|Home | About | Journals | Submit | Contact Us | Français|
The transcription factor CCAAT/Enhancer Binding Protein α (C/EBPα) is a critical regulator of myeloid development, directing granulocyte and monocyte differentiation. As such, it is dysregulated in over half of patients with acute myeloid leukemia (AML). C/EBPα expression is suppressed as result of common leukemia-associated genetic and epigenetic alterations such as AML1-ETO, BCR-ABL, FLT3-ITD, or CEBPA promoter methylation. In addition, 10–15% of patients with AML with intermediate risk cytogenetics are characterized by mutations of the CEBPA gene. Two classes of mutations are described. N-terminal changes result in expression of a truncated dominant negative C/EBPαp30 isoform. C-terminal mutations are in-frame insertions or deletions resulting in alteration of the leucine zipper preventing dimerization and DNA binding. Often, patients carry both N- and C-terminal mutations each affecting a different allele, and a mouse model recapitulates the human phenotype. Patients with mutated CEBPA AML comprise a clinically distinct group with favorable outcome consistently seen in patients with biallelic mutations. In addition, C/EBP family members are aberrantly expressing from the immunoglobulin heavy chain locus in 2% of pre-B ALLs. This review summarizes the normal hematopoietic developmental pathways regulated by C/EBPα and discusses the molecular pathways involved in mutated CEBPA AML and ALL.
CCAAT/Enhancer Binding Protein α (C/EBPα) contains 358 amino acid residues, with the intronless human CEBPA gene located on chromosome 19q13.1. C/EBPα has an α-helical, 86 residue, C-terminal basic region-leucine zipper (BR-LZ or bZIP) DNA-binding domain.1 The LZ contains a hydrophobic surface allowing homo-dimerization or hetero-dimerization with other bZIP proteins as a coiled-coil structure, thereby positioning the more N-terminal BR to enter the major groove and contact DNA (Figure 1).2 Additional members of the C/EBP family of bZIP transcription factors include C/EBPβ, C/EBPδ, and C/EBPε. C/EBP as obligatory homo- or hetero-dimers bind the DNA motif 5′-T(T/G)NNGNAA(T/G). C/EBP proteins also hetero-dimerize with members of the CREB or AP-1 families of bZIP proteins to bind hybrid DNA elements.3, 4 Once bound to DNA, C/EBPα activates transcription via its two N-terminal trans-activation domains.5
Full-length C/EBPα is 42 kd in molecular weight. Initiation of translation from an internal ATG located at amino acid 120 of the human protein leads to co-expression of a shorter, 30 kd isoform in a subset of normal tissues, though typically the p30 isoform is less abundant (Figure 1).6 C/EBPαp30 retains the ability to dimerize and bind DNA but lacks a potent trans-activation domain (TAD), allowing the p30 isoform to dominantly inhibit trans-activation by C/EBPα42, at least for a subset of C/EBP target genes.
Within hematopoiesis, C/EBPα is specifically expressed in granulocytes, monocytes, and eosinophils,7 though it is also found in hepatocytes, adipocytes, and type II pneumocytes.8 Low level C/EBPα expression is detectable in the hematopoietic stem cell (HSC) population, and expression increases as these cells develop into the common myeloid progenitor (CMP) and subsequently into the granulocyte-monocyte progenitor (GMP); C/EBPα levels finally diminish as immature myeloid cells mature to neutrophils or monocytes.7, 9 Deletion of the C/EBPα gene leads to arrest at the CMP to GMP transition, with reduced formation of both granulocytes and monocytes.9 When expressed in 32Dcl3 cells, representative of granulocytic progenitors, exogenous C/EBPα directs granulopoiesis.10 However, transduction of marrow cells with C/EBPα leads to increased monopoiesis at the expense of granulopoiesis11 – this may reflect formation of C/EBP:AP-1 hetero-dimers, as C/EBPα proteins containing artificial acid and basic LZs that force their homo-dimerization do not induce increased monopoiesis, whereas forced hetero-dimerization of C/EBPα with c-Jun or c-Fos strongly favors monocytic development.12 Induction of AP-1 proteins in myeloid cell lines using phorbol ester or IL-6 allows formation of endogenous C/EBP:AP-1 complexes during monopoiesis, whereas C/EBPα homodimers are more abundant in cells undergoing granulopoiesis in response to G-CSF.3 Reduced levels or activity of C/EBPα may be sufficient for monopoiesis via interaction with AP-1 proteins but not for granulopoiesis, as suggested by the finding that absence of RUNX1 or NF-κB p50 leads to about 25% of normal C/EBPα protein expression in marrow and diminished granulopoiesis.11, 13
In addition to interacting with AP-1 proteins, C/EBPα may stimulate monopoiesis by inducing transcription of the PU.1 gene via interaction with its promoter and -14 kb distal enhancer.14, 15 Notably, in contrast to C/EBPα reduced PU.1 protein levels due to gene or enhancer deletion favors granulopoiesis over monopoiesis,16–18 and exogenous C/EBPα induces granulopoiesis rather than monopoiesis in marrow cells lacking the PU.1 distal enhancer.15 A summary of the transcriptional control of myeloid development by RUNX1, C/EBPα, PU.1 and cooperating factors is shown (Figure 2).
