417675_a_at	100019	Mdn1	7.68	7.17	−1.42	0.0098793
1449623_at	232223	Txnrd3	6.54	6.01	−1.45	0.0063133
1424437_s_at	192663	Abcg4	6.73	6.19	−1.45	0.0023347
1424858_at	217666	L2hgdh	7.48	6.91	−1.49	0.0063305
1424722_at	71775	1300017J02Rik	7.94	7.36	−1.49	0.0078258
1436990_s_at	NA	NA	10.46	9.85	−1.52	0.0045047
1460371_at	72630	Hspa12b	7.80	7.19	−1.53	0.0085880
1421883_at	15569	Elavl2	4.72	4.09	−1.55	0.0070508
1420339_at	76890	Memo1	8.46	7.82	−1.56	0.0019363
1434959_at	13363	Dhh	7.28	6.51	−1.71	0.0000812
1422920_at	57255	Cldn13	7.03	6.21	−1.76	0.0048145
1451229_at	232232	Hdac11	6.72	5.89	−1.78	0.0045228
1424117_at	414077	BC056474	7.03	6.18	−1.80	0.0052470
1449424_at	27260	Plek2	8.34	7.47	−1.83	0.0025433
1417092_at	19228	Pth1r	6.10	5.22	−1.84	0.0083709
1452324_at	19296	Pvt1	8.26	7.38	−1.84	0.0001237
1424942_a_at	17869	Myc	7.35	6.46	−1.85	0.0021150
1437614_x_at	224454	Zdhhc14	7.76	6.86	−1.87	0.0032367
1438975_x_at	224454	Zdhhc14	7.81	6.84	−1.95	0.0010981
1418591_at	58233	Dnaja4	7.58	6.59	−1.99	0.0014377
1432332_a_at	110959	Nudt19	8.25	7.19	−2.09	0.0009313
1448300_at	66447	Mgst3	9.75	8.68	−2.09	0.0023340
1460223_a_at	13829	Epb4.9	7.45	6.30	−2.22	0.0004694
1437502_x_at	12484	Cd24a	10.12	8.94	−2.27	0.0001713
1460444_at	109689	Arrb1	6.05	4.83	−2.33	0.0049112
1448182_a_at	12484	Cd24a	9.31	8.05	−2.40	0.0011112
1448530_at	66355	Gmpr	8.13	6.85	−2.42	0.0001778
1438619_x_at	224454	Zdhhc14	7.45	6.16	−2.44	0.0000852
1416034_at	12484	Cd24a	8.64	7.27	−2.59	0.0000683
1438711_at	18770	Pklr	8.63	7.21	−2.66	0.0000103
1451415_at	69068	1810011O10Rik	8.14	6.71	−2.68	0.0000060
1418600_at	16596	Klf1	10.67	9.15	−2.87	0.0013763
1452661_at	22042	Tfrc	11.40	9.87	−2.89	0.0017499
1456307_s_at	11513	Adcy7	8.62	7.02	−3.05	0.0000228
1422930_at	78369	Icam4	8.33	6.70	−3.09	0.0000331
1417184_s_at	NA	NA	11.67	9.97	−3.24	0.0011955
1435945_a_at	16534	Kcnn4	10.26	8.46	−3.48	0.0000054
1427866_x_at	15130	Hbb-b2	8.07	6.15	−3.79	0.0048748
1418909_at	27028	Ermap	10.17	8.11	−4.16	0.0013191
1420468_at	66772	Asb17	7.24	4.93	−4.96	0.0000063
1449077_at	170812	Ahsp	12.98	8.11	−29.35	0.0000000

^aMean probe set expression summaries for WT and KLF^−/− cells were calculated by RMA. Some genes were detected by multiple AffyID probe sets, and there are multiple entries for these genes. Positive n-fold values indicate genes that are upregulated in KLF1^−/− versus WT erythroid cells; negative n-fold values indicate genes downregulated in KLF1^−/− versus WT erythroid cells.

To define genes that might be synergistically regulated by KLF1 and KLF2, the Jonckheere-Terpstra trend test was used to identify probe sets having significantly decreasing expression across the WT, KLF1^−/−, and KLF1^−/− KLF2^−/− E9.5 embryonic erythroid cell samples. There were 110 probe sets corresponding to 101 unique genes that had a significant decreasing trend of expression in WT, KLF1^−/−, and KLF1^−/− KLF2^−/− erythroid cell samples (Table 2). The major functional classifications of these genes include homeostasis, regulation of apoptosis, hemopoiesis, and positive regulation of biosynthetic and macromolecule metabolic processes (Fig. 1). Because KLF1 and KLF2 have generally positive roles in gene regulation, we focused on genes that are down- rather than upregulated in mutants, to increase the likelihood of discovering genes that are directly regulated by KLF1 and KLF2. Several of the downregulated genes encode important red cell membrane and cytoskeletal proteins, such as Tfrc, adducin, Epb4.9, and the solute carrier 4 anion exchanger (SLc4a1), indicating previously unrecognized roles for KLF1 and KLF2 in coordinately regulating these genes in primitive cells. Decreased expression of these cell membrane and cytoskeletal genes may influence the observed abnormal cell morphology in embryonic erythroid cells, which is more severe in KLF1^−/− KLF2^−/− cells than in KLF1^−/− or KLF2^−/− cells. Certain red cell enzyme genes have more pronounced downregulation in KLF1^−/− KLF2^−/− cells than in KLF1^−/− cells, including Adcy7 and Pklr.

Table 2.

Genes that are significantly^a decreased in expression across WT, KLF1^−/−, and KLF1^−/− KLF2^−/− embryonic erythroid cells

Gene symbol	Description	P value	AffyID
Elavl2	ELAVc	0.000518	1421883_at
Gmpr3	GMPR	0.000518	1448530_at
Zdhhc14	Zinc finger, DHHC domain containing 14	0.000518	1438619_x_at
Arrb1	Arrestin, beta 1	0.000518	1460444_at
Sgk3	Serum/glucocorticoid regulated kinase 3	0.000518	1420918_at
Cd24ab	CD24a antigen	0.000797	1437502_x_at
Memo1	Mediator of cell motility 1	0.000797	1420339_at
Klf1	KLF1 (erythroid)	0.000797	1418600_at
1600012F09Rik	RIKEN cDNA 1600012F09 gene	0.000797	1419660_at
Hdac11	Histone deacetylase 11	0.000797	1451229_at
Pvt1	Plasmacytoma variant translocation 1	0.001211	1452324_at
Eraf	Erythroid-associated factor	0.001211	1449077_at
Tfrc	TFRC	0.001211	1452661_at
Pklr	PKLR	0.001211	1438711_at
Orc2l	Origin recognition complex, subunit 2 like (Saccharomyces cerevisiae)	0.001211	1418226_at
Mmaa	Methylmalonic aciduria (cobalamin deficiency) type A	0.001211	1417857_at
Dhh	Desert hedgehog	0.001211	1434959_at
Icam4	Intercellular adhesion molecule 4, Landsteiner-Wiener blood group	0.001211	1422930_at
Epb4.9	Erythrocyte protein band 4.9	0.001211	1460223_a_at
Chchd10	Coiled-coil-helix-coiled-coil-helix domain containing 10	0.001211	1436990_s_at
Nudt19	Nudixd-type motif 19	0.001812	1432332_a_at
Wdr43	WD repeat domain 43	0.001812	1428390_at
Ankrd28	Ankyrin repeat domain 28	0.001812	1454801_at
Kcnn4	Potassium intermediate/small conductance calcium-activated channel, subfamily N, member 4	0.001812	1435945_a_at
Mphosph10	M-phase phosphoprotein 10 (U3 small nucleolar ribonucleoprotein)	0.001812	1429080_at
Nudcd1	NudC domain containing 1	0.001812	1429655_at
Nudt19	Nudix-type motif 19	0.001812	1434216_a_at
Pth1r	Parathyroid hormone 1 receptor	0.001812	1417092_at
Usp59	Ubiquitin-specific peptidase 49	0.001812	1453037_at
Add2	Adducin 2 (beta)	0.001812	1450442_at
Ddx3x	DEAD/H (Asp-Glu-Ala-Asp/His) box polypeptide 3, X linked	0.001812	1416467_at
AU021092	Expressed sequence AU021092	0.002673	1437661_at
BC048403	cDNA sequence BC048403	0.002673	1427954_at
Asb17	Ankyrin repeat and SOCS box-containing 17	0.002673	1420468_at
Cldn13	Claudin 13	0.002673	1422920_at
Zdhhc14	Zinc finger, DHHC domain containing 14	0.002673	1437614_x_at
Hspa12b	Heat shock protein 12B	0.002673	1460371_at
Dnaja4	DnaJ (Hsp40) homolog, subfamily A, member 4	0.002673	1418591_at
Slc4a1	Solute carrier family 4 (anion exchanger), member 1	0.002673	1416464_at
Atg5	Autophagy-related 5 (yeast)	0.002673	1418235_at
Myc	Myelocytomatosis oncogene	0.002673	1424942_a_at
Pfkp	Phosphofructokinase, platelet	0.002673	1430634_a_at
BC048355	cDNA sequence BC048355	0.002673	1460713_at
Exosc2	Exosome component 2	0.002673	1426630_at
Nme3	Nonmetastatic cells 3, protein expressed in	0.002673	1448905_at
Irf4	Interferon regulatory factor 4	0.002673	1421174_at
AA517858	Expressed sequence AA517858	0.003229	1420121_at
Adcy7	Adenylate cyclase 7	0.003888	1456307_s_at
BC056474	cDNA sequence BC056474	0.003888	1424117_at
Hbb-b1/Hbb-b2	Hemoglobin, beta adult major chain/hemoglobin, beta adult minor chain	0.003888	1417184_s_at
Cd24ab	CD24a antigen	0.003888	1416034_at
Abcg4	ATP-binding cassette, subfamily G (WHITE), member 4	0.003888	1424437_s_at
Elavl2	ELAV	0.003888	1421882_a_at
Dck	Deoxycytidine kinase	0.003888	1439012_a_at