As a major regulator of differentiation C/EBPα has a strong anti-proliferative effect, 19, 20 and in myeloid cells it inhibits the progression from G1 to S phase.10 This effect is independent of DNA binding 21 and occurs at least in part through protein:protein interaction with E2F mediated by the basic region.22 As a result, C/EBPα down regulates pro-proliferative transcription factors such as c-Myc. 23 Inhibition of cell cycle is part of the differentiation program induced by C/EBPα but is not sufficient by itself to direct terminal differentiation. 21
Cytokine signals cooperate with C/EBPα to regulate myeloid development. Study of individual GMP exposed to G-CSF or M-CSF demonstrates that these cytokines not only provide survival and proliferative signals but also contribute to lineage specification.24 Comparison of M-CSF with G-CSF signaling in lineage-negative marrow cells demonstrates that M-CSF more potently activates ERK via PLCγ, whereas G-CSF more potently activates STAT3 and specifically induces SHP2 tyrosine phosphorylation.25 Moreover, ERK inhibitors reduce formation of monocytic progenitor colonies whereas a SHP2 inhibitor reduces formation of granulocytic colonies. shRNA-mediated knockdown of SHP2 similarly reduces granulopoiesis via inhibition of CEBPA gene transcription.26 G-CSF activation of SHP2 may increase CEBPA mRNA expression due to the ability of SHP2 to increase the activity of RUNX1,27 with RUNX1 then directly inducing CEBPA gene transcription.13 ERK phosphorylates C/EBPα serine 21, reducing the activity of C/EBPα, perhaps by weakening its N-terminal TAD.28 Partial inactivation of C/EBPα by ERK downstream of M-CSF receptor would be expected to favor monopoiesis – in particular, weakened C/EBPα might be incapable of directing granulopoiesis but remain able to hetero-dimerize with AP-1 proteins to direct monopoiesis.
Two mechanisms affect C/EBPα function in AML: reduced expression as a downstream result of other AML-related mutation or mutation of the CEBPA gene.
Inhibition of C/EBPα expression or activity occurs via several mechanisms in different subsets of AML. As discussed above, deletion of the RUNX1 gene reduces CEBPA expression, and RUNX1 might directly activate CEBPA transcription. Related to these findings, mutation of RUNX1 leading to reduced RUNX1 levels or expression or fusion proteins that dominantly inhibit RUNX1 activity occurs in at least 30% of AML cases.29 In particular, t(8;21) leads to expression of RUNX1-ETO, which binds DNA via RUNX1 cis elements to repress target expression; inv(16) leads to expression of CBFβ-SMMHC, which interacts with RUNX1 to either sequester RUNX1 off chromatin or inhibit its activity on chromatin; t(3;21) expresses RUNX1-MDS1/EVI1 which also inhibits RUNX1 activity; and point mutations that inactivate RUNX1, the majority heterozygous, are present in a subset of patients.29, 30 RUNX1-ETO directly represses the CEBPA promoter, and blasts from patients with t(8;21)-associated AML indeed have reduced C/EBPα protein levels.31 Abnormal expression of EVI1 through translocations involving chromosome 3, is associated with high risk AML or MDS. EVI1 fusion protein expression is associated with translational suppression of C/EBPα expression. 32
Signaling pathways activated in AML can inhibit C/EBPα expression or activity. The activated receptor tyrosine kinase receptor mutant, FLT3-ITD, found in 30% of AML cases, reduces CEBPA transcription and leads to ERK modification of C/EBPα S21 to reduce the activity of C/EBPα.33, 34 BCR-ABL, an intracellular, constitutively active tyrosine kinase, inhibits translation of the CEBPA mRNA.35 Trib2 induces C/EBPα proteosomal degradation, dependent upon interaction with COP1.36
CEBPA promoter methylation is found in half of AML cases, often as the indirect consequence of several of the above mechanisms that down-modulate CEBPA transcription, and is most commonly associated with inv(16) and t(15;17). 37 In addition, a subgroup of AML whose gene expression profile aggregates with mutated CEBPA AML is characterized by CEBPA silencing through promoter hypermethylation and expression of T cell markers. The majority of these patients harbor activating NOTCH1 mutations and have poor outcome. 38, 39 Finally, the CEBPA gene open-reading frame itself is subject to mutation in approximately 10% of AML cases. Biologically and clinically this is a distinct subtype of AML as recognized in the 2008 World Health Organization classification of myeloid neoplasms 40 and elaborated on the next section.