Ctse	Cathepsin E	0.003888	1418989_at
Pklr	PKLR	0.003888	1421258_a_at
Ermap	Erythroblast membrane-associated protein	0.003888	1418909_at
2810453I06Rik	RIKEN cDNA 2810453I06 gene	0.003888	1418389_at
Zdhhc14	Zinc finger, DHHC domain containing 14	0.003888	1438975_x_at
1110007M04Rik	RIKEN cDNA 1110007M04 gene	0.003888	1427997_at
Gcnt2	Glucosaminyl (N-acetyl) transferase 2, I-branching enzyme	0.003888	1430826_s_at
1300017J02Rik	RIKEN cDNA 1300017J02 gene	0.003888	1424722_at
Fgf13	Fibroblast growth factor 13	0.003888	1418498_at
Atp13a2	ATPase type 13A2	0.003888	1452747_at
Txnrd3	Thioredoxin reductase 3	0.003888	1449623_at
Myo19	Myosin XIX	0.003888	1451183_at
Cd24a	CD24a antigen	0.005577	1448182_a_at
Nop10	NOP10 ribonucleoprotein homolog (yeast)	0.005577	1423210_a_at
Hbb-b2	Hemoglobin, beta adult minor chain	0.005577	1427866_x_at
Slc4a1	Solute carrier family 4 (anion exchanger), member 1	0.005577	1434502_x_at
Xpo5	Exportin 5	0.005577	1451468_s_at
Mapk6	Mitogen-activated protein kinase 6	0.005577	1419169_at
Rhd	Rh blood group, D antigen	0.005577	1417049_at
Mgst3	Microsomal glutathione S-transferase 3	0.005577	1448300_at
Traf6	Tumor necrosis factor receptor-associated factor 6	0.005577	1435350_at
Clps	Colipase, pancreatic	0.005577	1438612_a_at
Dnaja4	DnaJ (Hsp40) homolog, subfamily A, member 4	0.005577	1434196_at
Tlcd1	TLC domain containing 1	0.005577	1452132_at
Olfr78	Olfactory receptor 78	0.005577	1421507_at
Nol5a	Nucleolar protein 5A	0.005577	1455035_s_at
Igfals	Insulin-like growth factor binding protein, acid-labile subunit	0.005577	1422826_at
Traf4	Tumor necrosis factor receptor-associated factor 4	0.005577	1416571_at
Rrp1b	rRNA processing 1 homolog B (S. cerevisiae)	0.005577	1452119_at
Srfbp1	Serum response factor binding protein 1	0.005577	1420509_at
Tmem49	Transmembrane protein 49	0.005577	1421491_a_at
Grin2b	Glutamate receptor, ionotropic, NMDA2B (epsilon 2)	0.006644	1422223_at
Usp38	Ubiquitin-specific peptidase 38	0.007888	1428592_s_at
Bco2	Beta-carotene oxygenase 2	0.007888	1421221_at
Arl6ip6	ADP-ribosylation factor-like 6 interacting protein 6	0.007888	1424021_at
Sphk1	Sphingosine kinase 1	0.007888	1451596_a_at
Frrs1	Ferric-chelate reductase 1	0.007888	1423465_at
Pigq	Phosphatidylinositol glycan anchor biosynthesis, class Q	0.007888	1437999_x_at
Plek2	Pleckstrin 2	0.007888	1449424_at
Ppara	Peroxisome proliferator-activated receptor alpha	0.007888	1449051_at
Ela3	Elastase 3, pancreatic	0.007888	1415883_a_at
Tcp11	T-complex protein 11	0.007888	1420730_a_at
Unknown	Unknown	0.007888	1425499_at
Josd2	Josephin domain containing 2	0.007888	1449046_a_at
1810029B16Rik	RIKEN cDNA 1810029B16 gene	0.007888	1423289_a_at
Slc25a37	Solute carrier family 25, member 37	0.007888	1417750_a_at
Pde3a	Phosphodiesterase 3A, cGMP inhibited	0.007888	1431914_at
Echdc1	Enoyl coenzyme A hydratase domain containing 1	0.007888	1419552_at
Csda	Cold shock domain protein A	0.007888	1451012_a_at
Pura	Purine-rich element binding protein A	0.007888	1449934_at
Btg2	B-cell translocation gene 2, antiproliferative	0.007888	1448272_at
2610209A20Rik	RIKEN cDNA 2610209A20 gene	0.007888	1423357_at
Ndn	Necdin	0.007888	1456575_at
Isoc1	Isochorismatase domain containing 1	0.007888	1425051_at
Foxm1	Forkhead box M1	0.007888	1448833_at
1600014C10Rik	RIKEN cDNA 1600014C10 gene	0.007888	1436289_x_at

^aP < 0.01.

^bIncluded are the products of 101 unique genes, but some were detected by multiple AffyID probe sets and there are multiple entries for these.

^cELAV, embryonic lethal, abnormal vision, Drosophila-like 2 (Hu antigen B).

^dNudix, nucleoside diphosphate-linked moiety X.

Fig 1.

Functional classifications of genes exhibiting significantly decreasing expression across WT, KLF1^−/−, and KLF1^−/− KLF2^−/− embryonic erythroid cells. Functional gene categories were determined with GO (Gene Ontology) using DAVID (Database Annotation Visualization and Integrated Discovery, http://david.abcc.ncifcrf.gov/). Eleven of the 101 genes identified were not registered by DAVID. The graph illustrates categories with at least 5 genes. Some genes are in multiple categories. The values on the x axis are numbers of genes. ncRNA, noncoding RNA.

An embryonic erythroid cell regulatory network controlled by KLF1 and KLF2.

Figure 2A shows the values for the RMA expression summaries for 7 selected genes that are downregulated with a decreasing monotonic trend in WT, KLF1^−/−, and KLF1^−/− KLF2^−/− primitive erythroid precursor cells. These genes were chosen for further analysis because their P values were relatively low (Table 2), and except for that of the gene for Ddx3x, their expression is enriched in E9.5 yolk sac embryonic erythroid cells compared to epithelial cells, based on our previous data (63). The selected genes encode GMP reductase (Gmpr); Pklr; Tfrc; DEAD/H box polypeptide 3, X linked (Ddx3x); Myc; cathepsin E (Ctse); and cold shock protein A (Csda).

Fig 2.

(A) RMA expression summaries for selected genes regulated by KLF1 and KLF2. LCM was performed to collect independent E9.5 WT, KLF1^−/−, and KLF1^−/− KLF2^−/− mouse erythroid precursor cells. Peptidylprolyl isomerase A (PPIA) and glyceraldehyde 3-phosphate dehydrogenase (GAPDH) mRNAs were used as negative controls that are not regulated by KLF1 or KLF2. The average and normalized RMA expression summary is indicated on the y axis. Black bars indicate WT (n = 8) samples, gray bars indicate KLF1^−/− (n = 3) samples, and white bars indicate KLF1^−/− KLF2^−/− (n = 3) samples. The error bars indicate the standard errors. (B) qRT-PCR verification of KLF1- and KLF2-regulated genes identified by microarray analyses. Independent samples of E10.5 WT, KLF1^−/−, KLF2^−/−, and KLF1^−/− KLF2^−/− peripheral blood cells were collected. qRT-PCR was performed with at least 100 erythroid cells per assay. GAPDH mRNA was used to normalize the data. The expression amounts in the mutants are expressed relative to those in the WT (100%). PPIA mRNA was used as a negative control. Black bars indicate WT samples, striped bars indicate KLF2^−/− samples, gray bars indicate KLF1^−/− samples, and white bars indicate KLF1^−/− KLF2^−/− samples. The values shown are mean of at least 3 independent samples. The error bars indicate the standard errors. Asterisks indicate significant difference from the WT (P < 0.05).

To validate that the 7 genes are differentially expressed in WT, KLF2^−/−, KLF1^−/−, and KLF1^−/− KLF2^−/− cells, qRT-PCR analysis was performed with E10.5 peripheral blood cells (Fig. 2B). To avoid bias, the samples used for qRT-PCR were obtained independently from those used for the microarray analyses. The amount of mRNA in the WT was taken as 100%, and the expression in mutants was compared to that in the WT. For each gene, the qRT-PCR results confirm a decreasing trend in WT, KLF1^−/−, and KLF1^−/− KLF2^−/− erythroid cells. For 6 of the 7 genes (except that for Gmpr), expression was significantly lower in KLF1^−/− KLF2^−/− than in WT cells. The consistency between the qRT-PCR and microarray analyses strengthens the validity of the microarrays. In Fig. 2B, it is possible to directly compare KLF2^−/− to the other genotypes, and KLF2 has a positive role in gene expression, but in general, its effect is more modest than that of KLF1. This may be explained by the uniquely KLF1-mediated transcription factories that coordinate the transcription of highly transcribed erythroid cell genes, including those for globin and cell membrane structural proteins (70).

To determine which genes might be the cause of the more severe KLF1^−/− KLF2^−/− erythroid cell phenotype compared to that of single KOs, ingenuity pathway analysis (IPA) of the 101 unique genes in Table 2 was used to discover gene networks. Seventeen networks were identified by IPA, but one was by far the most significant, scoring 46, indicating a P value of 1E-46. The functions of this IPA network include behavior, cell cycle, connective tissue development, and function (Fig. 3). Five of the 7 genes which were verified by qRT-PCR appear in this network (those for Myc, Tfrc, Pklr, Ddx3x, and Csda). The gene for Myc has an integral role in the network, forming a nexus, whereas some of the other genes appear to be more peripheral. Furthermore, as shown in Fig. 2B, Myc mRNA is reduced by approximately 2-fold in KLF2^−/− erythroid cells and by 3-fold in KLF1^−/− erythroid cells but by 10-fold in KLF1^−/− KLF2^−/− erythroid cells, compared to that in the WT. Therefore, KLF1 and KLF2 have synergistic roles in Myc regulation, as the reduction in the amount of Myc mRNA in the double KO is more than additive compared to the single KOs. The number of circulating erythroid cells at E10.5 is reduced in KLF1^−/− (4.5 ± 0.9 × 10⁵, n = 8; P < 0.003) and KLF1^−/− KLF2^−/− (2.6 ± 0.4 × 10⁵, n = 4; P < 0.002) embryos and yolk sacs compared to that in WT embryos and yolk sacs (8.7 ± 0.7 × 10⁵, n = 29). These data are consistent with a role for Myc in the proliferation of primitive erythroid cells.

Fig 3.