As a key regulator of myeloid differentiation CEBPA is mutated in approximately 5–15% of patients with AML.41–45 Two categories CEBPA mutations are found in AML cases (Figure 1).42, 44 Interestingly, C/EBPα null mutations are rare, and the mutated proteins are expressed by the leukemic blasts, suggesting a selective pressure and an active role in leukemogenesis for the mutated C/EBPα proteins. N-terminal mutations typically lead to premature termination and translational reinitiation at methionine 120, leading to expression of the N-terminally-truncated C/EBPαp30, lacking a major TAD. C/EBPαp30 retains the capacity to bind DNA and to hetero-dimerize with C/EBPαp42, thereby dominantly inhibiting C/EBPαp42-mediated trans-activation. In addition, C/EBPαp30 induces expression of Ubc9, an E2 conjugating enzyme, which in turn sumoylates C/EBPαp42 on lysine 161. Sumoylated C/EBPαp42 has reduced capacity to activate transcription or slow proliferation. 46, 47 C-terminal mutations typically occur in the vicinity of the first α-helix of the LZ, preventing dimerization and therefore preclude DNA-binding. Strikingly, these C/EBPαLZ variants are in-frame insertions or deletions, indicating that the resulting proteins contribute to leukemic transformation. Indeed, although they themselves cannot bind DNA, several C/EBPαLZ oncoproteins inhibit apoptosis via induction of bcl-2 or Flice inhibitory protein, dependent upon interaction of their BR with NF-κB p50 bound to DNA in the promoter regions of these target genes.48, 49 Interaction of C/EBPαLZ oncoproteins with NF-κB p50 displaces HDACs, inducing transcriptional derepression, with the C/EBPα TAD then directing transcriptional activation.50
In two-thirds of AML cases harboring CEBPA gene mutations, one allele harbors an N-terminal variant and the other allele a C-terminal variant. Of note, in a mouse model, C/EBPαp30 and a C/EBPαLZ oncoprotein synergistically contribute to AML formation. 51 Patients with mutated CEBPA AML typically present with myeloblastic French-American-British types M1 or M2 morphology and associated with the following immunophenotype: HLA-DR(+), CD7(+), CD13(+), CD14(−), CD15(+), CD33(+), CD34(+).41–44, 52 CEBPA mutations occur almost exclusively in patients with intermediate-risk cytogenetics and predominantly in those with normal karyotype. Moreover, other Class II mutations such as Core Binding Factor (CBF) leukemia, mutated NPM1, or MLL-PTD only rarely overlap with mutated CEBPA.41, 43, 45, 53
Multiple cooperative groups reported that patients with mutCEBPA have significantly improved outcome compared to wtCEBPA, in par with patients with favorable risk cytogenetics, such as CBF leukemia. Further analysis of several,53–57 but not all, 43, 57 studies suggests that this benefit is restricted to patients with biallelic mutations. Further, patients with biallelic CEBPA mutations have a specific gene expression pattern, while monoallelic mutations do not aggregate in a specific pattern. 56, 57 Of particular importance to risk stratification of normal karyotype AML is the interplay of several prognostic markers. For example, mutated NPM1/wt-FLT3 cases have a favorable outcome while wt-NPM1/FLT-ITD is associated with a dismal prognosis. NPM1 is rarely mutated in patients with biallelic mutCEBPA, and these patients have a 2–4-fold lower incidence of FLT3-ITD.41, 45, 53, 57 Interpretation of the combinatorial effect of FLT3-ITD and mutCEBPA is complicated by inconsistent analysis of mono vs. biallelic mutations and FLT3-ITD allelic ratio. However, in several reports mutCEBPA predicts favorable outcome independent of FLT3 status.41, 43, 57–59 The incidence of FLT3-ITD or mutated NPM1 is similar in wtCEBPA and monoallelic mutCEBPA.53, 57
Germline CEBPA N-terminal mutation were found in pedigrees with familial AML, and progression to AML is typically associated with acquisition of a somatic C-terminal mutation.57, 60, 61 Importantly, approximately 10% of patients with mutCEBPA AML harbor a germline mutation.57, 61