IPA network of genes coordinately regulated by KLF1 and KLF2. The 101 genes from Table 1, which had a significant decreasing trend of expression in WT, KLF1^−/−, and KLF1^−/− KLF2^−/− erythroid cell samples, were subjected to IPA analysis. The most significant network is shown. It scored 46, indicating that the overall P value is 1E-46; P values for individual genes are shown in Table 2. The network functions include behavior, cell cycle, connective tissue development, and function. Solid lines indicate direct relationships, and dashed lines indicate indirect relationships. Genes with statistically significant P values from the microarray assays are shaded, and darker shading represents lower P values.

The Mycg gene is directly regulated by KLF1 and KLF2.

ChIP assays were used to determine if KLF1 and KLF2 bind to consensus sites in the Myc promoters in WT E11.5 circulating mouse red blood cells. The consensus sites that were selected for the ChIP assays reside in regions evolutionarily conserved between the mouse and human genes. The results in Fig. 4 indicate that KLF1 binds to a CACCC element in the P1 promoter of the gene for Myc located at base position −126 (significant 7-fold enrichment compared to the IgG control). This validates ChIP-seq experiments with definitive erythroid cells which suggest that KLF1 binds near the Myc gene (60). Interestingly, KLF2 also binds to the gene for Myc in primitive erythroid cells (Fig. 4) but at the P2 promoter at base position +2074 (significant 2-fold enrichment compared to IgG). Both the P1 and P2 promoters of the Myc gene are utilized in erythroid cells (66).

To determine whether KLF1 and KLF2 directly regulate the gene for Myc, transient-transfection assays were performed with K562 cells, a human cell line representing the embryonic/fetal erythroid cell compartment. As shown in Fig. 5A, a KLF1 cDNA expression vector transactivates two different Myc gene promoter-luciferase fusion constructs containing the KLF1 binding site at P1 (Del-2269 and Del-352) by 2.2- and 3.4-fold, respectively, indicating a direct effect of KLF1 on the regulation of the Myc gene. Because KLF2 (80), unlike KLF1 (20), is already expressed in K562 cells, knockdown experiments were undertaken with two KLF2 siRNAs. As shown in Fig. 5B for one of these, 5- or 10-fold KLF2 knockdown in K562 cells was achieved compared to the Scramble control siRNA, and there was a correlative 2- or 3-fold reduction in Myc mRNA. Reduced Myc mRNA was observed with the two independent KLF2 siRNAs, suggesting that KLF2 positively regulates the expression of the Myc gene.

Fig 5.

(A) KLF1 transactivates the promoters of the gene for Myc in K562 cells. The black bars represent transfections with the luciferase reporter gene only, and the gray bars represent transfections which additionally have the expression plasmid pSG5-mEKLF. pBv-luc is a promoterless luciferase construct. Del-2269 and Del-352 are two different Myc promoter-luciferase fusion constructs. Data are the mean values from at least 3 independent samples. Error bars indicate standard deviations. **, P < 0.0001; *, P = 0.0034. The same total amount of DNA was used in each transfection by including the pSG5 empty vector in the transfections without the expression construct. (B) Knockdown indicates that KLF2 regulates Myc expression in K562 cells. K562 cells were transfected with Scramble siRNA, KLF2 siRNA (10 nm), or KLF2 siRNA (20 nm). qRT-PCR assay of KLF2 (black bars) and Myc (c-Myc, white bars) mRNAs in transfected cells was performed using cyclophilin A mRNA as an internal standard. The KLF2-to-cyclophilin A mRNA (siRNA Scramble) and Myc-to-cyclophilin A mRNA (siRNA Scramble) ratios were taken as 100%. Error bars show standard deviations. Data are the mean values of 2 independent samples. An asterisk indicates significant difference from siRNA Scramble (P < 0.05).

In vivo characterization of the role of Myc in embryonic erythropoiesis.

To determine whether Myc downregulation in KLF1^−/− KFL2^−/− embryos plays a critical role in embryonic erythropoiesis, Myc expression was specifically suppressed in the late stages of erythropoiesis and maturation, because Myc null embryos have severe developmental abnormalities and die in mid-gestation before E10 (16). Cell-autonomous Myc ablation was induced by mating mice carrying the Myc conditional allele (Myc^fl/fl) and mice with the erythropoietin receptor Cre allele (EpoR-Cre), which express Cre from the proerythroblast onward (44). The Myc^fl/fl; EpoR-Cre mice are alive at E11.5 (n = 8/8) but die by E12.5 (n = 6/6). Similar to KLF1^−/− KLF2^−/− embryos, the Myc^fl/fl; EpoR-Cre mice were indistinguishable from controls at E9.5 but pale and apparently anemic at E10.5 (n = 26) and 11.5 (n = 19) (Fig. 6).

Fig 6.

Phenotypes of control and Myc^fl/fl; EpoR-Cre embryos. At E10.5, the Myc^fl/fl; EpoR-Cre embryo and yolk sac (right) are very pale and vessel arborization is difficult to define compared to that in controls (left). Images were obtained with a Leica MZ12 stereomicroscope. Magnification, ×2. At E11.5, dissected Myc^fl/fl; EpoR-Cre embryos (right), relative to controls (left), appeared to have a growth delay, exhibited a drastic decrease in erythroid cells within the vasculature, and were severely anemic, which is indicative of erythroid cell deficiency. Images were obtained with a Leica MZ12 stereomicroscope. Magnification, ×1.25.

Quantification of circulating primitive erythroid cells was carried out with E9.5 to E11.5 control and Myc^fl/fl; EpoR-Cre embryos (Table 3). The number of cells in controls increased markedly between E9.5 and E10.5 by ∼14-fold (P < 0.0002) and between E10.5 and E11.5 by an additional ∼4-fold (P < 0.05), concordant with previous studies (18). In contrast, the Myc^fl/fl; EpoR-Cre erythroid cell count did not increase with age. At E9.5, cell counts in Myc^fl/fl; EpoR-Cre and normal embryos were not significantly different. However, E10.5 and E11.5 Myc^fl/fl; EpoR-Cre embryos showed a progressive and significant reduction in cell counts, relative to control embryos (P < 0.002 and P < 0.01, respectively), consistent with their severely anemic appearance. The low circulating erythroid cell counts in Myc^fl/fl; EpoR-Cre embryos at E11.5 are consistent with a potential role for Myc in the expansion and/or maturation of the circulating primitive erythroid cell population.

Table 3.

Circulating erythroid cell counts in control and Myc^fl/fl; EpoR-Cre embryos

Embryonic stage	n	Control (104 cells)	n	Mycfl/fl; EpoR-Cre (104 cells)
E9.5	6	1.7 ± 0.4a	2	2.1 ± 0.4a
E10.5	6	24.6 ± 3.9	4	1.9 ± 1.4b
E11.5	9	99.5 ± 22.8	5	2.0 ± 0.0c

^aThe values shown are mean numbers of circulating erythroid cells ± the standard error of the mean.

^bP < 0.002 (in comparison with controls).

^cP < 0.01 (in comparison with controls).

Cytospin preparations of embryonic peripheral blood at E10.5 and E11.5 were stained with Giemsa dye (Fig. 7A), and erythroid cell maturation was monitored. The circulating erythroid cells from controls were mainly basophilic erythroblasts at E10.5 and E11.5. Interestingly, blood from Myc^fl/fl; EpoR-Cre embryos from the same litters displayed a change in Giemsa reactivity and a more mature phenotype at both E10.5 and E11.5, resembling E13.5 orthochromatophilic erythroblasts (27). The E10.5 and E11.5 Myc^fl/fl; EpoR-Cre cells have more condensed nuclei, a lower nucleus-to-cytoplasm ratio, and a higher-than-normal proportion of cells with nuclei at the periphery (Fig. 7B). In contrast to KLF1^−/− KLF2^−/− cells (5), there is no evidence of cytoplasmic blebbing or cell shape anomalies in Myc^fl/fl; EpoR-Cre circulating blood cells, suggesting that this phenotype is controlled not by the gene for Myc but rather by other genes regulated by KLF1 and KLF2. Nevertheless, the finding that Myc^fl/fl; EpoR-Cre embryos have more mature cells suggests that the absence of Myc in the late stages of primitive erythropoiesis promotes earlier erythroid cell maturation than normally observed in development.

Fig 7.

(A) Cytological analysis of circulating primitive erythroid cells. Giemsa-stained cytospin preparations of peripheral blood from E10.5 and E11.5 control (left) and Myc^fl/fl; EpoR-Cre (right) embryos. Primitive blood cell preparations of the control embryos at E10.5 and E11.5 exhibited numerous large erythroid cells with features characteristic of basophilic erythroblasts. In contrast, Myc^fl/fl; EpoR-Cre embryos had altered Giemsa reactivity and a more orthochromatophilic erythroblast appearance. The Myc^fl/fl; EpoR-Cre cells also displayed more condensed and eccentric nuclei in comparison to those of controls, indicating that the Myc^fl/fl; EpoR-Cre cells were more mature. Images were obtained with a Zeiss Axiophot microscope. Magnification, ×64. (B) Nuclear-to-cell morphometric analysis indicates a more mature erythroid cell profile in Myc^fl/fl; EpoR-Cre embryos. Histograms of ratios of nuclear to cell surface areas of circulating erythroid cells from WT controls (n = 3) and Myc^fl/fl; EpoR-Cre embryos (n = 3) at E10.5 and E11.5 were obtained for a minimum of 103 primitive erythroid cells (range, 103 to 203). Importantly, the ratio of nuclear to cell surface areas in Myc^fl/fl; EpoR-Cre embryos was more than 2-fold lower than that of controls at E10.5 and E11.5 (****; P < 0.0001). In controls, the ratio of nucleus to cell surface area decreased significantly between E10.5 to E11.5 (P < 0.006).

To evaluate the role of Myc in the primitive erythroid cell developmental program, we monitored the mRNA encoding the erythroid cell-specific membrane marker GPA in the peripheral blood of E9.5 to E11.5 controls and Myc^fl/fl; EpoR-Cre embryos (Fig. 8A). Controls have an approximately 3- and 5-fold increased GPA/S16 ratio at E10.5 and E11.5 compared to that at E9.5, consistent with a previous report (37). Strikingly, Myc^fl/fl; EpoR-Cre embryos exhibited similar GPA/S16 ratios at E10.5 and E11.5, compared to E9.5. Hence Myc^fl/fl; EpoR-Cre E9.5 to E11.5 circulating erythroid cells maintain stationary GPA mRNA amounts and an unchanged cell count, whereas normal embryos show increases in both parameters.