MicroRNAs (miRNA) are small (20–22 bp) noncoding RNAs that play a key role in transcriptional regulation. miRNAs silence target genes by binding untranslated regions of mRNA resulting in translation inhibition or cleavage of coding mRNA. Several miRNAs are induced by C/EBPα in the course of normal myeloid differentiation.62–64 Dysregulation of miRNA expression is increasingly appreciated as a common feature of cancer. Global miRNA expression patterns accurately classify molecular subtypes of AML, including mutCEBPA leukemia that is characterized by down regulation of multiple miRNAs including the 181 family and miR-34a.59, 65 C/EBPα induces expression of miR-34a to silence E2F3 and suppress proliferation during normal granulopoiesis. Restoration of miR-34a expression in mutCEBPA AML blasts slows proliferation and induces differentiation.64
In contrast to RUNX1 and CEBPA, mutation of the PU.1 gene is rare in AML cases, despite the finding that deletion of the PU.1 −14 kb distal enhancer in mice leads to 20% of control PU.1 expression and highly penetrant AML.18 Nevertheless, as both RUNX1 and C/EBPα activate the PU.1 distal enhancer,15, 66 their diminished expression or activity would be expected to lead to reduced PU.1 levels, contributing to AML formation.
Besides stimulating myeloid differentiation and inhibiting apoptosis in cooperation with NF-κB p50, C/EBPα inhibits G1 to S cell cycle transition via direct interaction with E2F in myeloid progenitors, and this effect requires integrity of the C/EBPα N-terminus via an unclear mechanism.20–22 Thus, reduced C/EBPα expression or its N-terminal mutation might contribute to the myeloid transformation in part via removal of this cell cycle inhibitory effect.
Within hematopoiesis, C/EBPα is restricted to the myeloid lineages and it is not expressed in lymphocytes or their progenitors or in the erythroid/megakaryocytic lineages 7. Moreover, ectopic expression of C/EBPα leads to supression of Pax5, a central transcription factor in B-cell development, and redirects the fate of B cells into macrophages 67. Overexpression of wild type C/EBPα occurs in B precursor ALL carrying the t(14;19)(q32;q13) translocation which juxtaposes CEBPA and the immunoglobulin heavy chain enhancer locus.68 The leukemic blasts from patients with pre-B ALL carrying this translocation express high levels of wild type C/EBPα mRNA and protein.68 Importantly, no mutations were found in CEBPA, and the AML-associated p30 isoform 44 was not excessively expressed. Akasaka et al extended these findings and demonstrated involvement of other C/EBP family members in translocations with the IgH locus in pre-B ALL patients, and consequently overexpression of wild type C/EBPα, C/EBPβ, C/EBPδ, C/EBPε, or C/EBPγ in approximately 2% of patients with pre B ALL.69 The similar phenotype resulting from translocations of the various C/EBPs suggests an important role for the bZIP domain that is highly conserved among the different family members. Of note, this region mediates protein:protein interaction with NF-κB p50 and consequent induction of bcl-2 and FLIP and protection from apoptosis.22, 48, 49
C/EBPα is a key mediator of normal myeloid differentiation, contributing to both granulopoiesis and monopoiesis. C/EBPα also inhibits cell cycle progression and stimulates cell survival in cooperation with NF-κB p50. Alterations in the CEBPA gene or in pathways that down-modulate C/EBPα expression at the transcriptional, translational, or post-translational levels likely contribute to myeloid transformation by inhibiting myeloid differentiation while favoring myeloid progenitor cell cycle progression. In addition, C/EBPαp30, C/EBPαLZ variants, or residual wild-type C/EBPα may also contribute to myeloid transformation by inhibiting apoptosis. Restoration of C/EBPα expression in AML might provide a means to induce cell differentiation and slow cell proliferation to contribute to AML therapy.
I.P.P. is a St. Baldrick’s Foundation Scholar and the recipient of a Hyundai Hope on Wheels Award. A.D.F. is supported by NIH grants R01 HL089176 and U01 HL099775/HL100397 and the Maryland Stem Cell Research Foundation.