Fig 8.

(A) Quantitative analyses of mRNA expression in Myc^fl/fl; EpoR-Cre and control embryos. Shown is the erythroid-cell-specific GPA expression detected by real-time PCR and normalized to S16 from peripheral blood of Myc^fl/fl; EpoR-Cre and control mice at E9.5, E10.5, and E11.5. GPA transcript levels increased significantly in controls, whereas in Myc^fl/fl; EpoR-Cre embryos, the levels were virtually unchanged from E9.5. (B) Quantification of embryonic (Ey, βh1, ζ) and adult β-globin expression. qRT-PCR was performed with peripheral blood at E9.5 and E11.5 and normalized to S16. A significant increase in globin gene expression was observed in control embryos from E9.5 to E11.5, while in Myc^fl/fl; EpoR-Cre embryos, the expression of none of these genes was increased and expression at E11.5 was comparable to that at E9.5. Data are presented as means ± standard deviations (n ≥ 6). Asterisks indicate significant difference from the control. *, P < 0.02; **, P < 0.01; ***, P < 0.0001.

To determine whether Myc deficiency could be dysregulating endogenous embryonic α-like and β-like globin gene expression, real-time PCR was performed to quantify embryonic globin mRNA levels in E9.5 and E11.5 peripheral blood. Figure 8B shows that the expression of the embryonic Ey, βh1, and ζ-globin genes is similar in control and Myc^fl/fl; EpoR-Cre embryos at E9.5. At E11.5, the expression of all three of these genes in controls increased by ∼8-, 3-, and 5-fold, respectively, consistent with proliferating and maturing circulating erythroid cells. In contrast, E11.5 Myc^fl/fl; EpoR-Cre embryos had levels similar to those measured at E9.5, thereby exhibiting less globin expression than controls (P < 0.007 for each globin). Further, the adult murine β-globin gene was also assessed in Myc^fl/fl; EpoR-Cre and control embryos. While the expression of the gene for β-globin was also significantly increased in controls between E9.5 and 11.5, levels were unchanged for Myc^fl/fl; EpoR-Cre, leading to reduced expression in comparison to that in controls. In fact, the globin gene expression profiles of Myc^fl/fl; EpoR-Cre appear almost identical between E9.5 and E11.5. Although the circulating primitive erythroid cells in Myc^fl/fl; EpoR-Cre embryos display a developmental block in globin gene expression and switching, there is an apparent accelerated maturation, as well as impaired expansion, of the cell population and this provides a rationale for the severe anemia seen at E11.5.

DISCUSSION

To elucidate the complex network of genes controlled by KLF1 and KLF2 in the primitive erythropoietic program, a global transcriptional analysis, followed by functional characterization, was undertaken. The regulatory network that was uncovered reveals potential mechanisms controlling the processes of differentiation, expansion, and maturation of the primitive erythroid cell population. A pathway of major significance was discovered that involves KLF1, KLF2, and Myc. Our data indicate that KLF1 and KLF2 directly regulate the gene for Myc, which, in turn, controls some of the phenotypes observed in KLF1^−/− KLF2^−/− embryos. Importantly, cell-autonomous Myc ablation at the proerythroblast stage demonstrates that Myc is an essential contributor to the developmental program of primitive erythropoiesis.

A unique expression profiling approach comparing WT, KLF1^−/−, and KLF1^−/− KLF2^−/− erythroid cells was devised to enhance the identification of key regulators in primitive erythropoiesis. A number of genes that are now recognized to be coordinately regulated by KLF1 and KLF2 in primitive erythroid cells (this report and reference 62), were also identified in expression profiling studies of KLF1^−/− definitive erythroid cells (22, 34, 61, 73). These include Epb4.9 and Tfrc (22, 34, 73), CD24a antigen and Kcnn4 (22), microsomal glutathione S-transferase 3 (Mgst3), solute carrier family 25 member 27 (Slc25a37), and erythroblast membrane-associated protein (Ermap) (73). These targets represent a subclass of genes directly or indirectly controlled by KLF1 and KLF2 in primitive erythropoiesis and by KLF1 in definitive erythropoiesis. This correlates with the fact that KLF1 and KLF2 expression is similar in primitive erythroid cells, whereas KLF1 expression is upregulated in definitive erythroid cells (2). Relatively few genes were determined to be coordinately regulated by KLF1 and KLF2, and some of these are important red cell genes, like those for Tfrc, Pklr, and Myc. Of these genes, that for Myc is synergistically regulated by KLF1 and KLF2. In addition, Myc is central to the pathway of genes regulated by KLF1 and KLF2 and is a critical determinant of primitive erythropoiesis.

The transcription factor Myc plays an important role in development and hematopoiesis (19, 50). Misexpression of Myc is associated with various forms of human cancer, as well as other genetic diseases (43, 57, 74, 82). Numerous studies have implicated Myc in the regulation of stem cell self-renewal, lineage differentiation, and cell proliferation and growth (e.g., 13, 15, 24). These cellular processes/functions are ongoing during primitive erythropoiesis, consistent with the observed expression and involvement of Myc.

The role of Myc in erythropoiesis has previously been explored. Several reports examined the role of ectopic Myc overexpression and suggested, despite conflicting outcomes, that Myc alters the differentiation rate (1, 78), self-renewal potential (12, 78), cell cycle arrest, proliferation, and terminal maturation (1, 38, 67). Discrepancies between these studies may result from the use of different erythroid cell culture systems at various differentiation stages or from dose-dependent responses (38). The relevance of these results to the in vivo activity of Myc in primitive erythropoiesis has not been addressed.

Epiblast-restricted Myc conditional ablation leads to fetal demise at approximately E11.5, and the yolk sac blood islands appear to have a reduced number of red blood cells (23). However, it was not clear whether this is due to the absence of Myc in blood cells or in other cell types. Hematopoietic Myc conditional KO mice generated using Vav-iCre precluded analysis of primitive erythropoiesis because the Vav gene promoter is activated during definitive erythropoiesis (30). In the present study, we demonstrated that Myc has a cell-autonomous role in primitive erythroid cells in vivo by specifically ablating the Myc gene at the proerythroblast stage. These results also provide the first evidence that Myc is an essential regulator in the late stages of embryonic erythroid cell maturation. Indeed, the constant number of erythroblasts in Myc^fl/fl; EpoR-Cre between E9.5 and E11.5 suggests that Myc plays a crucial role in controlling terminal cell division in circulating primitive erythroid cells. The finding that there are fewer circulating erythroid cells at E10.5 in both KLF1^−/− KLF2^−/− and Myc^fl/fl; EpoR-Cre embryos is consistent with the observed hierarchical regulation of Myc by KLF1 and KLF2. In the context of the enormous growth of the embryo between E9.5 and E11.5, failure to expand the synchronous circulating primitive erythroid cell population likely leads to embryonic death in both Myc^fl/fl; EpoR-Cre and KLF1^−/− KLF2^−/− mice. In addition, Myc^fl/fl; EpoR-Cre circulating primitive erythroid cells resemble orthochromatophilic erythroblasts of normal morphology, with nuclear condensation and asymmetric nuclear localization. This accelerated erythroid cell maturation suggests that upon cell cycle arrest, the onset of terminal maturation is triggered in circulating Myc^fl/fl; EpoR-Cre erythroid cells. This correlates with the reported requirement of Myc downregulation for terminal maturation in definitive erythroid cells (38).

Myc^fl/fl; EpoR-Cre circulating cells fail to upregulate the embryonic globin genes between E9.5 and E11.5, even though they appear to mature morphologically, and it is likely that this contributes to the paler appearance of embryos. In addition, the circulating Myc^fl/fl; EpoR-Cre cells do not initiate the embryonic-to-adult switch normally detected by E11.5. Consequently, the absence of Myc in primitive proerythroblasts appears to abrogate not only the developmental program but also indirectly the globin developmental switching pattern. Taken together, our results indicate that the absence of KLF1 and KLF2 in KO mice leads to downregulation of Myc, and this, in turn, causes a reduced number, an abnormal developmental program, and accelerated maturation of primitive erythroid cells after E9.5.

Pathway analysis mapped Myc to a nexus in the KLF1 and KLF2 regulatory network in embryonic erythroid cells. Interestingly, Myc had previously been shown to regulate genes important in red cells, such as the genes for the Tfrc, deoxycytidine kinase (dck), nucleolar protein 5A, Mgst3, and Foxm1 (26, 53). These genes are also coordinately regulated by KLF1 and KLF2 in primitive erythroid cells. The similarity in gene targets identified for KLF1 and KLF2 in embryonic erythropoiesis, and for KLF1 in definitive erythropoiesis (60, 61, 73), suggests a potential in vivo functional role for Myc in definitive cells. Significantly, the more severe erythroid cell abnormalities observed in KLF1^−/− KLF2^−/− compared to Myc^fl/fl; EpoR-Cre primitive erythroid cells likely stem from the dysregulation of genes other than that for Myc. Nevertheless, the overlap in gene targets is consistent with the hierarchical role of KLF1 and KLF2 in regulating Myc in embryonic erythropoiesis.

Our interaction network model of primitive erythropoiesis has at its core the coordinate roles of KLF1 and KLF2 in the direct regulation of the Myc gene. An important subset of genes in the broader network could be indirectly controlled by KLF1 and KLF2 but directly regulated by Myc. KLF1 and KLF2, as well as Myc, have roles in the expansion and maturation of the erythroid cell population. While KLF1 and KLF2 have been shown to directly stimulate globin gene expression in primitive erythroid cells, Myc may potentially influence hemoglobin production indirectly, by modulating heme synthesis or iron-controlling genes (54, 79). KLF1, KLF2, and Myc apparently regulate genes controlling cytoskeleton and membrane formation, and this could be directly or indirectly required for the expansion of primitive erythroid cells. These experiments integrate, for the first time, the roles of KLF1, KLF2, and Myc in driving primitive erythropoiesis.

In summary, we have defined a regulatory network controlled by KLF1 and KLF2 in primitive erythroid cells. The Myc gene is a direct target that is synergistically regulated by KLF1 and KLF2. Myc is an essential regulator in primitive erythroid cells and controls the expansion of the circulating erythroid cell population, the developmental program, and the kinetics of erythroid cell maturation. The identification of the KLF1, KLF2, and Myc network provides mechanistic insights into the dynamic sequential events that control the developmental program of primitive erythropoiesis.